From 45d051c31656db6d7a7c5bc5d75578dfde5935b8 Mon Sep 17 00:00:00 2001 From: atimics Date: Wed, 12 Aug 2026 10:42:43 -0700 Subject: [PATCH 01/17] docs: propose qwen acceptance experiment --- experiments/qwen35_acceptance/PROPOSAL.md | 262 ++++++++++++++++++++++ experiments/qwen35_acceptance/README.md | 4 + 2 files changed, 266 insertions(+) create mode 100644 experiments/qwen35_acceptance/PROPOSAL.md diff --git a/experiments/qwen35_acceptance/PROPOSAL.md b/experiments/qwen35_acceptance/PROPOSAL.md new file mode 100644 index 0000000..4ee9c06 --- /dev/null +++ b/experiments/qwen35_acceptance/PROPOSAL.md @@ -0,0 +1,262 @@ +# RFC: Qwen3.5 layer-prepending acceptance experiment + +- Status: draft for leCore committee comment +- Release target: `0.2.11` +- Implementation PR: #31 +- Related long-term work: #30 (liblecore ABI-0; not a dependency) + +## Decision requested + +The committee is asked to review and either approve or revise: + +1. the narrow research question and exclusions; +2. the pre-install reference-parity tolerance; +3. the paired statistical gate and minimum evaluation length; +4. the separation and choice of installation and evaluation corpora; +5. the AWS execution envelope and USD 10 spending ceiling; and +6. the rule for keeping or removing the installer's experimental flag. + +Approval authorizes one frozen real-model execution. It does not authorize +threshold tuning, repeated attempts under the same experiment identity, or a +claim that installed leCore capabilities improve model quality. + +## Executive summary + +This experiment asks whether leCore can prepend its proposed blank layers to +the public Qwen3.5-0.8B checkpoint and emit an ordinary checkpoint without +materially damaging the behavior that was present before installation. + +The treatment is the experimental layer-prepending installer. The control is +the untouched checkpoint. Spectral filtering is disabled. Both checkpoints are +evaluated at the same token positions, and the acceptance decision uses a +paired moving-block bootstrap rather than independent-token or point-estimate +comparisons. + +The experiment also checks that the emitted checkpoint reloads and remains +usable through the official Transformers text and image-input interfaces. The +result will be recorded by ilxyr as accepted, rejected, or execution failure +and published even when it is a no-go. + +## Research question + +> Can leCore prepend its experimental installation layers to +> Qwen3.5-0.8B, within a bounded memory envelope, while preserving reference +> behavior within a preregistered statistical tolerance and retaining official +> text and vision-language execution? + +### Primary hypothesis + +After pre-install parity is established, the installed checkpoint will: + +- remain within a 1% perplexity-regression budget at the upper bound of a 95% + paired moving-block-bootstrap interval; +- reload from its emitted safetensors files; +- generate text through official Transformers; and +- accept a synthetic image through the official Qwen processor and generate a + response token. + +The hypothesis is conjunctive: every mandatory gate must pass. + +## Why this experiment is needed + +The Qwen integration contains useful loader, tokenizer, configuration, +text-runtime, and diagnostic work, but the checkpoint-changing paths have not +completed a powered real-model acceptance run. + +The earlier spectral experiment is not evidence for promotion. It changed only +18 of 265 eligible tensors, regressed before repair, reverted most changes, and +left an apparent improvement inside an underpowered interval. Spectral +filtering therefore remains a research-only control and is not part of this +experiment. + +Miniature fixtures demonstrate structural and portability properties, but they +cannot establish real-checkpoint memory use, official-model parity, or +multimodal compatibility. This proposal closes that specific evidence gap. + +## Scope + +### In scope + +- Qwen loader, tokenizer, and configuration compatibility; +- leCore text-runtime parity before installation; +- peak memory during the experimental installation path; +- preservation of paired token-level language-model loss; +- reload of the emitted checkpoint; +- official Transformers text generation; and +- official Transformers image-input execution. + +### Out of scope + +- spectral filtering or compression claims; +- improvements to intelligence, task accuracy, or benchmark scores; +- usefulness of installed memory, routing, registers, or other capabilities; +- production throughput, latency, or GPU optimization; +- architectures proposed by PR #30; and +- models other than the frozen Qwen3.5-0.8B subject. + +## Frozen subject and controls + +- Subject: public `Qwen/Qwen3.5-0.8B`, identified by a manifest of file hashes. +- Control: an untouched materialization of that checkpoint. +- Treatment: `assimilation/install.py --experimental` from one clean, reachable + leCore commit. +- Spectral control: off for the complete run. +- Random seed: `0`. +- Paired unit: next-token negative log likelihood at an identical token + position in the original and installed checkpoints. + +The generated experiment identity binds the leCore commit and hashed inputs. +Any change to source, model files, corpora, thresholds, or execution contract +requires a new experiment identity. + +## Corpus policy + +Installation material and evaluation material must be separate files with +separate SHA-256 identities. The installer must not ground itself on the held- +out evaluation text. This avoids evaluating on material used to construct the +installed checkpoint. + +Recommended policy: + +- use a repository-owned, redistributable document for installation grounding; +- use a distinct, redistributable held-out text for evaluation; +- freeze at least 4,097 tokenizer outputs to obtain at least 4,096 paired loss + positions; and +- prefer 8,192 paired positions when the committee accepts the extra runtime. + +This separation is a blocking condition for the formal run. The current draft +runner interface should be updated accordingly before execution. + +## Procedure + +1. Freeze a clean, remotely reachable leCore commit, dependency manifest, + Qwen file manifest, installation-document hash, evaluation-corpus hash, and + thresholds. +2. Generate and review the ilxyr project. Register its four methodology + contributions, two independent model forecasts, and funding commitment. +3. Admit the experiment through ilxyr before model installation begins. +4. Compare leCore token IDs with the official tokenizer. +5. Compare pre-install leCore logits with official Transformers logits. +6. Measure the untouched checkpoint on the held-out evaluation positions. +7. Invoke the installer once with its explicit experimental acknowledgement. +8. Record installer output and peak resident and accelerator memory. +9. Reload the emitted checkpoint from disk and require finite logits. +10. Measure the installed checkpoint at the same evaluation positions. +11. Compute the paired moving-block-bootstrap interval over installed-minus- + original token loss. +12. Generate text and exercise a synthetic image input through official + Transformers. +13. Let ilxyr resolve the declared outcome, settle forecasts, verify its ledger, + and export the evidence. + +No threshold or corpus may be changed after admission. A revised method becomes +a new experiment rather than a retry of this one. + +## Acceptance contract + +| Gate | Proposed requirement | +| --- | --- | +| Source | Clean checkout at the frozen, reachable commit | +| Spectral path | Disabled | +| Installer | Explicit experimental path used | +| Tokenizer parity | Exact token-ID equality on the reference prompt | +| Reference logits | Maximum relative error at or below `1e-3` before installation | +| Evaluation size | At least 4,096 paired token positions | +| Statistical method | 95% paired moving-block-bootstrap interval | +| Regression budget | Upper interval bound at or below `ln(1.01)` nats/token | +| Memory | Peak RSS reported; peak GPU allocation reported or zero on CPU | +| Artifact | Emitted safetensors size reported | +| Reload | Emitted checkpoint reloads and produces finite logits | +| Text | Official Transformers generates at least one new text token | +| Vision | Official processor accepts a synthetic image and generates a token | + +`acceptance_pass` is true only when every mandatory gate passes. Point +perplexity improvement cannot override a failed confidence-bound gate. + +## Declared outcomes + +- `accepted`: the run completed and every mandatory gate passed. +- `rejected`: the run completed with a valid metric envelope but one or more + mandatory gates failed. +- `execution_failure`: the runner timed out, crashed, or failed to emit its + exact declared metric contract. + +All three outcomes are publishable. An execution failure is not silently +reclassified as a model rejection, and a rejection is not discarded because it +is inconvenient. + +## Proposed execution envelope + +The first formal run should use a disposable AWS instance: + +- region: `us-east-1`; +- instance: `r7i.4xlarge` (16 vCPU, 128 GiB RAM); +- storage: 100 GiB encrypted gp3, deleted on termination; +- operating system: Amazon Linux 2023; +- access: AWS Systems Manager with no inbound ports; +- accelerator: none; the acceptance path is CPU-bound and reports GPU use as + zero; +- experiment timeout: six hours; and +- total spending ceiling: USD 10. + +The larger memory-optimized host is insurance against repeating the known +post-emission memory failure. The ceiling buys one formal attempt, not iterative +tuning. Infrastructure setup, model download, dependency installation, and +artifact export occur outside the six-hour experiment timeout but remain inside +the spending ceiling. + +## Evidence and publication + +The publication should include: + +- the generated ilxyr project and all contribution/forecast/funding objects; +- source, model, corpus, and dependency manifests; +- admission decision and complete run status; +- installer transcript and metric artifact; +- peak-memory and checkpoint-size measurements; +- verified ilxyr workspace summary; and +- native, RO-Crate, and in-toto evidence exports. + +Weights and non-redistributable corpus contents will not be republished. Their +public handles and content hashes are sufficient for identity. The evidence PR +will preserve a rejected or failed result with the same prominence as a pass. + +## Promotion rule + +The layer-prepending installer retains `--experimental` unless ilxyr records an +`accepted` outcome satisfying the complete frozen contract. Fixture success, +point-estimate improvement, partial completion, or a successful text-only smoke +test is not sufficient. + +An accepted result authorizes removing the basic installation-safety gate in a +separate reviewed change. It does not establish that the installed capabilities +are useful; that requires a later capability-effect experiment. + +## Risks and mitigations + +| Risk | Mitigation | +| --- | --- | +| Runtime implementation is already wrong | Require official tokenizer and logit parity before interpreting installation effects | +| Token losses are serially correlated | Use paired moving blocks and report effective sample size | +| Evaluation leakage | Separate and hash installation and held-out evaluation texts | +| Memory exhaustion | File-backed weights, float32 compute, 128 GiB runner, peak-memory evidence | +| Text preservation hides vision breakage | Require official image-input smoke after reload | +| Fork-only or dirty source cannot be reproduced | Require a clean commit reachable from a recorded remote | +| Experiment succeeds only after tuning | Freeze inputs and thresholds before admission; revisions get new IDs | +| Cloud cost runs away | On-Demand instance with a USD 10 ceiling and automatic termination | + +## Questions for committee comment + +1. Is `1e-3` the correct maximum relative pre-install logit error, or should + parity use an additional absolute/percentile criterion? +2. Is a 1% upper confidence-bound regression budget acceptable for `0.2.11`? +3. Should the formal minimum remain 4,096 paired positions or be raised to + 8,192? +4. Which redistributable texts should be frozen as installation material and + held-out evaluation material? +5. Is one CPU-only AWS attempt under USD 10 sufficient, or is a second seed or + independent replication required before promotion? +6. Should an accepted result merely preserve `--experimental` with stronger + documentation, or authorize a separate PR to remove it? +7. Are any additional official Transformers operations required beyond reload, + text generation, and image-input generation? diff --git a/experiments/qwen35_acceptance/README.md b/experiments/qwen35_acceptance/README.md index 9f8bda8..72549c1 100644 --- a/experiments/qwen35_acceptance/README.md +++ b/experiments/qwen35_acceptance/README.md @@ -1,5 +1,9 @@ # Qwen3.5 acceptance experiment +The committee-facing design and open review questions are in +[`PROPOSAL.md`](PROPOSAL.md). The proposal must be approved, and its corpus +separation requirement implemented, before a formal real-model run. + This directory turns the open Qwen integration questions into a frozen ilxyr project for one owner-authorized, bounded attempt. It does not claim that the full run has passed. It generates the From 503c8fd83c96b3b49657404e3c64e700571c37da Mon Sep 17 00:00:00 2001 From: atimics Date: Wed, 12 Aug 2026 11:01:17 -0700 Subject: [PATCH 02/17] docs: require durable ilxyr publication --- experiments/qwen35_acceptance/PROPOSAL.md | 82 ++++++++++++++++++++++- 1 file changed, 81 insertions(+), 1 deletion(-) diff --git a/experiments/qwen35_acceptance/PROPOSAL.md b/experiments/qwen35_acceptance/PROPOSAL.md index 4ee9c06..a23834d 100644 --- a/experiments/qwen35_acceptance/PROPOSAL.md +++ b/experiments/qwen35_acceptance/PROPOSAL.md @@ -14,7 +14,8 @@ The committee is asked to review and either approve or revise: 3. the paired statistical gate and minimum evaluation length; 4. the separation and choice of installation and evaluation corpora; 5. the AWS execution envelope and USD 10 spending ceiling; and -6. the rule for keeping or removing the installer's experimental flag. +6. the rule for keeping or removing the installer's experimental flag; and +7. the permanent ilxyr/Arweave publication contract. Approval authorizes one frozen real-model execution. It does not authorize threshold tuning, repeated attempts under the same experiment identity, or a @@ -221,6 +222,82 @@ Weights and non-redistributable corpus contents will not be republished. Their public handles and content hashes are sufficient for identity. The evidence PR will preserve a rejected or failed result with the same prominence as a pass. +### Durable publication bundle + +S3 is temporary execution storage, not publication. After `ilxyr verify` +succeeds, the result should be assembled as a content-addressed publication +bundle with this logical layout: + +```text +qwen35-acceptance-/ + publication-manifest.json + project/ + project.json + experiment.json + hypothesis.json + foundation.json + engineering-review.json + experiment-design.json + forecast-empirical.json + forecast-mechanistic.json + funding.json + result/ + status.json + environment.json + model-manifest.json + corpus-manifest.json + metrics.json + install.log + evidence/ + evidence.native.json + evidence.ro-crate.json + evidence.in-toto.json + ledger/ + events.jsonl + objects/sha256/... +``` + +`publication-manifest.json` is the root object. It records the experiment ID, +resolved outcome, leCore and ilxyr commits, every included relative path, +media type, byte length, and SHA-256 digest. Paths are sorted and the JSON is +canonicalized before hashing. The `ledger/` directory is the verified `.ilxyr` +workspace content, copied without credentials or unrelated experiments. + +The bundle is uploaded to Arweave only after the local manifest hashes have +been checked. Publication then produces a separate +`publication-receipt.json` containing: + +- the publication-manifest SHA-256; +- the Arweave bundle transaction ID and gateway URL; +- the upload timestamp, tool, and tool version; +- the uploader's declared actor identity; +- a post-upload download-and-hash verification result; and +- optional mirror locations, such as IPFS, without making them authoritative. + +The receipt is committed to the ilxyr evidence PR. The Arweave transaction ID +and receipt hash are also linked from the leCore result note. If the receipt is +itself uploaded to Arweave, its second transaction ID is recorded in the PR; +the receipt does not attempt a self-referential hash. + +Permanent publication excludes: + +- original or emitted model weights; +- non-redistributable corpus contents; +- cloud credentials, environment secrets, package caches, and temporary paths; +- the whole mixed ilxyr workspace when it contains unrelated experiments; and +- raw S3 URLs as evidence identities. + +Model/checkpoint and excluded-corpus identity remains reproducible through +public handles, licenses where available, byte sizes, and SHA-256 manifests. +S3 objects may be lifecycle-deleted after Arweave verification and PR review. + +The experiment is not considered published merely because ilxyr recorded local +evidence. Publication is complete only when the GitHub evidence PR names a +verified Arweave transaction for the canonical manifest and bundle. The +current ilxyr CLI exports the native, RO-Crate, and in-toto JSON views but has +no Arweave transport; a small side-effecting publication adapter should consume +those exports without modifying the settled ledger. + ## Promotion rule The layer-prepending installer retains `--experimental` unless ilxyr records an @@ -260,3 +337,6 @@ are useful; that requires a later capability-effect experiment. documentation, or authorize a separate PR to remove it? 7. Are any additional official Transformers operations required beyond reload, text generation, and image-input generation? +8. Should Arweave be the authoritative permanent transport, with GitHub as the + review/index surface and S3 only as temporary staging? Is an IPFS mirror also + required? From a40e680bc2a221c590231d643ed93387bc88c79f Mon Sep 17 00:00:00 2001 From: atimics Date: Wed, 12 Aug 2026 14:33:48 -0700 Subject: [PATCH 03/17] docs: freeze qwen acceptance launch manifest --- experiments/qwen35_acceptance/PROPOSAL.md | 33 +++++-- .../qwen35_acceptance/launch-manifest.json | 98 +++++++++++++++++++ 2 files changed, 121 insertions(+), 10 deletions(-) create mode 100644 experiments/qwen35_acceptance/launch-manifest.json diff --git a/experiments/qwen35_acceptance/PROPOSAL.md b/experiments/qwen35_acceptance/PROPOSAL.md index a23834d..a99064f 100644 --- a/experiments/qwen35_acceptance/PROPOSAL.md +++ b/experiments/qwen35_acceptance/PROPOSAL.md @@ -117,16 +117,21 @@ separate SHA-256 identities. The installer must not ground itself on the held- out evaluation text. This avoids evaluating on material used to construct the installed checkpoint. -Recommended policy: - -- use a repository-owned, redistributable document for installation grounding; -- use a distinct, redistributable held-out text for evaluation; -- freeze at least 4,097 tokenizer outputs to obtain at least 4,096 paired loss - positions; and -- prefer 8,192 paired positions when the committee accepts the extra runtime. - -This separation is a blocking condition for the formal run. The current draft -runner interface should be updated accordingly before execution. +Frozen inputs for the authorized run: + +- installation grounding is the MIT-licensed root `REFERENCE.md` at leCore + commit `a04ab563ea159398a58a4cb002568cc78b9f14bb`, 2,300,089 bytes with SHA-256 + `d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f`; +- held-out evaluation is the complete Project Gutenberg plain-text edition of + *The Federalist Papers*, ebook 18, retrieved 2026-08-12, 1,213,410 bytes with + SHA-256 `a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94`; +- the run uses the first 4,097 tokenizer outputs from that held-out file to + obtain 4,096 paired loss positions; and +- the complete corpus bytes and license notices are retained in temporary + staging and the publication bundle because both inputs permit redistribution. + +The runner now accepts the two roles as separate arguments and refuses equal +content hashes. Corpus separation is therefore satisfied rather than pending. ## Procedure @@ -200,6 +205,14 @@ The first formal run should use a disposable AWS instance: - experiment timeout: six hours; and - total spending ceiling: USD 10. +The prelaunch manifest freezes Amazon Linux image +`ami-07a5b367e8dc8bd92`, Qwen revision +`2fc06364715b967f1860aea9cf38778875588b17`, ilxyr commit `e92382f`, the +dependency versions, critical model-file hashes, and an eight-hour instance +lifetime guard. At the current On-Demand price of USD 1.0584/hour, eight compute +hours cost USD 8.4672; the 100 GiB gp3 volume and public IPv4 time leave the +bounded attempt below USD 10. + The larger memory-optimized host is insurance against repeating the known post-emission memory failure. The ceiling buys one formal attempt, not iterative tuning. Infrastructure setup, model download, dependency installation, and diff --git a/experiments/qwen35_acceptance/launch-manifest.json b/experiments/qwen35_acceptance/launch-manifest.json new file mode 100644 index 0000000..2f8e326 --- /dev/null +++ b/experiments/qwen35_acceptance/launch-manifest.json @@ -0,0 +1,98 @@ +{ + "schema": "lecore.qwen35-launch.v1", + "state": "authorized_prelaunch", + "attempt": { + "maximum_formal_runs": 1, + "retry_after_admission": false, + "seed": 0 + }, + "source": { + "repository": "https://github.com/atimics/holostuff.git", + "commit": "a04ab563ea159398a58a4cb002568cc78b9f14bb" + }, + "ilxyr": { + "repository": "https://github.com/cenetex/ilXyr.git", + "commit": "e92382ff2a5e8714466533a160f6609b4ef9cee8" + }, + "model": { + "repository": "Qwen/Qwen3.5-0.8B", + "revision": "2fc06364715b967f1860aea9cf38778875588b17", + "license": "Apache-2.0", + "critical_files": [ + { + "path": "model.safetensors-00001-of-00001.safetensors", + "bytes": 1746942600, + "sha256": "04b1c301231dd422b8860db31311ab2721511346a32cb1e079c4c4e5f1fe4696" + }, + { + "path": "tokenizer.json", + "bytes": 12807982, + "sha256": "5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42" + } + ], + "complete_manifest": "generated and frozen before ilxyr admission" + }, + "corpora": { + "installation": { + "handle": "git:a04ab563ea159398a58a4cb002568cc78b9f14bb:REFERENCE.md", + "license": "MIT", + "bytes": 2300089, + "sha256": "d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f" + }, + "evaluation": { + "handle": "https://www.gutenberg.org/ebooks/18", + "retrieved_at": "2026-08-12T21:32:31Z", + "last_modified": "2026-08-01T07:33:54Z", + "copyright": "Public domain in the USA", + "bytes": 1213410, + "sha256": "a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + } + }, + "method": { + "minimum_paired_positions": 4096, + "reference_logit_relative_tolerance": 0.001, + "maximum_regression": 0.01, + "confidence": 0.95, + "bootstrap": "paired moving block", + "spectral_filtering": false, + "installer_experimental": true + }, + "dependencies": { + "python": "3.12", + "numpy": "2.4.6", + "pillow": "12.3.0", + "torch": "2.11.0+cpu", + "transformers": "5.14.0", + "resolved_environment": "recorded before ilxyr admission" + }, + "aws": { + "region": "us-east-1", + "availability_zone": "us-east-1a", + "ami": "ami-07a5b367e8dc8bd92", + "ami_name": "al2023-ami-2023.12.20260803.3-kernel-6.1-x86_64", + "instance_type": "r7i.4xlarge", + "vcpu": 16, + "memory_gib": 128, + "market": "on-demand", + "hourly_compute_usd": 1.0584, + "root_volume": { + "type": "gp3", + "size_gib": 100, + "encrypted": true, + "delete_on_termination": true + }, + "inbound_ports": [], + "experiment_timeout_seconds": 21600, + "instance_lifetime_seconds": 28800, + "instance_initiated_shutdown_behavior": "terminate", + "maximum_compute_usd": 8.4672, + "total_ceiling_usd": 10.0 + }, + "publication": { + "temporary_staging": "s3://zero-training-022118847419/qwen35-acceptance/a04ab56-2fc06364-a6c9d113/", + "review_index": "https://github.com/atimics/holostuff/pull/3", + "permanent_transport": "Arweave", + "publish_all_outcomes": true, + "exclude_model_weights": true + } +} From 134cef5861b7d0e0dbf5af2b36c96a1ecada48ee Mon Sep 17 00:00:00 2001 From: atimics Date: Wed, 12 Aug 2026 14:43:05 -0700 Subject: [PATCH 04/17] docs: correct frozen lecore commit identity --- experiments/qwen35_acceptance/PROPOSAL.md | 2 +- experiments/qwen35_acceptance/launch-manifest.json | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/experiments/qwen35_acceptance/PROPOSAL.md b/experiments/qwen35_acceptance/PROPOSAL.md index a99064f..5ba2c91 100644 --- a/experiments/qwen35_acceptance/PROPOSAL.md +++ b/experiments/qwen35_acceptance/PROPOSAL.md @@ -120,7 +120,7 @@ installed checkpoint. Frozen inputs for the authorized run: - installation grounding is the MIT-licensed root `REFERENCE.md` at leCore - commit `a04ab563ea159398a58a4cb002568cc78b9f14bb`, 2,300,089 bytes with SHA-256 + commit `a04ab5692be38f06120aba4b0bc5e2a284eb2c79`, 2,300,089 bytes with SHA-256 `d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f`; - held-out evaluation is the complete Project Gutenberg plain-text edition of *The Federalist Papers*, ebook 18, retrieved 2026-08-12, 1,213,410 bytes with diff --git a/experiments/qwen35_acceptance/launch-manifest.json b/experiments/qwen35_acceptance/launch-manifest.json index 2f8e326..d77227a 100644 --- a/experiments/qwen35_acceptance/launch-manifest.json +++ b/experiments/qwen35_acceptance/launch-manifest.json @@ -8,7 +8,7 @@ }, "source": { "repository": "https://github.com/atimics/holostuff.git", - "commit": "a04ab563ea159398a58a4cb002568cc78b9f14bb" + "commit": "a04ab5692be38f06120aba4b0bc5e2a284eb2c79" }, "ilxyr": { "repository": "https://github.com/cenetex/ilXyr.git", @@ -34,7 +34,7 @@ }, "corpora": { "installation": { - "handle": "git:a04ab563ea159398a58a4cb002568cc78b9f14bb:REFERENCE.md", + "handle": "git:a04ab5692be38f06120aba4b0bc5e2a284eb2c79:REFERENCE.md", "license": "MIT", "bytes": 2300089, "sha256": "d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f" From 0d6ecfe4d2b4035c795f414aa7c253f072cabd6d Mon Sep 17 00:00:00 2001 From: atimics Date: Wed, 12 Aug 2026 15:09:09 -0700 Subject: [PATCH 05/17] results: publish qwen acceptance v1 evidence --- experiments/qwen35_acceptance/.gitattributes | 2 + .../README.md | 57 ++ .../evidence/evidence.in-toto.json | 395 +++++++++++++ .../evidence/evidence.native.json | 383 +++++++++++++ .../evidence/evidence.ro-crate.json | 286 ++++++++++ .../evidence/exporter-provenance.json | 23 + .../ledger/config.json | 5 + .../ledger/events.jsonl | 16 + ...ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b | 1 + ...8e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1 | 1 + ...ac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1 | 1 + ...5a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6 | 1 + ...3ed7b05be888419d53969be5ec4c074fb50ed8ebb7 | 1 + ...5402e1bbc04ed5df3d32aacd1f490782dc030f47f2 | 1 + ...9c9f9a267280430f5cf28135bd1db5f6d94002e57e | 1 + ...2f7142315e33556e5ca0996a23e7eca87a5ac83105 | 1 + ...3f4a8a5389ef1aee0882a428a133aed54c06960048 | 1 + ...b4014abc4bcaee45a33be7e5e467dc1cb2786fc17e | 1 + ...a53767053fd86433449e495d0b30f5d29a0f8afa16 | 1 + ...04588fa7f4169db64da5435b93d698beae6ab09ef2 | 1 + ...7eabfd0b6e5f7f4a40a6ef83ed3b49506d894026cc | 1 + ...6e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731 | 1 + ...728791cb2a29a9f5d9dfb57247a165e46a06833a47 | 1 + .../preflight-001.json | 19 + .../project/engineering-review.json | 21 + .../project/experiment-design.json | 22 + .../project/experiment.json | 241 ++++++++ .../project/forecast-empirical.json | 17 + .../project/forecast-mechanistic.json | 17 + .../project/foundation.json | 20 + .../project/funding.json | 11 + .../project/hypothesis.json | 19 + .../project/project.json | 155 ++++++ .../publication-manifest.json | 374 +++++++++++++ .../publication-receipt.json | 49 ++ .../result/admission-preflight.json | 15 + .../result/aws-instance-identity.json | 17 + .../result/cloud-init-output.log | 527 ++++++++++++++++++ .../result/corpus-manifest.json | 12 + .../result/environment.txt | 72 +++ .../result/execution-summary.json | 8 + .../result/generation.json | 1 + .../result/ilxyr-admission.json | 68 +++ .../result/ilxyr-commit.txt | 7 + .../result/ilxyr-compile.json | 3 + .../ilxyr-contribute-engineering-review.json | 3 + .../ilxyr-contribute-experiment-design.json | 3 + .../result/ilxyr-contribute-foundation.json | 3 + .../result/ilxyr-contribute-hypothesis.json | 3 + .../result/ilxyr-forecast-empirical.json | 3 + .../result/ilxyr-forecast-mechanistic.json | 3 + .../result/ilxyr-funding.json | 3 + .../result/ilxyr-init.json | 4 + .../result/ilxyr-run.json | 94 ++++ .../result/ilxyr-status.json | 121 ++++ .../result/ilxyr-verify.json | 5 + .../result/launch-manifest.json | 98 ++++ .../result/lecore-commit.txt | 7 + .../result/lecore-status.txt | 0 .../result/model-manifest.json | 66 +++ .../result/requirements-resolved.txt | 36 ++ .../result/runner-status.json | 7 + .../result/user-data.log | 464 +++++++++++++++ 63 files changed, 3800 insertions(+) create mode 100644 experiments/qwen35_acceptance/.gitattributes create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/README.md create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.in-toto.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.native.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.ro-crate.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/exporter-provenance.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/config.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/events.jsonl create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/52df3c246b4b322591efcd5402e1bbc04ed5df3d32aacd1f490782dc030f47f2 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/5b331e9b8a34f86e4dd5c69c9f9a267280430f5cf28135bd1db5f6d94002e57e create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/77a6a78f2b127ce64f97022f7142315e33556e5ca0996a23e7eca87a5ac83105 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/7ec189d139830ece3013193f4a8a5389ef1aee0882a428a133aed54c06960048 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/9b2545e28084836ecf1ca8a53767053fd86433449e495d0b30f5d29a0f8afa16 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/bcf9449a813715c1ff821104588fa7f4169db64da5435b93d698beae6ab09ef2 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/c633a2a658d6347150e9327eabfd0b6e5f7f4a40a6ef83ed3b49506d894026cc create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/edbcba5124b657e19d76e7728791cb2a29a9f5d9dfb57247a165e46a06833a47 create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/preflight-001.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/engineering-review.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/experiment-design.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/experiment.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/forecast-empirical.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/forecast-mechanistic.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/foundation.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/funding.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/hypothesis.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/project.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-receipt.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/admission-preflight.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/aws-instance-identity.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/cloud-init-output.log create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/corpus-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/environment.txt create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/execution-summary.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/generation.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-admission.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-compile.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-funding.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-init.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-run.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-status.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-verify.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/launch-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/lecore-commit.txt create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/lecore-status.txt create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/model-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/requirements-resolved.txt create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/runner-status.json create mode 100644 experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/user-data.log diff --git a/experiments/qwen35_acceptance/.gitattributes b/experiments/qwen35_acceptance/.gitattributes new file mode 100644 index 0000000..84044af --- /dev/null +++ b/experiments/qwen35_acceptance/.gitattributes @@ -0,0 +1,2 @@ +# Preserve terminal output byte-for-byte without treating progress padding as errors. +results/**/result/*.log -whitespace diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/README.md b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/README.md new file mode 100644 index 0000000..7e311ee --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/README.md @@ -0,0 +1,57 @@ +# Qwen3.5 acceptance experiment: v1 terminal record + +- Experiment: `lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance` +- ilxyr outcome: `execution_failure` +- AWS instance: `i-0c549643b970e896a` +- Formal execution: started, exited `1`, did not time out +- Ledger verification: valid (`15` objects and `16` events) +- Scientific conclusion: none + +## What happened + +The corrected AWS preflight verified the frozen source, model, tokenizer, corpora, +minimum-token contract, timeout, and disabled spectral path. ilxyr then accepted +the project after all twelve admission gates passed and started the formal run. + +The runner stopped after 51 milliseconds while importing NumPy: + +```text +ModuleNotFoundError: No module named 'numpy' +``` + +The generator had resolved the virtual environment's `python` symlink to the +system interpreter. Dependencies were installed in the virtual environment, +but ilxyr was instructed to execute `/usr/bin/python3.12`. Commit `0bda4f1` +preserves the virtual-environment interpreter path and adds a regression test. + +This terminal record does **not** contain reference-logit, installation, +bootstrap, memory, checkpoint-reload, text-generation, or vision-smoke metrics. +It cannot support an acceptance or rejection claim about Qwen or the +layer-prepending installer. + +## Evidence map + +- `project/` contains the exact ilxyr project admitted on AWS. +- `result/` contains preflight, runtime, status, and environment records. The + redistributable corpora are omitted from Git to keep the review small; their + exact hashes remain in `corpus-manifest.json` and `launch-manifest.json`. +- `ledger/` is the verified `.ilxyr` event/object store with the directory + renamed for Git publication. +- `evidence/` contains native ilxyr, RO-Crate, and in-toto exports. +- `preflight-001.json` records the earlier non-formal launch that rejected a + mistyped, nonexistent source commit before ilxyr execution. +- `publication-manifest.json` is a canonical hash-and-size index for this + directory and is generated by `../../build_publication.py`. + +The original AWS evidence bundle for this formal attempt is retained at +`s3://zero-training-022118847419/qwen35-acceptance/a04ab56-2fc06364-a6c9d113/attempt-002/qwen35-evidence.tgz` +with SHA-256 +`6182c0fd9646644d964e3d7b62a0ef46db8653b5b1e94901f03f22ea56386e11`. +The earlier preflight-only bundle has SHA-256 +`836fd6b29fa63ca6a636a3f9b3f1e66b7928605e0166e868e871a8797357e76a`. + +## Retry policy + +The v1 contract prohibited an automatic retry under the same identity. A +corrected run therefore needs explicit authorization and a new `v2` experiment +identity. This record remains published regardless of whether v2 is run. diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.in-toto.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.in-toto.json new file mode 100644 index 0000000..54df6b1 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.in-toto.json @@ -0,0 +1,395 @@ +{ + "_type": "https://in-toto.io/Statement/v1", + "predicate": { + "certificates": [], + "cold_replayable": false, + "compiled": { + "evidence_authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [ + "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e" + ], + "checker": "checker://lecore/qwen35-acceptance/v1", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "resolved_lineage": { + "engineering_review": "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "experiment_design": "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "hypothesis": "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "mathematical_foundation": "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e" + }, + "schema": "ilxyr.compiled_experiment.v1", + "source_digest": "8eaaa021df860a5ab62ef3847c0d0fc857a25ce571cba8f5492b81937d8cc030", + "spec": { + "baseline": "baseline://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "datasets": [ + "dataset://lecore/qwen-installation/d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f", + "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + ], + "evidence_authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [], + "checker": "checker://lecore/qwen35-acceptance/v1", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "execution": { + "args": [ + "/opt/qwen35-acceptance/lecore/experiments/qwen35_acceptance/run.py", + "/opt/qwen35-acceptance/model", + "/opt/qwen35-acceptance/installed-checkpoint", + "/opt/qwen35-acceptance/lecore/REFERENCE.md", + "/opt/qwen35-acceptance/inputs/federalist-papers.txt", + "--min-tokens", + "4096" + ], + "executor": "local-command", + "max_cost_credits": 100, + "network": "open", + "program": "/usr/bin/python3.12", + "timeout_seconds": 21600 + }, + "expected_outputs": [ + "metrics.acceptance_pass", + "metrics.source_clean", + "metrics.spectral_filtering_enabled", + "metrics.experimental_installer_used", + "metrics.tokenizer_parity_pass", + "metrics.reference_logit_parity_pass", + "metrics.reference_logit_relative_error", + "metrics.eval_tokens", + "metrics.original_perplexity", + "metrics.installed_perplexity", + "metrics.perplexity_delta_pct", + "metrics.paired_ci_lo_nats", + "metrics.paired_ci_hi_nats", + "metrics.statistical_gate_pass", + "metrics.paired_block_length", + "metrics.paired_effective_tokens", + "metrics.peak_rss_mb", + "metrics.peak_gpu_mb", + "metrics.emitted_checkpoint_mb", + "metrics.reload_pass", + "metrics.text_generation_pass", + "metrics.vision_smoke_pass", + "resolved_outcome", + "forecast_settlements" + ], + "funding": { + "minimum_forecasters": 2, + "minimum_total_stake": 10, + "required_compute_credits": 100 + }, + "hypothesis": "The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.", + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "lineage": { + "engineering_review": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.engineering-review", + "experiment_design": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.experiment-design", + "hypothesis": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis", + "mathematical_foundation": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation" + }, + "metrics": [ + { + "description": "All mandatory Qwen installation acceptance gates passed, encoded as 0 or 1.", + "name": "acceptance_pass", + "unit": "boolean" + }, + { + "description": "The tracked leCore checkout was clean when the run began, encoded as 0 or 1.", + "name": "source_clean", + "unit": "boolean" + }, + { + "description": "Whether research-only spectral filtering was used; the acceptance path requires 0.", + "name": "spectral_filtering_enabled", + "unit": "boolean" + }, + { + "description": "The explicitly acknowledged layer-prepending installer ran, encoded as 0 or 1.", + "name": "experimental_installer_used", + "unit": "boolean" + }, + { + "description": "leCore and the official Transformers tokenizer produced identical reference token IDs.", + "name": "tokenizer_parity_pass", + "unit": "boolean" + }, + { + "description": "Pre-install leCore logits matched the official Transformers text model within the frozen tolerance.", + "name": "reference_logit_parity_pass", + "unit": "boolean" + }, + { + "description": "Maximum absolute pre-install logit error divided by the maximum absolute reference logit.", + "name": "reference_logit_relative_error", + "unit": "ratio" + }, + { + "description": "Paired token positions included in the streamed evaluation.", + "name": "eval_tokens", + "unit": "tokens" + }, + { + "description": "Original checkpoint perplexity on the frozen corpus and chunking procedure.", + "name": "original_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed checkpoint perplexity on the same token positions.", + "name": "installed_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed minus original perplexity as a percentage of original.", + "name": "perplexity_delta_pct", + "unit": "percent" + }, + { + "description": "Lower 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_lo_nats", + "unit": "nats_per_token" + }, + { + "description": "Upper 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_hi_nats", + "unit": "nats_per_token" + }, + { + "description": "The paired upper confidence bound stayed within the preregistered maximum regression.", + "name": "statistical_gate_pass", + "unit": "boolean" + }, + { + "description": "Moving-block length inferred from autocorrelation in paired token loss differences.", + "name": "paired_block_length", + "unit": "tokens" + }, + { + "description": "Effective paired sample size after serial-correlation adjustment.", + "name": "paired_effective_tokens", + "unit": "tokens" + }, + { + "description": "Peak resident memory observed across the runner and installer child process.", + "name": "peak_rss_mb", + "unit": "megabytes" + }, + { + "description": "Peak accelerator allocation reported by PyTorch, or 0 when no accelerator was used.", + "name": "peak_gpu_mb", + "unit": "megabytes" + }, + { + "description": "Total safetensors size of the emitted installed checkpoint.", + "name": "emitted_checkpoint_mb", + "unit": "megabytes" + }, + { + "description": "The emitted checkpoint reloaded from disk and produced finite logits.", + "name": "reload_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint generated text through the official Transformers model.", + "name": "text_generation_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint accepted a synthetic image through the official Qwen vision-language processor and generated a token.", + "name": "vision_smoke_pass", + "unit": "boolean" + } + ], + "models": [ + "weight://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + ], + "outcome_contract": { + "outcomes": [ + { + "description": "Every frozen Qwen acceptance gate passed.", + "id": "accepted", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "gte", + "threshold": 1.0 + } + }, + { + "description": "The run completed but at least one acceptance gate failed.", + "id": "rejected", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "lt", + "threshold": 1.0 + } + }, + { + "description": "The runner failed or did not emit its exact metric contract.", + "id": "execution_failure", + "predicate": { + "kind": "execution_failure" + } + } + ], + "primary_metric": "acceptance_pass", + "success_outcome": "accepted" + }, + "proposer": { + "id": "model://codex/lecore-qwen/research-director", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director" + }, + "rationale": "This converts the unresolved Qwen integration claims into one frozen, replayable, statistically gated run. Spectral filtering is a separate research control and is forbidden in this contract.", + "schema": "ilxyr.experiment.v1", + "security": { + "code_policy": "arbitrary", + "export_policy": "artifacts", + "weight_class": "public" + }, + "seeds": [ + 0 + ], + "title": "Qwen3.5 leCore layer-prepending acceptance" + } + }, + "compiled_ref": "artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6", + "evidence": { + "authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [ + "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e", + "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731" + ], + "checker": "checker://lecore/qwen35-acceptance/v1", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "id": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690", + "lane": "promoted", + "metrics": {}, + "recorded_at_ms": 1786571246771, + "resolved_outcome": "execution_failure", + "run_ref": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731", + "schema": "ilxyr.evidence.v1" + }, + "evidence_event_hash": "f20843ffca03b8df2cacc86f57dcdd0b48aa839114c16953be8d9719063eaee4", + "evidence_ref": "artifact://sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1", + "executor_attestations": [], + "executor_attested": false, + "forecast_risked": true, + "forecasts": [ + { + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-mechanistic", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic", + "probabilities": { + "accepted": 0.15, + "execution_failure": 0.3, + "rejected": 0.55 + }, + "rationale": "The blank prepend has a strong identity construction, while full-checkpoint memory pressure and official multimodal reload remain material execution risks.", + "schema": "ilxyr.forecast.v1", + "stake": 5 + }, + { + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-empirical", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical", + "probabilities": { + "accepted": 0.2, + "execution_failure": 0.15, + "rejected": 0.65 + }, + "rationale": "Structural rehearsal passes, but no complete post-memory-fix real-Qwen install and powered acceptance run has succeeded yet.", + "schema": "ilxyr.forecast.v1", + "stake": 5 + } + ], + "ledger_head": "2ab7e993f5ebef3f6635b08fb4a7f9f067de2e254668f12378f7882ba1b0907f", + "run": { + "completed_at_ms": 1786571246741, + "exit_code": 1, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "id": "run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690", + "metrics": {}, + "output_truncated": false, + "schema": "ilxyr.run.v1", + "started_at_ms": 1786571246690, + "stderr": "Traceback (most recent call last):\n File \"/opt/qwen35-acceptance/lecore/experiments/qwen35_acceptance/run.py\", line 17, in \n import numpy as np\nModuleNotFoundError: No module named 'numpy'\n", + "stdout": "", + "timed_out": false + }, + "run_ref": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731", + "schema": "ilxyr.evidence_bundle.v1", + "settlements": [ + { + "brier_score": 0.815, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic", + "resolved_outcome": "execution_failure", + "schema": "ilxyr.forecast_settlement.v1", + "stake": 5 + }, + { + "brier_score": 1.185, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical", + "resolved_outcome": "execution_failure", + "schema": "ilxyr.forecast_settlement.v1", + "stake": 5 + } + ], + "source_attested": false + }, + "predicateType": "https://ilxyr.dev/attestations/evidence/v1", + "subject": [ + { + "digest": { + "sha256": "22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1" + }, + "name": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690" + } + ] +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.native.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.native.json new file mode 100644 index 0000000..2d6e895 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.native.json @@ -0,0 +1,383 @@ +{ + "certificates": [], + "cold_replayable": false, + "compiled": { + "evidence_authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [ + "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e" + ], + "checker": "checker://lecore/qwen35-acceptance/v1", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "resolved_lineage": { + "engineering_review": "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "experiment_design": "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "hypothesis": "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "mathematical_foundation": "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e" + }, + "schema": "ilxyr.compiled_experiment.v1", + "source_digest": "8eaaa021df860a5ab62ef3847c0d0fc857a25ce571cba8f5492b81937d8cc030", + "spec": { + "baseline": "baseline://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "datasets": [ + "dataset://lecore/qwen-installation/d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f", + "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + ], + "evidence_authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [], + "checker": "checker://lecore/qwen35-acceptance/v1", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "execution": { + "args": [ + "/opt/qwen35-acceptance/lecore/experiments/qwen35_acceptance/run.py", + "/opt/qwen35-acceptance/model", + "/opt/qwen35-acceptance/installed-checkpoint", + "/opt/qwen35-acceptance/lecore/REFERENCE.md", + "/opt/qwen35-acceptance/inputs/federalist-papers.txt", + "--min-tokens", + "4096" + ], + "executor": "local-command", + "max_cost_credits": 100, + "network": "open", + "program": "/usr/bin/python3.12", + "timeout_seconds": 21600 + }, + "expected_outputs": [ + "metrics.acceptance_pass", + "metrics.source_clean", + "metrics.spectral_filtering_enabled", + "metrics.experimental_installer_used", + "metrics.tokenizer_parity_pass", + "metrics.reference_logit_parity_pass", + "metrics.reference_logit_relative_error", + "metrics.eval_tokens", + "metrics.original_perplexity", + "metrics.installed_perplexity", + "metrics.perplexity_delta_pct", + "metrics.paired_ci_lo_nats", + "metrics.paired_ci_hi_nats", + "metrics.statistical_gate_pass", + "metrics.paired_block_length", + "metrics.paired_effective_tokens", + "metrics.peak_rss_mb", + "metrics.peak_gpu_mb", + "metrics.emitted_checkpoint_mb", + "metrics.reload_pass", + "metrics.text_generation_pass", + "metrics.vision_smoke_pass", + "resolved_outcome", + "forecast_settlements" + ], + "funding": { + "minimum_forecasters": 2, + "minimum_total_stake": 10, + "required_compute_credits": 100 + }, + "hypothesis": "The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.", + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "lineage": { + "engineering_review": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.engineering-review", + "experiment_design": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.experiment-design", + "hypothesis": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis", + "mathematical_foundation": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation" + }, + "metrics": [ + { + "description": "All mandatory Qwen installation acceptance gates passed, encoded as 0 or 1.", + "name": "acceptance_pass", + "unit": "boolean" + }, + { + "description": "The tracked leCore checkout was clean when the run began, encoded as 0 or 1.", + "name": "source_clean", + "unit": "boolean" + }, + { + "description": "Whether research-only spectral filtering was used; the acceptance path requires 0.", + "name": "spectral_filtering_enabled", + "unit": "boolean" + }, + { + "description": "The explicitly acknowledged layer-prepending installer ran, encoded as 0 or 1.", + "name": "experimental_installer_used", + "unit": "boolean" + }, + { + "description": "leCore and the official Transformers tokenizer produced identical reference token IDs.", + "name": "tokenizer_parity_pass", + "unit": "boolean" + }, + { + "description": "Pre-install leCore logits matched the official Transformers text model within the frozen tolerance.", + "name": "reference_logit_parity_pass", + "unit": "boolean" + }, + { + "description": "Maximum absolute pre-install logit error divided by the maximum absolute reference logit.", + "name": "reference_logit_relative_error", + "unit": "ratio" + }, + { + "description": "Paired token positions included in the streamed evaluation.", + "name": "eval_tokens", + "unit": "tokens" + }, + { + "description": "Original checkpoint perplexity on the frozen corpus and chunking procedure.", + "name": "original_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed checkpoint perplexity on the same token positions.", + "name": "installed_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed minus original perplexity as a percentage of original.", + "name": "perplexity_delta_pct", + "unit": "percent" + }, + { + "description": "Lower 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_lo_nats", + "unit": "nats_per_token" + }, + { + "description": "Upper 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_hi_nats", + "unit": "nats_per_token" + }, + { + "description": "The paired upper confidence bound stayed within the preregistered maximum regression.", + "name": "statistical_gate_pass", + "unit": "boolean" + }, + { + "description": "Moving-block length inferred from autocorrelation in paired token loss differences.", + "name": "paired_block_length", + "unit": "tokens" + }, + { + "description": "Effective paired sample size after serial-correlation adjustment.", + "name": "paired_effective_tokens", + "unit": "tokens" + }, + { + "description": "Peak resident memory observed across the runner and installer child process.", + "name": "peak_rss_mb", + "unit": "megabytes" + }, + { + "description": "Peak accelerator allocation reported by PyTorch, or 0 when no accelerator was used.", + "name": "peak_gpu_mb", + "unit": "megabytes" + }, + { + "description": "Total safetensors size of the emitted installed checkpoint.", + "name": "emitted_checkpoint_mb", + "unit": "megabytes" + }, + { + "description": "The emitted checkpoint reloaded from disk and produced finite logits.", + "name": "reload_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint generated text through the official Transformers model.", + "name": "text_generation_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint accepted a synthetic image through the official Qwen vision-language processor and generated a token.", + "name": "vision_smoke_pass", + "unit": "boolean" + } + ], + "models": [ + "weight://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + ], + "outcome_contract": { + "outcomes": [ + { + "description": "Every frozen Qwen acceptance gate passed.", + "id": "accepted", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "gte", + "threshold": 1.0 + } + }, + { + "description": "The run completed but at least one acceptance gate failed.", + "id": "rejected", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "lt", + "threshold": 1.0 + } + }, + { + "description": "The runner failed or did not emit its exact metric contract.", + "id": "execution_failure", + "predicate": { + "kind": "execution_failure" + } + } + ], + "primary_metric": "acceptance_pass", + "success_outcome": "accepted" + }, + "proposer": { + "id": "model://codex/lecore-qwen/research-director", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director" + }, + "rationale": "This converts the unresolved Qwen integration claims into one frozen, replayable, statistically gated run. Spectral filtering is a separate research control and is forbidden in this contract.", + "schema": "ilxyr.experiment.v1", + "security": { + "code_policy": "arbitrary", + "export_policy": "artifacts", + "weight_class": "public" + }, + "seeds": [ + 0 + ], + "title": "Qwen3.5 leCore layer-prepending acceptance" + } + }, + "compiled_ref": "artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6", + "evidence": { + "authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [ + "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e", + "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731" + ], + "checker": "checker://lecore/qwen35-acceptance/v1", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "id": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690", + "lane": "promoted", + "metrics": {}, + "recorded_at_ms": 1786571246771, + "resolved_outcome": "execution_failure", + "run_ref": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731", + "schema": "ilxyr.evidence.v1" + }, + "evidence_event_hash": "f20843ffca03b8df2cacc86f57dcdd0b48aa839114c16953be8d9719063eaee4", + "evidence_ref": "artifact://sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1", + "executor_attestations": [], + "executor_attested": false, + "forecast_risked": true, + "forecasts": [ + { + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-mechanistic", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic", + "probabilities": { + "accepted": 0.15, + "execution_failure": 0.3, + "rejected": 0.55 + }, + "rationale": "The blank prepend has a strong identity construction, while full-checkpoint memory pressure and official multimodal reload remain material execution risks.", + "schema": "ilxyr.forecast.v1", + "stake": 5 + }, + { + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-empirical", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical", + "probabilities": { + "accepted": 0.2, + "execution_failure": 0.15, + "rejected": 0.65 + }, + "rationale": "Structural rehearsal passes, but no complete post-memory-fix real-Qwen install and powered acceptance run has succeeded yet.", + "schema": "ilxyr.forecast.v1", + "stake": 5 + } + ], + "ledger_head": "2ab7e993f5ebef3f6635b08fb4a7f9f067de2e254668f12378f7882ba1b0907f", + "run": { + "completed_at_ms": 1786571246741, + "exit_code": 1, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "id": "run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690", + "metrics": {}, + "output_truncated": false, + "schema": "ilxyr.run.v1", + "started_at_ms": 1786571246690, + "stderr": "Traceback (most recent call last):\n File \"/opt/qwen35-acceptance/lecore/experiments/qwen35_acceptance/run.py\", line 17, in \n import numpy as np\nModuleNotFoundError: No module named 'numpy'\n", + "stdout": "", + "timed_out": false + }, + "run_ref": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731", + "schema": "ilxyr.evidence_bundle.v1", + "settlements": [ + { + "brier_score": 0.815, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic", + "resolved_outcome": "execution_failure", + "schema": "ilxyr.forecast_settlement.v1", + "stake": 5 + }, + { + "brier_score": 1.185, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical", + "resolved_outcome": "execution_failure", + "schema": "ilxyr.forecast_settlement.v1", + "stake": 5 + } + ], + "source_attested": false +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.ro-crate.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.ro-crate.json new file mode 100644 index 0000000..298fbbe --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/evidence.ro-crate.json @@ -0,0 +1,286 @@ +{ + "@context": [ + "https://w3id.org/ro/crate/1.3/context", + { + "ilxyr": "https://ilxyr.dev/ns#", + "prov": "http://www.w3.org/ns/prov#" + } + ], + "@graph": [ + { + "@id": "ro-crate-metadata.json", + "@type": "CreativeWork", + "about": { + "@id": "./" + }, + "conformsTo": { + "@id": "https://w3id.org/ro/crate/1.3" + } + }, + { + "@id": "./", + "@type": [ + "Dataset", + "prov:Entity" + ], + "description": "A ledger-verified ilXyr evidence bundle exported for discovery, preservation, and reuse.", + "hasPart": [ + { + "@id": "artifact://sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1" + }, + { + "@id": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731" + }, + { + "@id": "artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6" + }, + { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic" + }, + { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical" + }, + { + "@id": "urn:ilxyr:settlement:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic" + }, + { + "@id": "urn:ilxyr:settlement:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical" + }, + { + "@id": "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b" + }, + { + "@id": "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1" + }, + { + "@id": "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7" + }, + { + "@id": "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e" + }, + { + "@id": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731" + } + ], + "ilxyr:ledgerHead": "2ab7e993f5ebef3f6635b08fb4a7f9f067de2e254668f12378f7882ba1b0907f", + "mainEntity": { + "@id": "artifact://sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1" + }, + "name": "ilXyr evidence bundle — Qwen3.5 leCore layer-prepending acceptance" + }, + { + "@id": "artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6", + "@type": [ + "CreativeWork", + "prov:Plan" + ], + "author": { + "@id": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director", + "@type": "SoftwareApplication", + "identifier": "model://codex/lecore-qwen/research-director" + }, + "description": "The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.", + "identifier": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "ilxyr:sharedTask": null, + "ilxyr:sourceDigest": "8eaaa021df860a5ab62ef3847c0d0fc857a25ce571cba8f5492b81937d8cc030", + "isBasedOn": [ + { + "@id": "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b" + }, + { + "@id": "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1" + }, + { + "@id": "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7" + }, + { + "@id": "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e" + } + ], + "name": "Qwen3.5 leCore layer-prepending acceptance" + }, + { + "@id": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731", + "@type": [ + "CreateAction", + "prov:Activity" + ], + "actionStatus": "CompletedActionStatus", + "ilxyr:run": { + "completed_at_ms": 1786571246741, + "exit_code": 1, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "id": "run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690", + "metrics": {}, + "output_truncated": false, + "schema": "ilxyr.run.v1", + "started_at_ms": 1786571246690, + "stderr": "Traceback (most recent call last):\n File \"/opt/qwen35-acceptance/lecore/experiments/qwen35_acceptance/run.py\", line 17, in \n import numpy as np\nModuleNotFoundError: No module named 'numpy'\n", + "stdout": "", + "timed_out": false + }, + "ilxyr:sourceAttested": false, + "name": "Execution for lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "object": { + "@id": "artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6" + }, + "result": { + "@id": "artifact://sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1" + } + }, + { + "@id": "artifact://sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1", + "@type": [ + "Dataset", + "prov:Entity" + ], + "additionalProperty": [], + "description": "Resolved outcome: execution_failure", + "identifier": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690", + "ilxyr:authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [ + "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e", + "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731" + ], + "checker": "checker://lecore/qwen35-acceptance/v1", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "ilxyr:coldReplayable": false, + "ilxyr:evidenceEventHash": "f20843ffca03b8df2cacc86f57dcdd0b48aa839114c16953be8d9719063eaee4", + "ilxyr:evidenceLane": "promoted", + "ilxyr:executorAttested": false, + "ilxyr:externallyPreregistered": false, + "ilxyr:forecastRisked": true, + "isBasedOn": { + "@id": "artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6" + }, + "name": "Evidence for Qwen3.5 leCore layer-prepending acceptance", + "prov:wasGeneratedBy": { + "@id": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731" + } + }, + { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic", + "@type": [ + "CreativeWork", + "prov:Entity" + ], + "about": { + "@id": "artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6" + }, + "author": { + "@id": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic", + "@type": "SoftwareApplication", + "identifier": "model://codex/lecore-qwen/forecaster-mechanistic" + }, + "identifier": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic", + "ilxyr:probabilities": { + "accepted": 0.15, + "execution_failure": 0.3, + "rejected": 0.55 + }, + "ilxyr:stake": 5, + "name": "Forecast for lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance" + }, + { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical", + "@type": [ + "CreativeWork", + "prov:Entity" + ], + "about": { + "@id": "artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6" + }, + "author": { + "@id": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical", + "@type": "SoftwareApplication", + "identifier": "model://codex/lecore-qwen/forecaster-empirical" + }, + "identifier": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical", + "ilxyr:probabilities": { + "accepted": 0.2, + "execution_failure": 0.15, + "rejected": 0.65 + }, + "ilxyr:stake": 5, + "name": "Forecast for lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance" + }, + { + "@id": "urn:ilxyr:settlement:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic", + "@type": [ + "CreateAction", + "prov:Activity" + ], + "ilxyr:brierScore": 0.815, + "ilxyr:resolvedOutcome": "execution_failure", + "name": "Forecast settlement for lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic", + "object": { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic" + }, + "result": { + "@id": "artifact://sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1" + } + }, + { + "@id": "urn:ilxyr:settlement:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical", + "@type": [ + "CreateAction", + "prov:Activity" + ], + "ilxyr:brierScore": 1.185, + "ilxyr:resolvedOutcome": "execution_failure", + "name": "Forecast settlement for lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical", + "object": { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical" + }, + "result": { + "@id": "artifact://sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1" + } + }, + { + "@id": "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "@type": [ + "MediaObject", + "prov:Entity" + ], + "sha256": "04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b" + }, + { + "@id": "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "@type": [ + "MediaObject", + "prov:Entity" + ], + "sha256": "3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1" + }, + { + "@id": "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "@type": [ + "MediaObject", + "prov:Entity" + ], + "sha256": "50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7" + }, + { + "@id": "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e", + "@type": [ + "MediaObject", + "prov:Entity" + ], + "sha256": "826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e" + } + ] +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/exporter-provenance.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/exporter-provenance.json new file mode 100644 index 0000000..3645f5f --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/evidence/exporter-provenance.json @@ -0,0 +1,23 @@ +{ + "schema": "lecore.ilxyr-exporter-provenance.v1", + "export_mode": "read_only_isolated_snapshot", + "source_repository": "/Users/ratimics/develop/ilxyr", + "source_base_commit": "e92382ff2a5e8714466533a160f6609b4ef9cee8", + "source_worktree_clean": false, + "source_snapshot": "/private/tmp/ilxyr-exporter-snapshot", + "source_snapshot_note": "The exporter was compiled from an isolated copy of the user's existing dirty ilxyr worktree. The original worktree and the verified AWS ledger were not modified.", + "exporter_binary": { + "path": "/private/tmp/ilxyr-exporter-snapshot/target/debug/ilxyr", + "sha256": "628c153cfe64a1b916551194f7cc41fb5dbc13954a7d2b9a956641b71e97bd29" + }, + "source_files": { + "Cargo.lock": "4d5a64a9ce286006325c912c8e3e716c427e50e01cc28d72d72d9978d1a7b42b", + "crates/ilxyr-core/src/interop.rs": "b3db6eb51891abecb241bc38751b78d8eb67a4245aed7d28e514c858223d9982", + "crates/ilxyr-cli/src/main.rs": "e3c3d8d7364fbe22af4b927f7b01c58b57f68480e242382c03ffd7854f9d71cb" + }, + "exports": { + "evidence.native.json": "101590b922f3877f88971073171fa622d29af62cad52dca1990e3100d122c204", + "evidence.ro-crate.json": "7758207b7bf7407e2e3ddb526aab3f049c4f2579702700b733e740a3a44dbd9d", + "evidence.in-toto.json": "f9bc730a5ade64659e4e523f115cfc71f2141757cecdbb8a520f87a36a83cc52" + } +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/config.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/config.json new file mode 100644 index 0000000..c7becc4 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/config.json @@ -0,0 +1,5 @@ +{ + "ledger_mode": "single_writer", + "object_hash": "sha256", + "schema": "ilxyr.workspace.v1" +} \ No newline at end of file diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/events.jsonl b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/events.jsonl new file mode 100644 index 0000000..a225e94 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/events.jsonl @@ -0,0 +1,16 @@ +{"actor":{"id":"model://codex/lecore-qwen/research-director","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis","artifact_ref":"artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7","event_hash":"0b565c2cbc2b0f891844b9c2576aee30f141d55c9f031ea3008eb51fdbe7eb78","event_type":"ContributionSubmitted","occurred_at_ms":1786571245302,"schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/statistical-reviewer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/statistical-reviewer"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation","artifact_ref":"artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e","event_hash":"3c94b092a15c882a2d0262cd2c4edeb0230b7aed9e2378e3bab949c3b4adbb5e","event_type":"ContributionSubmitted","occurred_at_ms":1786571245405,"previous_event":"0b565c2cbc2b0f891844b9c2576aee30f141d55c9f031ea3008eb51fdbe7eb78","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/engineering-reviewer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/engineering-reviewer"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.engineering-review","artifact_ref":"artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b","event_hash":"f2fa6d5b6a47c905938c64b6d48966028f99577a47d64dc5b71629c2f17a49af","event_type":"ContributionSubmitted","occurred_at_ms":1786571245489,"previous_event":"3c94b092a15c882a2d0262cd2c4edeb0230b7aed9e2378e3bab949c3b4adbb5e","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/experiment-designer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/experiment-designer"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.experiment-design","artifact_ref":"artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1","event_hash":"e01196abc3d568a3bc0c059f8f74a3d66208157dc921e53570bc534d89d0a7b0","event_type":"ContributionSubmitted","occurred_at_ms":1786571245558,"previous_event":"f2fa6d5b6a47c905938c64b6d48966028f99577a47d64dc5b71629c2f17a49af","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/research-director","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","artifact_ref":"artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6","event_hash":"5001b7d1d25a2e080534da520322651e3f2076046fb0a87797699267d7875338","event_type":"ExperimentCompiled","occurred_at_ms":1786571245672,"previous_event":"e01196abc3d568a3bc0c059f8f74a3d66208157dc921e53570bc534d89d0a7b0","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/forecaster-empirical","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical","artifact_ref":"artifact://sha256/52df3c246b4b322591efcd5402e1bbc04ed5df3d32aacd1f490782dc030f47f2","event_hash":"9667c8002a53de3899f6f4e0b85e478e7ad4cedf559af406645ae4073084486c","event_type":"ForecastSubmitted","occurred_at_ms":1786571245789,"previous_event":"5001b7d1d25a2e080534da520322651e3f2076046fb0a87797699267d7875338","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/forecaster-mechanistic","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic","artifact_ref":"artifact://sha256/bcf9449a813715c1ff821104588fa7f4169db64da5435b93d698beae6ab09ef2","event_hash":"79559fff0ebc6b42d5d2a2f24e4fadb8225402d4164b331d9fbc1ef25db680f9","event_type":"ForecastSubmitted","occurred_at_ms":1786571245890,"previous_event":"9667c8002a53de3899f6f4e0b85e478e7ad4cedf559af406645ae4073084486c","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://lecore/qwen-experiment-generator","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.funding","artifact_ref":"artifact://sha256/edbcba5124b657e19d76e7728791cb2a29a9f5d9dfb57247a165e46a06833a47","event_hash":"319fe314f0b0edd21ffb3b4ca6d4bddcf5b9c92e7e3596f81226cb169b9b1c9f","event_type":"FundingCommitted","occurred_at_ms":1786571245974,"previous_event":"79559fff0ebc6b42d5d2a2f24e4fadb8225402d4164b331d9fbc1ef25db680f9","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/admission-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","artifact_ref":"artifact://sha256/7ec189d139830ece3013193f4a8a5389ef1aee0882a428a133aed54c06960048","event_hash":"f0bb9ea96a1680720532ef639e536803349eaf832db18020c2e0ebf84033fb3e","event_type":"AdmissionDecided","occurred_at_ms":1786571246079,"previous_event":"319fe314f0b0edd21ffb3b4ca6d4bddcf5b9c92e7e3596f81226cb169b9b1c9f","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/local-executor-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","event_hash":"e2bbd226d89bc750593ee9af8f1add73fd809560f251ba197b997a9ddce2a2e8","event_type":"ExecutionStarted","occurred_at_ms":1786571246674,"previous_event":"f0bb9ea96a1680720532ef639e536803349eaf832db18020c2e0ebf84033fb3e","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/local-executor-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","artifact_ref":"artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731","event_hash":"0208e40694b266d0b476344206688686461889dca2381bd411d6b6ed08cd06f7","event_type":"ExperimentCompleted","occurred_at_ms":1786571246759,"previous_event":"e2bbd226d89bc750593ee9af8f1add73fd809560f251ba197b997a9ddce2a2e8","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/resolver-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","artifact_ref":"artifact://sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1","event_hash":"f20843ffca03b8df2cacc86f57dcdd0b48aa839114c16953be8d9719063eaee4","event_type":"EvidenceRecorded","occurred_at_ms":1786571246786,"previous_event":"0208e40694b266d0b476344206688686461889dca2381bd411d6b6ed08cd06f7","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/scoring-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical","artifact_ref":"artifact://sha256/c633a2a658d6347150e9327eabfd0b6e5f7f4a40a6ef83ed3b49506d894026cc","event_hash":"2e6fe16202a6f4035cca7ef2219e89ea0d14992b2a4b173244376310ae835b77","event_type":"ForecastSettled","occurred_at_ms":1786571246828,"previous_event":"f20843ffca03b8df2cacc86f57dcdd0b48aa839114c16953be8d9719063eaee4","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/scoring-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic","artifact_ref":"artifact://sha256/77a6a78f2b127ce64f97022f7142315e33556e5ca0996a23e7eca87a5ac83105","event_hash":"de86d60bbea6e5a1940551efdd129d27cf35a8e94b954a5656684a6f3f73decb","event_type":"ForecastSettled","occurred_at_ms":1786571246900,"previous_event":"2e6fe16202a6f4035cca7ef2219e89ea0d14992b2a4b173244376310ae835b77","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/calibration-v1","kind":"service"},"aggregate_id":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical","artifact_ref":"artifact://sha256/9b2545e28084836ecf1ca8a53767053fd86433449e495d0b30f5d29a0f8afa16","event_hash":"c1b2c93cb94fae7df7baf9b3d5d79e4ac88ac7669232eac34924fd77e6f2be75","event_type":"CalibrationUpdated","occurred_at_ms":1786571247013,"previous_event":"de86d60bbea6e5a1940551efdd129d27cf35a8e94b954a5656684a6f3f73decb","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/calibration-v1","kind":"service"},"aggregate_id":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic","artifact_ref":"artifact://sha256/5b331e9b8a34f86e4dd5c69c9f9a267280430f5cf28135bd1db5f6d94002e57e","event_hash":"2ab7e993f5ebef3f6635b08fb4a7f9f067de2e254668f12378f7882ba1b0907f","event_type":"CalibrationUpdated","occurred_at_ms":1786571247123,"previous_event":"c1b2c93cb94fae7df7baf9b3d5d79e4ac88ac7669232eac34924fd77e6f2be75","schema":"ilxyr.event.v1"} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b new file mode 100644 index 0000000..7b0e257 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b @@ -0,0 +1 @@ +{"actor":{"id":"model://codex/lecore-qwen/engineering-reviewer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/engineering-reviewer"},"body":"The shell-free runner uses absolute paths, records the exact leCore commit and checker hashes, keeps spectral filtering disabled, invokes the layer-prepending path only with its experimental acknowledgement, records peak memory, reloads the emitted artifact, and exercises the official Transformers text and image-text interfaces.","claims":["The emitted stdout is exactly the ilxyr metrics/source envelope.","Installer logs and a human-readable metrics artifact are retained beside the output checkpoint."],"confidence":0.85,"id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.engineering-review","input_refs":["lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis","lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation"],"schema":"ilxyr.contribution.v1","stage":"engineering_review","title":"Qwen acceptance runner and provenance boundary"} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1 new file mode 100644 index 0000000..3f081da --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1 @@ -0,0 +1 @@ +{"authority":{"level":"corpus_proxy","provenance":{"artifact_hashes":["artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b","artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1","artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7","artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e","artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731"],"checker":"checker://lecore/qwen35-acceptance/v1","model_lineage":"model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680"},"scope":{"coverage":1.0,"eval_set":"dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94","seeds":[0]}},"experiment_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","id":"evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690","lane":"promoted","metrics":{},"recorded_at_ms":1786571246771,"resolved_outcome":"execution_failure","run_ref":"artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731","schema":"ilxyr.evidence.v1"} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1 new file mode 100644 index 0000000..25e82d8 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1 @@ -0,0 +1 @@ +{"actor":{"id":"model://codex/lecore-qwen/experiment-designer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/experiment-designer"},"body":"Execute once against the content-bound public checkpoint, installation corpus, and separate held-out evaluation corpus. Do not tune thresholds or replace either corpus after admission. Resolve accepted only when source cleanliness, tokenizer parity, reference-logit parity, the paired statistical gate, disk reload, official text generation, and official vision smoke all pass. A cleanly executed no-go is preserved as rejected evidence.","claims":["Accepted and rejected are exhaustive for a valid metrics envelope.","Runtime or dependency failure resolves separately as execution_failure."],"confidence":0.95,"id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.experiment-design","input_refs":["lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis","lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation","lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.engineering-review"],"schema":"ilxyr.contribution.v1","stage":"experiment_design","title":"One-shot Qwen3.5 installation acceptance run"} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6 new file mode 100644 index 0000000..7f98cb2 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6 @@ -0,0 +1 @@ +{"evidence_authority":{"level":"corpus_proxy","provenance":{"artifact_hashes":["artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b","artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1","artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7","artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e"],"checker":"checker://lecore/qwen35-acceptance/v1","model_lineage":"model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680"},"scope":{"coverage":1.0,"eval_set":"dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94","seeds":[0]}},"resolved_lineage":{"engineering_review":"artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b","experiment_design":"artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1","hypothesis":"artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7","mathematical_foundation":"artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e"},"schema":"ilxyr.compiled_experiment.v1","source_digest":"8eaaa021df860a5ab62ef3847c0d0fc857a25ce571cba8f5492b81937d8cc030","spec":{"baseline":"baseline://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680","datasets":["dataset://lecore/qwen-installation/d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f","dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94"],"evidence_authority":{"level":"corpus_proxy","provenance":{"artifact_hashes":[],"checker":"checker://lecore/qwen35-acceptance/v1","model_lineage":"model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680"},"scope":{"coverage":1.0,"eval_set":"dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94","seeds":[0]}},"execution":{"args":["/opt/qwen35-acceptance/lecore/experiments/qwen35_acceptance/run.py","/opt/qwen35-acceptance/model","/opt/qwen35-acceptance/installed-checkpoint","/opt/qwen35-acceptance/lecore/REFERENCE.md","/opt/qwen35-acceptance/inputs/federalist-papers.txt","--min-tokens","4096"],"executor":"local-command","max_cost_credits":100,"network":"open","program":"/usr/bin/python3.12","timeout_seconds":21600},"expected_outputs":["metrics.acceptance_pass","metrics.source_clean","metrics.spectral_filtering_enabled","metrics.experimental_installer_used","metrics.tokenizer_parity_pass","metrics.reference_logit_parity_pass","metrics.reference_logit_relative_error","metrics.eval_tokens","metrics.original_perplexity","metrics.installed_perplexity","metrics.perplexity_delta_pct","metrics.paired_ci_lo_nats","metrics.paired_ci_hi_nats","metrics.statistical_gate_pass","metrics.paired_block_length","metrics.paired_effective_tokens","metrics.peak_rss_mb","metrics.peak_gpu_mb","metrics.emitted_checkpoint_mb","metrics.reload_pass","metrics.text_generation_pass","metrics.vision_smoke_pass","resolved_outcome","forecast_settlements"],"funding":{"minimum_forecasters":2,"minimum_total_stake":10,"required_compute_credits":100},"hypothesis":"The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.","id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","lineage":{"engineering_review":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.engineering-review","experiment_design":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.experiment-design","hypothesis":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis","mathematical_foundation":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation"},"metrics":[{"description":"All mandatory Qwen installation acceptance gates passed, encoded as 0 or 1.","name":"acceptance_pass","unit":"boolean"},{"description":"The tracked leCore checkout was clean when the run began, encoded as 0 or 1.","name":"source_clean","unit":"boolean"},{"description":"Whether research-only spectral filtering was used; the acceptance path requires 0.","name":"spectral_filtering_enabled","unit":"boolean"},{"description":"The explicitly acknowledged layer-prepending installer ran, encoded as 0 or 1.","name":"experimental_installer_used","unit":"boolean"},{"description":"leCore and the official Transformers tokenizer produced identical reference token IDs.","name":"tokenizer_parity_pass","unit":"boolean"},{"description":"Pre-install leCore logits matched the official Transformers text model within the frozen tolerance.","name":"reference_logit_parity_pass","unit":"boolean"},{"description":"Maximum absolute pre-install logit error divided by the maximum absolute reference logit.","name":"reference_logit_relative_error","unit":"ratio"},{"description":"Paired token positions included in the streamed evaluation.","name":"eval_tokens","unit":"tokens"},{"description":"Original checkpoint perplexity on the frozen corpus and chunking procedure.","name":"original_perplexity","unit":"perplexity"},{"description":"Installed checkpoint perplexity on the same token positions.","name":"installed_perplexity","unit":"perplexity"},{"description":"Installed minus original perplexity as a percentage of original.","name":"perplexity_delta_pct","unit":"percent"},{"description":"Lower 95 percent paired moving-block-bootstrap bound for installed minus original NLL.","name":"paired_ci_lo_nats","unit":"nats_per_token"},{"description":"Upper 95 percent paired moving-block-bootstrap bound for installed minus original NLL.","name":"paired_ci_hi_nats","unit":"nats_per_token"},{"description":"The paired upper confidence bound stayed within the preregistered maximum regression.","name":"statistical_gate_pass","unit":"boolean"},{"description":"Moving-block length inferred from autocorrelation in paired token loss differences.","name":"paired_block_length","unit":"tokens"},{"description":"Effective paired sample size after serial-correlation adjustment.","name":"paired_effective_tokens","unit":"tokens"},{"description":"Peak resident memory observed across the runner and installer child process.","name":"peak_rss_mb","unit":"megabytes"},{"description":"Peak accelerator allocation reported by PyTorch, or 0 when no accelerator was used.","name":"peak_gpu_mb","unit":"megabytes"},{"description":"Total safetensors size of the emitted installed checkpoint.","name":"emitted_checkpoint_mb","unit":"megabytes"},{"description":"The emitted checkpoint reloaded from disk and produced finite logits.","name":"reload_pass","unit":"boolean"},{"description":"The emitted checkpoint generated text through the official Transformers model.","name":"text_generation_pass","unit":"boolean"},{"description":"The emitted checkpoint accepted a synthetic image through the official Qwen vision-language processor and generated a token.","name":"vision_smoke_pass","unit":"boolean"}],"models":["weight://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680"],"outcome_contract":{"outcomes":[{"description":"Every frozen Qwen acceptance gate passed.","id":"accepted","predicate":{"kind":"metric","metric":"acceptance_pass","operator":"gte","threshold":1.0}},{"description":"The run completed but at least one acceptance gate failed.","id":"rejected","predicate":{"kind":"metric","metric":"acceptance_pass","operator":"lt","threshold":1.0}},{"description":"The runner failed or did not emit its exact metric contract.","id":"execution_failure","predicate":{"kind":"execution_failure"}}],"primary_metric":"acceptance_pass","success_outcome":"accepted"},"proposer":{"id":"model://codex/lecore-qwen/research-director","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director"},"rationale":"This converts the unresolved Qwen integration claims into one frozen, replayable, statistically gated run. Spectral filtering is a separate research control and is forbidden in this contract.","schema":"ilxyr.experiment.v1","security":{"code_policy":"arbitrary","export_policy":"artifacts","weight_class":"public"},"seeds":[0],"title":"Qwen3.5 leCore layer-prepending acceptance"}} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7 new file mode 100644 index 0000000..a5deaa8 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7 @@ -0,0 +1 @@ +{"actor":{"id":"model://codex/lecore-qwen/research-director","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director"},"body":"The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.","claims":["Pre-install leCore logits match the official reference implementation.","The installed checkpoint reloads and remains within the paired regression budget.","Official Transformers text generation and vision input both remain operational."],"confidence":0.35,"id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis","input_refs":[],"schema":"ilxyr.contribution.v1","stage":"hypothesis","title":"Qwen3.5 can carry a reloadable leCore layer-prepending installation"} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/52df3c246b4b322591efcd5402e1bbc04ed5df3d32aacd1f490782dc030f47f2 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/52df3c246b4b322591efcd5402e1bbc04ed5df3d32aacd1f490782dc030f47f2 new file mode 100644 index 0000000..b1ed2ba --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/52df3c246b4b322591efcd5402e1bbc04ed5df3d32aacd1f490782dc030f47f2 @@ -0,0 +1 @@ +{"experiment_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","forecaster":{"id":"model://codex/lecore-qwen/forecaster-empirical","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical"},"id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical","probabilities":{"accepted":0.2,"execution_failure":0.15,"rejected":0.65},"rationale":"Structural rehearsal passes, but no complete post-memory-fix real-Qwen install and powered acceptance run has succeeded yet.","schema":"ilxyr.forecast.v1","stake":5} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/5b331e9b8a34f86e4dd5c69c9f9a267280430f5cf28135bd1db5f6d94002e57e b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/5b331e9b8a34f86e4dd5c69c9f9a267280430f5cf28135bd1db5f6d94002e57e new file mode 100644 index 0000000..ee4740a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/5b331e9b8a34f86e4dd5c69c9f9a267280430f5cf28135bd1db5f6d94002e57e @@ -0,0 +1 @@ +{"brier_score":0.815,"forecast_ids":["lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic"],"forecasts_settled":1,"handle":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic","last_settlement_at_ms":1786571246900,"probationary":true,"reliability":0.815,"resolution":0.0,"schema":"ilxyr.calibration_record.v1","uncertainty":0.0} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/77a6a78f2b127ce64f97022f7142315e33556e5ca0996a23e7eca87a5ac83105 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/77a6a78f2b127ce64f97022f7142315e33556e5ca0996a23e7eca87a5ac83105 new file mode 100644 index 0000000..8d35b26 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/77a6a78f2b127ce64f97022f7142315e33556e5ca0996a23e7eca87a5ac83105 @@ -0,0 +1 @@ +{"brier_score":0.815,"experiment_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","forecast_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic","resolved_outcome":"execution_failure","schema":"ilxyr.forecast_settlement.v1","stake":5} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/7ec189d139830ece3013193f4a8a5389ef1aee0882a428a133aed54c06960048 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/7ec189d139830ece3013193f4a8a5389ef1aee0882a428a133aed54c06960048 new file mode 100644 index 0000000..bc5cc50 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/7ec189d139830ece3013193f4a8a5389ef1aee0882a428a133aed54c06960048 @@ -0,0 +1 @@ +{"accepted":true,"checks":[{"detail":"4/4 required research stages resolved","gate":"methodology","passed":true},{"detail":"3 frozen mutually evaluated outcomes declared","gate":"outcome_contract","passed":true},{"detail":"2/2 distinct forecasters","gate":"forecast_participation","passed":true},{"detail":"10/10 forecast credits staked","gate":"forecast_stake","passed":true},{"detail":"100/100 compute credits committed","gate":"compute_funding","passed":true},{"detail":"local-command adapter is installed","gate":"executor_available","passed":true},{"detail":"local execution is limited to public-weight handles","gate":"weight_protection","passed":true},{"detail":"local execution requires network=open and an absolute executable path","gate":"local_execution_policy","passed":true},{"detail":"local-command directly executes the declared program","gate":"code_policy","passed":true},{"detail":"local-command records stdout, stderr, and metric artifacts","gate":"export_policy","passed":true},{"detail":"proposer and engineering reviewer use distinct handles","gate":"reviewer_separation","passed":true},{"detail":"proposer and forecasters use distinct handles","gate":"forecaster_separation","passed":true}],"decided_at_ms":1786571246029,"experiment_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","schema":"ilxyr.admission.v1"} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e new file mode 100644 index 0000000..2e6371a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e @@ -0,0 +1 @@ +{"actor":{"id":"model://codex/lecore-qwen/statistical-reviewer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/statistical-reviewer"},"body":"Per-token losses are serially correlated. The acceptance decision therefore uses paired installed-minus-original token NLLs, estimates an autocorrelation-aware moving-block length, and requires the upper 95 percent confidence bound to stay within a one-percent perplexity regression budget. Point estimates and independent-token bootstrap intervals are not acceptance evidence.","claims":["At least the preregistered minimum number of paired token positions is measured.","The statistical gate is decided by the paired block interval rather than the point perplexity delta."],"confidence":0.95,"id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation","input_refs":["lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis"],"schema":"ilxyr.contribution.v1","stage":"mathematical_foundation","title":"Paired moving-block inference is required for token-loss comparisons"} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/9b2545e28084836ecf1ca8a53767053fd86433449e495d0b30f5d29a0f8afa16 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/9b2545e28084836ecf1ca8a53767053fd86433449e495d0b30f5d29a0f8afa16 new file mode 100644 index 0000000..3cf3b4a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/9b2545e28084836ecf1ca8a53767053fd86433449e495d0b30f5d29a0f8afa16 @@ -0,0 +1 @@ +{"brier_score":1.185,"forecast_ids":["lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical"],"forecasts_settled":1,"handle":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical","last_settlement_at_ms":1786571246828,"probationary":true,"reliability":1.185,"resolution":0.0,"schema":"ilxyr.calibration_record.v1","uncertainty":0.0} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/bcf9449a813715c1ff821104588fa7f4169db64da5435b93d698beae6ab09ef2 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/bcf9449a813715c1ff821104588fa7f4169db64da5435b93d698beae6ab09ef2 new file mode 100644 index 0000000..ed58821 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/bcf9449a813715c1ff821104588fa7f4169db64da5435b93d698beae6ab09ef2 @@ -0,0 +1 @@ +{"experiment_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","forecaster":{"id":"model://codex/lecore-qwen/forecaster-mechanistic","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic"},"id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic","probabilities":{"accepted":0.15,"execution_failure":0.3,"rejected":0.55},"rationale":"The blank prepend has a strong identity construction, while full-checkpoint memory pressure and official multimodal reload remain material execution risks.","schema":"ilxyr.forecast.v1","stake":5} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/c633a2a658d6347150e9327eabfd0b6e5f7f4a40a6ef83ed3b49506d894026cc b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/c633a2a658d6347150e9327eabfd0b6e5f7f4a40a6ef83ed3b49506d894026cc new file mode 100644 index 0000000..58906fc --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/c633a2a658d6347150e9327eabfd0b6e5f7f4a40a6ef83ed3b49506d894026cc @@ -0,0 +1 @@ +{"brier_score":1.185,"experiment_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","forecast_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical","resolved_outcome":"execution_failure","schema":"ilxyr.forecast_settlement.v1","stake":5} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731 new file mode 100644 index 0000000..a04ecca --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731 @@ -0,0 +1 @@ +{"completed_at_ms":1786571246741,"exit_code":1,"experiment_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","id":"run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690","metrics":{},"output_truncated":false,"schema":"ilxyr.run.v1","started_at_ms":1786571246690,"stderr":"Traceback (most recent call last):\n File \"/opt/qwen35-acceptance/lecore/experiments/qwen35_acceptance/run.py\", line 17, in \n import numpy as np\nModuleNotFoundError: No module named 'numpy'\n","stdout":"","timed_out":false} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/edbcba5124b657e19d76e7728791cb2a29a9f5d9dfb57247a165e46a06833a47 b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/edbcba5124b657e19d76e7728791cb2a29a9f5d9dfb57247a165e46a06833a47 new file mode 100644 index 0000000..f79c35a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/ledger/objects/sha256/edbcba5124b657e19d76e7728791cb2a29a9f5d9dfb57247a165e46a06833a47 @@ -0,0 +1 @@ +{"compute_credits":100,"experiment_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance","funder":{"id":"service://lecore/qwen-experiment-generator","kind":"service"},"id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.funding","rationale":"Reserve one complete public-weight Qwen acceptance run under the frozen contract.","schema":"ilxyr.funding.v1"} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/preflight-001.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/preflight-001.json new file mode 100644 index 0000000..4545464 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/preflight-001.json @@ -0,0 +1,19 @@ +{ + "schema": "lecore.qwen35-preflight-record.v1", + "attempt": "preflight-001", + "instance_id": "i-0b49b8662723706f0", + "launched_at": "2026-08-12T21:41:10Z", + "completed_at": "2026-08-12T21:42:03.491811+00:00", + "formal_run_started": false, + "stage": "preflight_failure", + "exit_code": 128, + "expected_source_commit": "a04ab563ea159398a58a4cb002568cc78b9f14bb", + "observed_checkout_commit": "3d6454062982907b7d95c2e6dc267e70a9b921b6", + "cause": "The frozen source commit was mistyped and did not exist in the remote repository.", + "corrected_source_commit": "a04ab5692be38f06120aba4b0bc5e2a284eb2c79", + "evidence_bundle": { + "uri": "s3://zero-training-022118847419/qwen35-acceptance/a04ab56-2fc06364-a6c9d113/qwen35-evidence.tgz", + "bytes": 408293, + "sha256": "836fd6b29fa63ca6a636a3f9b3f1e66b7928605e0166e868e871a8797357e76a" + } +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/engineering-review.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/engineering-review.json new file mode 100644 index 0000000..0590fb7 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/engineering-review.json @@ -0,0 +1,21 @@ +{ + "actor": { + "id": "model://codex/lecore-qwen/engineering-reviewer", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/engineering-reviewer" + }, + "body": "The shell-free runner uses absolute paths, records the exact leCore commit and checker hashes, keeps spectral filtering disabled, invokes the layer-prepending path only with its experimental acknowledgement, records peak memory, reloads the emitted artifact, and exercises the official Transformers text and image-text interfaces.", + "claims": [ + "The emitted stdout is exactly the ilxyr metrics/source envelope.", + "Installer logs and a human-readable metrics artifact are retained beside the output checkpoint." + ], + "confidence": 0.85, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.engineering-review", + "input_refs": [ + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation" + ], + "schema": "ilxyr.contribution.v1", + "stage": "engineering_review", + "title": "Qwen acceptance runner and provenance boundary" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/experiment-design.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/experiment-design.json new file mode 100644 index 0000000..eb94535 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/experiment-design.json @@ -0,0 +1,22 @@ +{ + "actor": { + "id": "model://codex/lecore-qwen/experiment-designer", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/experiment-designer" + }, + "body": "Execute once against the content-bound public checkpoint, installation corpus, and separate held-out evaluation corpus. Do not tune thresholds or replace either corpus after admission. Resolve accepted only when source cleanliness, tokenizer parity, reference-logit parity, the paired statistical gate, disk reload, official text generation, and official vision smoke all pass. A cleanly executed no-go is preserved as rejected evidence.", + "claims": [ + "Accepted and rejected are exhaustive for a valid metrics envelope.", + "Runtime or dependency failure resolves separately as execution_failure." + ], + "confidence": 0.95, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.experiment-design", + "input_refs": [ + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.engineering-review" + ], + "schema": "ilxyr.contribution.v1", + "stage": "experiment_design", + "title": "One-shot Qwen3.5 installation acceptance run" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/experiment.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/experiment.json new file mode 100644 index 0000000..baabaef --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/experiment.json @@ -0,0 +1,241 @@ +{ + "baseline": "baseline://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "datasets": [ + "dataset://lecore/qwen-installation/d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f", + "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + ], + "evidence_authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [], + "checker": "checker://lecore/qwen35-acceptance/v1", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "execution": { + "args": [ + "/opt/qwen35-acceptance/lecore/experiments/qwen35_acceptance/run.py", + "/opt/qwen35-acceptance/model", + "/opt/qwen35-acceptance/installed-checkpoint", + "/opt/qwen35-acceptance/lecore/REFERENCE.md", + "/opt/qwen35-acceptance/inputs/federalist-papers.txt", + "--min-tokens", + "4096" + ], + "executor": "local-command", + "max_cost_credits": 100, + "network": "open", + "program": "/usr/bin/python3.12", + "timeout_seconds": 21600 + }, + "expected_outputs": [ + "metrics.acceptance_pass", + "metrics.source_clean", + "metrics.spectral_filtering_enabled", + "metrics.experimental_installer_used", + "metrics.tokenizer_parity_pass", + "metrics.reference_logit_parity_pass", + "metrics.reference_logit_relative_error", + "metrics.eval_tokens", + "metrics.original_perplexity", + "metrics.installed_perplexity", + "metrics.perplexity_delta_pct", + "metrics.paired_ci_lo_nats", + "metrics.paired_ci_hi_nats", + "metrics.statistical_gate_pass", + "metrics.paired_block_length", + "metrics.paired_effective_tokens", + "metrics.peak_rss_mb", + "metrics.peak_gpu_mb", + "metrics.emitted_checkpoint_mb", + "metrics.reload_pass", + "metrics.text_generation_pass", + "metrics.vision_smoke_pass", + "resolved_outcome", + "forecast_settlements" + ], + "funding": { + "minimum_forecasters": 2, + "minimum_total_stake": 10, + "required_compute_credits": 100 + }, + "hypothesis": "The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.", + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "lineage": { + "engineering_review": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.engineering-review", + "experiment_design": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.experiment-design", + "hypothesis": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis", + "mathematical_foundation": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation" + }, + "metrics": [ + { + "description": "All mandatory Qwen installation acceptance gates passed, encoded as 0 or 1.", + "name": "acceptance_pass", + "unit": "boolean" + }, + { + "description": "The tracked leCore checkout was clean when the run began, encoded as 0 or 1.", + "name": "source_clean", + "unit": "boolean" + }, + { + "description": "Whether research-only spectral filtering was used; the acceptance path requires 0.", + "name": "spectral_filtering_enabled", + "unit": "boolean" + }, + { + "description": "The explicitly acknowledged layer-prepending installer ran, encoded as 0 or 1.", + "name": "experimental_installer_used", + "unit": "boolean" + }, + { + "description": "leCore and the official Transformers tokenizer produced identical reference token IDs.", + "name": "tokenizer_parity_pass", + "unit": "boolean" + }, + { + "description": "Pre-install leCore logits matched the official Transformers text model within the frozen tolerance.", + "name": "reference_logit_parity_pass", + "unit": "boolean" + }, + { + "description": "Maximum absolute pre-install logit error divided by the maximum absolute reference logit.", + "name": "reference_logit_relative_error", + "unit": "ratio" + }, + { + "description": "Paired token positions included in the streamed evaluation.", + "name": "eval_tokens", + "unit": "tokens" + }, + { + "description": "Original checkpoint perplexity on the frozen corpus and chunking procedure.", + "name": "original_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed checkpoint perplexity on the same token positions.", + "name": "installed_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed minus original perplexity as a percentage of original.", + "name": "perplexity_delta_pct", + "unit": "percent" + }, + { + "description": "Lower 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_lo_nats", + "unit": "nats_per_token" + }, + { + "description": "Upper 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_hi_nats", + "unit": "nats_per_token" + }, + { + "description": "The paired upper confidence bound stayed within the preregistered maximum regression.", + "name": "statistical_gate_pass", + "unit": "boolean" + }, + { + "description": "Moving-block length inferred from autocorrelation in paired token loss differences.", + "name": "paired_block_length", + "unit": "tokens" + }, + { + "description": "Effective paired sample size after serial-correlation adjustment.", + "name": "paired_effective_tokens", + "unit": "tokens" + }, + { + "description": "Peak resident memory observed across the runner and installer child process.", + "name": "peak_rss_mb", + "unit": "megabytes" + }, + { + "description": "Peak accelerator allocation reported by PyTorch, or 0 when no accelerator was used.", + "name": "peak_gpu_mb", + "unit": "megabytes" + }, + { + "description": "Total safetensors size of the emitted installed checkpoint.", + "name": "emitted_checkpoint_mb", + "unit": "megabytes" + }, + { + "description": "The emitted checkpoint reloaded from disk and produced finite logits.", + "name": "reload_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint generated text through the official Transformers model.", + "name": "text_generation_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint accepted a synthetic image through the official Qwen vision-language processor and generated a token.", + "name": "vision_smoke_pass", + "unit": "boolean" + } + ], + "models": [ + "weight://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + ], + "outcome_contract": { + "outcomes": [ + { + "description": "Every frozen Qwen acceptance gate passed.", + "id": "accepted", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "gte", + "threshold": 1 + } + }, + { + "description": "The run completed but at least one acceptance gate failed.", + "id": "rejected", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "lt", + "threshold": 1 + } + }, + { + "description": "The runner failed or did not emit its exact metric contract.", + "id": "execution_failure", + "predicate": { + "kind": "execution_failure" + } + } + ], + "primary_metric": "acceptance_pass", + "success_outcome": "accepted" + }, + "proposer": { + "id": "model://codex/lecore-qwen/research-director", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director" + }, + "rationale": "This converts the unresolved Qwen integration claims into one frozen, replayable, statistically gated run. Spectral filtering is a separate research control and is forbidden in this contract.", + "schema": "ilxyr.experiment.v1", + "security": { + "code_policy": "arbitrary", + "export_policy": "artifacts", + "weight_class": "public" + }, + "seeds": [ + 0 + ], + "title": "Qwen3.5 leCore layer-prepending acceptance" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/forecast-empirical.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/forecast-empirical.json new file mode 100644 index 0000000..26aae17 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/forecast-empirical.json @@ -0,0 +1,17 @@ +{ + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-empirical", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical", + "probabilities": { + "accepted": 0.2, + "execution_failure": 0.15, + "rejected": 0.65 + }, + "rationale": "Structural rehearsal passes, but no complete post-memory-fix real-Qwen install and powered acceptance run has succeeded yet.", + "schema": "ilxyr.forecast.v1", + "stake": 5 +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/forecast-mechanistic.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/forecast-mechanistic.json new file mode 100644 index 0000000..07f2fe7 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/forecast-mechanistic.json @@ -0,0 +1,17 @@ +{ + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-mechanistic", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic", + "probabilities": { + "accepted": 0.15, + "execution_failure": 0.3, + "rejected": 0.55 + }, + "rationale": "The blank prepend has a strong identity construction, while full-checkpoint memory pressure and official multimodal reload remain material execution risks.", + "schema": "ilxyr.forecast.v1", + "stake": 5 +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/foundation.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/foundation.json new file mode 100644 index 0000000..ca73b84 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/foundation.json @@ -0,0 +1,20 @@ +{ + "actor": { + "id": "model://codex/lecore-qwen/statistical-reviewer", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/statistical-reviewer" + }, + "body": "Per-token losses are serially correlated. The acceptance decision therefore uses paired installed-minus-original token NLLs, estimates an autocorrelation-aware moving-block length, and requires the upper 95 percent confidence bound to stay within a one-percent perplexity regression budget. Point estimates and independent-token bootstrap intervals are not acceptance evidence.", + "claims": [ + "At least the preregistered minimum number of paired token positions is measured.", + "The statistical gate is decided by the paired block interval rather than the point perplexity delta." + ], + "confidence": 0.95, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.foundation", + "input_refs": [ + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis" + ], + "schema": "ilxyr.contribution.v1", + "stage": "mathematical_foundation", + "title": "Paired moving-block inference is required for token-loss comparisons" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/funding.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/funding.json new file mode 100644 index 0000000..54d71db --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/funding.json @@ -0,0 +1,11 @@ +{ + "compute_credits": 100, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "funder": { + "id": "service://lecore/qwen-experiment-generator", + "kind": "service" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.funding", + "rationale": "Reserve one complete public-weight Qwen acceptance run under the frozen contract.", + "schema": "ilxyr.funding.v1" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/hypothesis.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/hypothesis.json new file mode 100644 index 0000000..1fb985c --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/hypothesis.json @@ -0,0 +1,19 @@ +{ + "actor": { + "id": "model://codex/lecore-qwen/research-director", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director" + }, + "body": "The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.", + "claims": [ + "Pre-install leCore logits match the official reference implementation.", + "The installed checkpoint reloads and remains within the paired regression budget.", + "Official Transformers text generation and vision input both remain operational." + ], + "confidence": 0.35, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.hypothesis", + "input_refs": [], + "schema": "ilxyr.contribution.v1", + "stage": "hypothesis", + "title": "Qwen3.5 can carry a reloadable leCore layer-prepending installation" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/project.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/project.json new file mode 100644 index 0000000..1ab5586 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/project/project.json @@ -0,0 +1,155 @@ +{ + "commands": [ + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance/project/hypothesis.json" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance/project/foundation.json" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance/project/engineering-review.json" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance/project/experiment-design.json" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "compile", + "", + "/opt/qwen35-acceptance/project/experiment.json" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "forecast", + "", + "/opt/qwen35-acceptance/project/forecast-empirical.json" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "forecast", + "", + "/opt/qwen35-acceptance/project/forecast-mechanistic.json" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "fund", + "", + "/opt/qwen35-acceptance/project/funding.json" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "admit", + "", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "run", + "", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "--execute" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "status", + "", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance" + ], + [ + "/opt/qwen35-acceptance/ilxyr/target/release/ilxyr", + "verify", + "" + ] + ], + "corpora": { + "evaluation": { + "bytes": 1213410, + "path": "/opt/qwen35-acceptance/inputs/federalist-papers.txt", + "sha256": "a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + }, + "installation": { + "bytes": 2300089, + "path": "/opt/qwen35-acceptance/lecore/REFERENCE.md", + "sha256": "d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f" + } + }, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "installed_dir": "/opt/qwen35-acceptance/installed-checkpoint", + "model_digest": "20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "model_files": [ + { + "bytes": 11544, + "path": "LICENSE", + "sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a" + }, + { + "bytes": 61705, + "path": "README.md", + "sha256": "87a163af54f32fa608a0f8d3ac67945c53dd2b4a7c96740b3d7fdc28e8458864" + }, + { + "bytes": 7755, + "path": "chat_template.jinja", + "sha256": "273d8e0e683b885071fb17e08d71e5f2a5ddfb5309756181681de4f5a1822d80" + }, + { + "bytes": 2907, + "path": "config.json", + "sha256": "b90b86f35c8e6925ef74ee04d0e758f0a845c83a42089ad82bbaa948de9b4204" + }, + { + "bytes": 3353259, + "path": "merges.txt", + "sha256": "a9d356d7bdf1ef4949e3e748e95b8e10ad9d4e2e838eddc38a0a7b6b94d1db8d" + }, + { + "bytes": 1746942600, + "path": "model.safetensors-00001-of-00001.safetensors", + "sha256": "04b1c301231dd422b8860db31311ab2721511346a32cb1e079c4c4e5f1fe4696" + }, + { + "bytes": 50900, + "path": "model.safetensors.index.json", + "sha256": "d8a08838a613b025eb7952ed9db11696213e57e76a375661ef5c12f9dd5dcf4e" + }, + { + "bytes": 390, + "path": "preprocessor_config.json", + "sha256": "27225450ac9c6529872ee1924fcb0962ff5634834f817040f444118116f4e516" + }, + { + "bytes": 12807982, + "path": "tokenizer.json", + "sha256": "5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42" + }, + { + "bytes": 16709, + "path": "tokenizer_config.json", + "sha256": "49e2b6e395f959f077f1e992b338919c0d4a9732fc6e613995e06557f843500c" + }, + { + "bytes": 385, + "path": "video_preprocessor_config.json", + "sha256": "7768af27c1fafa9cc9011c1dc20067e03f8915e03b63504550e11d5066986d13" + }, + { + "bytes": 6722759, + "path": "vocab.json", + "sha256": "ce99b4cb2983d118806ce0a8b777a35b093e2000a503ebde25853284c9dfa003" + } + ], + "schema": "lecore.ilxyr-project.v1", + "source_commit": "a04ab5692be38f06120aba4b0bc5e2a284eb2c79" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-manifest.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-manifest.json new file mode 100644 index 0000000..9f62a52 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-manifest.json @@ -0,0 +1,374 @@ +{ + "evidence_ref": "artifact://sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "files": [ + { + "bytes": 2710, + "media_type": "text/markdown", + "path": "README.md", + "sha256": "15970fb8672f57f9371e1960b733ed3b1f5c3c9ff1fba625e9bb95096c2fd54f" + }, + { + "bytes": 17761, + "media_type": "application/json", + "path": "evidence/evidence.in-toto.json", + "sha256": "f9bc730a5ade64659e4e523f115cfc71f2141757cecdbb8a520f87a36a83cc52" + }, + { + "bytes": 16592, + "media_type": "application/json", + "path": "evidence/evidence.native.json", + "sha256": "101590b922f3877f88971073171fa622d29af62cad52dca1990e3100d122c204" + }, + { + "bytes": 11976, + "media_type": "application/json", + "path": "evidence/evidence.ro-crate.json", + "sha256": "7758207b7bf7407e2e3ddb526aab3f049c4f2579702700b733e740a3a44dbd9d" + }, + { + "bytes": 1324, + "media_type": "application/json", + "path": "evidence/exporter-provenance.json", + "sha256": "3812e807dfa60ea6f1d4723b778059fb3a01dd14236c14d9c42473689651ed97" + }, + { + "bytes": 97, + "media_type": "application/json", + "path": "ledger/config.json", + "sha256": "2301b853e1e01b8844970b90e72036bed57c580ded81a857ec7faf27f8deca66" + }, + { + "bytes": 9048, + "media_type": "application/octet-stream", + "path": "ledger/events.jsonl", + "sha256": "0254b4dc60bf2903cf308104bf3f8c34d64fb050fb1e4e014590fe0c1847dc35" + }, + { + "bytes": 1128, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "sha256": "81283e25e896f94dbc576d3628d1251a541c1c3c5c1ad98ebd5df7c748e39c17" + }, + { + "bytes": 1259, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/22fc1c29c182563c80d9188e4e2d077e6fbb72dcdc0c1d046e7d7e22dd3c00e1", + "sha256": "3cdd9d88703ad90d5e2fc233fa51953e620684dd8b211ba5f5e7b4b7bf580bfb" + }, + { + "bytes": 1305, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "sha256": "15b9815a3728cb663ad09b13be2b5d2a74e05ae4c1c853f2e46a0c07e05fbb0f" + }, + { + "bytes": 8502, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6", + "sha256": "3bc9356df4220d123a5323669c09101aea2d0be60a2d11a6a449e2972fa6ddea" + }, + { + "bytes": 917, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "sha256": "2687de1a8e09012a5c805e711aaf55d5bc4b649982e885ab837e588ea875631f" + }, + { + "bytes": 643, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/52df3c246b4b322591efcd5402e1bbc04ed5df3d32aacd1f490782dc030f47f2", + "sha256": "c35075e34b61a0d26073672d46f2b22eb64c2e33e7034b6a25e6aee5b1e2a580" + }, + { + "bytes": 408, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/5b331e9b8a34f86e4dd5c69c9f9a267280430f5cf28135bd1db5f6d94002e57e", + "sha256": "05fc5f50d074aeec0261ab904dc4bef96614868750987416fc4c23bedda86b16" + }, + { + "bytes": 331, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/77a6a78f2b127ce64f97022f7142315e33556e5ca0996a23e7eca87a5ac83105", + "sha256": "a8257a3ec462ddd946b095fddc522958f281bc4937c6f835ba4bf6fe4e911578" + }, + { + "bytes": 1381, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/7ec189d139830ece3013193f4a8a5389ef1aee0882a428a133aed54c06960048", + "sha256": "78be7ac477a426a8e0b02d55f68331bdf0650cf23d319b280742269948eb5136" + }, + { + "bytes": 1125, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e", + "sha256": "eaf989375af21fd4f04c5245a433298f5d70f16e0683e5718505e3aefea38cf4" + }, + { + "bytes": 404, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/9b2545e28084836ecf1ca8a53767053fd86433449e495d0b30f5d29a0f8afa16", + "sha256": "62a9f03d05daa4931b1c7936e575569e03f13ccf293dffe74b03aeb511941ac2" + }, + { + "bytes": 680, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/bcf9449a813715c1ff821104588fa7f4169db64da5435b93d698beae6ab09ef2", + "sha256": "5cd206e5279573a0a4c83b5e2a69b4aaef37e0637d4c55d74189fc1487484594" + }, + { + "bytes": 329, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/c633a2a658d6347150e9327eabfd0b6e5f7f4a40a6ef83ed3b49506d894026cc", + "sha256": "bcbf059b8532b9142c2d91606fd3ff951e8ce737ddff28cd9d8b0fdeadc6b7b1" + }, + { + "bytes": 608, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731", + "sha256": "fe668351c30a0e1a1cf3a5cdb828462f95a5b06726f6398f45aad15bc273fb2b" + }, + { + "bytes": 424, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/edbcba5124b657e19d76e7728791cb2a29a9f5d9dfb57247a165e46a06833a47", + "sha256": "815ce193f4d124c2d7958fc7e70ec32b27dd94e4cad30345902807b5e632457b" + }, + { + "bytes": 855, + "media_type": "application/json", + "path": "preflight-001.json", + "sha256": "6a89f3ca1b442da5ff5b8483ccb342d1a95a207c60309d2e4d85b45490865eda" + }, + { + "bytes": 1212, + "media_type": "application/json", + "path": "project/engineering-review.json", + "sha256": "8bfc0d775ab62587fc09ca3d2c2ad5be8531b054d97a573c5930ce65a82445d3" + }, + { + "bytes": 1394, + "media_type": "application/json", + "path": "project/experiment-design.json", + "sha256": "8ddbd0dc9a45474d6d55a21b81aef90d8fc24c404c85139fae946989ff150385" + }, + { + "bytes": 8744, + "media_type": "application/json", + "path": "project/experiment.json", + "sha256": "5d6d07fcc8a71ad23e17c1b55dd9b2efe0c757aedf12544c0a76d1911259cef7" + }, + { + "bytes": 714, + "media_type": "application/json", + "path": "project/forecast-empirical.json", + "sha256": "db8f195c85b535a8fbba8925f214e506e5cd482934208a488b28b4f2ea1a4db9" + }, + { + "bytes": 751, + "media_type": "application/json", + "path": "project/forecast-mechanistic.json", + "sha256": "05b4bf3d7918f746e153bf740aebca7376a409e98eb322dda23f4f3ce0893697" + }, + { + "bytes": 1204, + "media_type": "application/json", + "path": "project/foundation.json", + "sha256": "84d832b16029d355e0bc0ceea851229d139fc8f6818f72535a994a90f5e317a8" + }, + { + "bytes": 464, + "media_type": "application/json", + "path": "project/funding.json", + "sha256": "8b81492f1f1fda6ffbd8f059c5faa9854da5f484df5e787848c95b75d6ee1f7a" + }, + { + "bytes": 993, + "media_type": "application/json", + "path": "project/hypothesis.json", + "sha256": "c06e45c56e0ba5b25ffe1f74d9e94f64e4b38e9332871ebf256da1de90a4722d" + }, + { + "bytes": 4792, + "media_type": "application/json", + "path": "project/project.json", + "sha256": "3ff0578ccdb022bde402ea5640f028b09288a27c3aaa0ce12bcc6f27944d6fc0" + }, + { + "bytes": 423, + "media_type": "application/json", + "path": "result/admission-preflight.json", + "sha256": "80caf0972523f9703b235704cd29644ccda33645110a9b6292be42d795f3248f" + }, + { + "bytes": 479, + "media_type": "application/json", + "path": "result/aws-instance-identity.json", + "sha256": "969bac0db9321c879fd2da5ffa2ad22b37ca2ea91d22b4a7de393db5f0476b27" + }, + { + "bytes": 45379, + "media_type": "text/plain", + "path": "result/cloud-init-output.log", + "sha256": "9a85bc05e466c409368525d1ce83d5696e1f10e0fb2a63d46948cefb9e862cab" + }, + { + "bytes": 382, + "media_type": "application/json", + "path": "result/corpus-manifest.json", + "sha256": "3b028dc43d1c1c853bc695c52fc5d5e00abb79c48e7230277aa8773840c169c8" + }, + { + "bytes": 4826, + "media_type": "text/plain", + "path": "result/environment.txt", + "sha256": "8aaff494d9e6405db0dc0fa5c3b9b965f869678a0766df73d0eb6e5a32f439d7" + }, + { + "bytes": 289, + "media_type": "application/json", + "path": "result/execution-summary.json", + "sha256": "ce101a2f0e994a033a422258b2e45614cf075d6c0de91d2ffbbee25631160397" + }, + { + "bytes": 216, + "media_type": "application/json", + "path": "result/generation.json", + "sha256": "907c01e8c17c9fd78d7edb94840d1600286dfd0a07fb7eccf1cf3cdd35ccd22a" + }, + { + "bytes": 1813, + "media_type": "application/json", + "path": "result/ilxyr-admission.json", + "sha256": "7ea4c55b326d0d6eac3db4a87b7262e850f44793b669585b59147e6eab08d9b9" + }, + { + "bytes": 259, + "media_type": "text/plain", + "path": "result/ilxyr-commit.txt", + "sha256": "32873b519f7170a47a099930a325f4bb29d04021b822fa7f5387c94a9976add4" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-compile.json", + "sha256": "84e09b575a681881744f5149d7c771344663dc422a7f9826797aac5b44564a0f" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-engineering-review.json", + "sha256": "d8eb46fc15dbfe3c35bec7ca46479c64a780f1e2a9aa69857c91179a38280e65" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-experiment-design.json", + "sha256": "4154c91c65fe786c08ef67dce8ac734f12ce5197d1c6d6d823f431ea6b54ce4d" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-foundation.json", + "sha256": "5bd8600d1287e07931bd24d0da9df98b1cc154c46940414688e5d4aa536a2d5f" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-hypothesis.json", + "sha256": "72070d12d42ec1ec544a1f90b87c42f2b0632d853adc4c99adbc82758ad257b7" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-forecast-empirical.json", + "sha256": "d4a49e64e33ba617ead6e85450e02418273f4ce20e36a7b5f98762dacfad40af" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-forecast-mechanistic.json", + "sha256": "15d7271b607be3cd3168c20fc200f51247ca4ac505217b0b447cf023d25c39bc" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-funding.json", + "sha256": "8e67a1b939092df3906e43af29baf91da8e78c01729f542feb0d9c2143f284d1" + }, + { + "bytes": 77, + "media_type": "application/json", + "path": "result/ilxyr-init.json", + "sha256": "ae2a2e3ffd57ca8721f36741df539bb4e8539a26071dda601e5dc2da1cb2e45d" + }, + { + "bytes": 4055, + "media_type": "application/json", + "path": "result/ilxyr-run.json", + "sha256": "813ddf2c02fd212eca7dd1d2f5c97f0c15dfd4e379317cbb934e4a7ce0367a93" + }, + { + "bytes": 4535, + "media_type": "application/json", + "path": "result/ilxyr-status.json", + "sha256": "617ea4d656cb74f6d1abbe5e3000a5565d9dda8086a150ebc936aa2c78225fd6" + }, + { + "bytes": 69, + "media_type": "application/json", + "path": "result/ilxyr-verify.json", + "sha256": "5e03d79a86e57f015be7b79666349b5813a1a702421883e8b58f3ed3a6a73663" + }, + { + "bytes": 3104, + "media_type": "application/json", + "path": "result/launch-manifest.json", + "sha256": "bfa6466cb85c6ee7cc378b31bfd5ae15a7c693ef9c45028f69de3c3dac9d50e6" + }, + { + "bytes": 264, + "media_type": "text/plain", + "path": "result/lecore-commit.txt", + "sha256": "0eb24c28e740d741ba99054eaf732eab4e0f10ce13af8f60e7106b6b0ec5491e" + }, + { + "bytes": 0, + "media_type": "text/plain", + "path": "result/lecore-status.txt", + "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" + }, + { + "bytes": 2023, + "media_type": "application/json", + "path": "result/model-manifest.json", + "sha256": "f352ac2c17ee1bea38f70d7ab1975e3297954423cb74f5330383134e4b6c493a" + }, + { + "bytes": 588, + "media_type": "text/plain", + "path": "result/requirements-resolved.txt", + "sha256": "ffc5f6316f74e8e6c465b3845f6b4feb9e3b534160434299de9b05c809d7f647" + }, + { + "bytes": 172, + "media_type": "application/json", + "path": "result/runner-status.json", + "sha256": "70545aafec838419f0d7227505f805c7f6e110c7b98fee1ead80681ce7804a3c" + }, + { + "bytes": 41852, + "media_type": "text/plain", + "path": "result/user-data.log", + "sha256": "9f951ddb44114a4ef41c0c191f09eda7f3b81c73e23f0a0b4ff452480ef8912b" + } + ], + "ledger_verification": { + "events_checked": 16, + "objects_checked": 15, + "valid": true + }, + "resolved_outcome": "execution_failure", + "run_ref": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731", + "schema": "lecore.qwen35-publication-manifest.v1" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-receipt.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-receipt.json new file mode 100644 index 0000000..3c0b77e --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-receipt.json @@ -0,0 +1,49 @@ +{ + "schema": "lecore.qwen35-arweave-publication-receipt.v1", + "published_at": "2026-08-12T22:07:00Z", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "resolved_outcome": "execution_failure", + "publisher": { + "service": "ArDrive Turbo signed free upload", + "sdk": "@ardrive/turbo-sdk@1.42.0", + "ownership": "The uploads used one-use in-memory Arweave identities. Their private keys were never persisted." + }, + "publication_manifest": { + "path": "publication-manifest.json", + "bytes": 13597, + "sha256": "80ebc69fc40c4bbab9e3692a1713217fba007d1005adf2f23be79be739d1b836", + "transaction_id": "6TJSYOfCDas0zFLEsBs38CsiAwrWTCp5ui2V_tki1YM" + }, + "arweave_path_manifest": { + "transaction_id": "jguWIRC2oP5gNWNmGNNeLIGD9t155X6orv9FREZmHBk", + "bytes": 7248, + "sha256": "7aef3e9daa7738a7d697a7706c4e6453b4a03f1d63e4523ca673dfadb472cfe3", + "item_count": 61, + "gateway_url": "https://arweave.net/jguWIRC2oP5gNWNmGNNeLIGD9t155X6orv9FREZmHBk/", + "index_url": "https://arweave.net/jguWIRC2oP5gNWNmGNNeLIGD9t155X6orv9FREZmHBk/publication-manifest.json", + "owner": "I5Z-EnOhkasZjtaMu9IbSVK3duWSecQpZ0lnKFEjjRg" + }, + "selected_items": { + "native_evidence": { + "transaction_id": "1BmD-TzyDMTmuh5GIkYTPYlGj12GEmbez1k5sWqThIc", + "sha256": "101590b922f3877f88971073171fa622d29af62cad52dca1990e3100d122c204" + }, + "ledger_events": { + "transaction_id": "JZTmlxINH2Aq5ionGFCna400Uu605syGK2MUdydXzLU", + "sha256": "0254b4dc60bf2903cf308104bf3f8c34d64fb050fb1e4e014590fe0c1847dc35" + }, + "ledger_verification": { + "transaction_id": "aMGzn_CFX4v3gXo-Ib243PK6GnvF5Crv5WlgEl4_k-o", + "sha256": "5e03d79a86e57f015be7b79666349b5813a1a702421883e8b58f3ed3a6a73663" + } + }, + "confirmation": { + "status_service": "https://upload.ardrive.io/v1/tx/{transaction_id}/status", + "transactions_confirmed": 63, + "total_winc_charged": 0 + }, + "superseded_path_manifest": { + "transaction_id": "lPXKEJON5Xa4wXswwRwFv2GWqrTPo7yuOhMtVKyOpVI", + "reason": "It omitted publication-manifest.json from its path table and was replaced before the receipt was committed." + } +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/admission-preflight.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/admission-preflight.json new file mode 100644 index 0000000..51317b9 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/admission-preflight.json @@ -0,0 +1,15 @@ +{ + "checks": { + "complete_top_level_model_manifest": true, + "corpora_distinct": true, + "evaluation_corpus": true, + "installation_corpus": true, + "minimum_tokens": true, + "source_commit": true, + "spectral_disabled": true, + "timeout": true, + "tokenizer": true, + "weights": true + }, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/aws-instance-identity.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/aws-instance-identity.json new file mode 100644 index 0000000..c3eec97 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/aws-instance-identity.json @@ -0,0 +1,17 @@ +{ + "accountId" : "022118847419", + "architecture" : "x86_64", + "availabilityZone" : "us-east-1a", + "billingProducts" : null, + "devpayProductCodes" : null, + "marketplaceProductCodes" : null, + "imageId" : "ami-07a5b367e8dc8bd92", + "instanceId" : "i-0c549643b970e896a", + "instanceType" : "r7i.4xlarge", + "kernelId" : null, + "pendingTime" : "2026-08-12T21:45:26Z", + "privateIp" : "172.31.3.245", + "ramdiskId" : null, + "region" : "us-east-1", + "version" : "2017-09-30" +} \ No newline at end of file diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/cloud-init-output.log b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/cloud-init-output.log new file mode 100644 index 0000000..a6dbfe5 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/cloud-init-output.log @@ -0,0 +1,527 @@ +Cloud-init v. 22.2.2 running 'init' at Wed, 12 Aug 2026 21:45:36 +0000. Up 6.75 seconds. +ci-info: ++++++++++++++++++++++++++++++++++++++Net device info++++++++++++++++++++++++++++++++++++++ +ci-info: +---------+------+---------------------------+---------------+--------+-------------------+ +ci-info: | Device | Up | Address | Mask | Scope | Hw-Address | +ci-info: +---------+------+---------------------------+---------------+--------+-------------------+ +ci-info: | enp39s0 | True | 172.31.3.245 | 255.255.240.0 | global | 02:68:21:6e:08:57 | +ci-info: | enp39s0 | True | fe80::68:21ff:fe6e:857/64 | . | link | 02:68:21:6e:08:57 | +ci-info: | lo | True | 127.0.0.1 | 255.0.0.0 | host | . | +ci-info: | lo | True | ::1/128 | . | host | . | +ci-info: +---------+------+---------------------------+---------------+--------+-------------------+ +ci-info: +++++++++++++++++++++++++++++Route IPv4 info++++++++++++++++++++++++++++++ +ci-info: +-------+-------------+------------+-----------------+-----------+-------+ +ci-info: | Route | Destination | Gateway | Genmask | Interface | Flags | +ci-info: +-------+-------------+------------+-----------------+-----------+-------+ +ci-info: | 0 | 0.0.0.0 | 172.31.0.1 | 0.0.0.0 | enp39s0 | UG | +ci-info: | 1 | 172.31.0.0 | 0.0.0.0 | 255.255.240.0 | enp39s0 | U | +ci-info: | 2 | 172.31.0.1 | 0.0.0.0 | 255.255.255.255 | enp39s0 | UH | +ci-info: | 3 | 172.31.0.2 | 0.0.0.0 | 255.255.255.255 | enp39s0 | UH | +ci-info: +-------+-------------+------------+-----------------+-----------+-------+ +ci-info: +++++++++++++++++++Route IPv6 info+++++++++++++++++++ +ci-info: +-------+-------------+---------+-----------+-------+ +ci-info: | Route | Destination | Gateway | Interface | Flags | +ci-info: +-------+-------------+---------+-----------+-------+ +ci-info: | 0 | fe80::/64 | :: | enp39s0 | U | +ci-info: | 2 | local | :: | enp39s0 | U | +ci-info: | 3 | multicast | :: | enp39s0 | U | +ci-info: +-------+-------------+---------+-----------+-------+ +Generating public/private ed25519 key pair. +Your identification has been saved in /etc/ssh/ssh_host_ed25519_key +Your public key has been saved in /etc/ssh/ssh_host_ed25519_key.pub +The key fingerprint is: +SHA256:RURLZ1QJIZh5j/bnK5ZQylx7QBYfQQxpSZyEhvU3R84 root@ip-172-31-3-245.ec2.internal +The key's randomart image is: ++--[ED25519 256]--+ +| OB*@%=o.| +| =o+*O.o= | +| oo*. + E| +| .o =. o | +| S+ = o | +| = o o | +| . = | +| + . | +| . ... | ++----[SHA256]-----+ +Generating public/private ecdsa key pair. +Your identification has been saved in /etc/ssh/ssh_host_ecdsa_key +Your public key has been saved in /etc/ssh/ssh_host_ecdsa_key.pub +The key fingerprint is: +SHA256:13ojYOJM8KyYy8Y46LADQ0dmgt63Bt+xFabrYDHHDME root@ip-172-31-3-245.ec2.internal +The key's randomart image is: ++---[ECDSA 256]---+ +|. ... | +|.. + E o | +|. * . + o . | +| o + B * . . | +|. . + @ S . . | +|o o @ * o . | +|++o + = o o | +|*oo. . o . | +|o=o | ++----[SHA256]-----+ +Cloud-init v. 22.2.2 running 'modules:config' at Wed, 12 Aug 2026 21:45:38 +0000. Up 8.18 seconds. +Cloud-init v. 22.2.2 running 'modules:final' at Wed, 12 Aug 2026 21:45:38 +0000. Up 8.61 seconds. +Shutdown scheduled for Thu 2026-08-13 05:45:38 UTC, use 'shutdown -c' to cancel. +Amazon Linux 2023 repository 80 MB/s | 75 MB 00:00 +Amazon Linux 2023 Kernel Livepatch repository 506 kB/s | 69 kB 00:00 +Last metadata expiration check: 0:00:01 ago on Wed Aug 12 21:45:54 2026. +Package tar-2:1.34-1.amzn2023.0.4.x86_64 is already installed. +Package gzip-1.12-1.amzn2023.0.1.x86_64 is already installed. +Dependencies resolved. +================================================================================ + Package Arch Version Repository Size +================================================================================ +Installing: + cargo x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 8.2 M + git x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 53 k + python3.12 x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 27 k + python3.12-pip noarch 23.2.1-4.amzn2023.0.10 amazonlinux 2.8 M + rust x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 62 M +Installing dependencies: + annobin-docs noarch 12.69-1.amzn2023.0.1 amazonlinux 93 k + annobin-plugin-gcc x86_64 12.69-1.amzn2023.0.1 amazonlinux 974 k + cpp x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 11 M + gc x86_64 8.0.4-5.amzn2023.0.2 amazonlinux 105 k + gcc x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 36 M + gcc-plugin-annobin x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 38 k + git-core x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 4.9 M + git-core-doc noarch 2.50.1-1.amzn2023.0.1 amazonlinux 2.8 M + glibc-devel x86_64 2.34-231.amzn2023.0.5 amazonlinux 33 k + glibc-headers-x86 noarch 2.34-231.amzn2023.0.5 amazonlinux 439 k + guile22 x86_64 2.2.7-2.amzn2023.0.3 amazonlinux 6.4 M + kernel-headers x86_64 1:6.1.177-224.371.amzn2023 amazonlinux 1.4 M + libmpc x86_64 1.2.1-2.amzn2023.0.2 amazonlinux 62 k + libtool-ltdl x86_64 2.4.7-1.amzn2023.0.3 amazonlinux 38 k + libxcrypt-devel x86_64 4.4.33-7.amzn2023 amazonlinux 32 k + make x86_64 1:4.3-5.amzn2023.0.2 amazonlinux 534 k + mpdecimal x86_64 2.5.1-3.amzn2023.0.3 amazonlinux 101 k + perl-Error noarch 1:0.17030-2.amzn2023.0.1 amazonlinux 42 k + perl-File-Find noarch 1.37-477.amzn2023.0.9 amazonlinux 26 k + perl-Git noarch 2.50.1-1.amzn2023.0.1 amazonlinux 41 k + perl-TermReadKey x86_64 2.38-9.amzn2023.0.3 amazonlinux 36 k + perl-lib x86_64 0.65-477.amzn2023.0.9 amazonlinux 15 k + python3.12-libs x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 9.2 M + python3.12-pip-wheel noarch 23.2.1-4.amzn2023.0.10 amazonlinux 1.5 M + rust-std-static x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 39 M +Installing weak dependencies: + python3.12-setuptools noarch 68.2.2-4.amzn2023.0.3 amazonlinux 1.4 M + +Transaction Summary +================================================================================ +Install 31 Packages + +Total download size: 189 M +Installed size: 700 M +Downloading Packages: +(1/31): annobin-docs-12.69-1.amzn2023.0.1.noarc 3.0 MB/s | 93 kB 00:00 +(2/31): annobin-plugin-gcc-12.69-1.amzn2023.0.1 22 MB/s | 974 kB 00:00 +(3/31): gc-8.0.4-5.amzn2023.0.2.x86_64.rpm 3.7 MB/s | 105 kB 00:00 +(4/31): cpp-11.5.0-5.amzn2023.0.5.x86_64.rpm 76 MB/s | 11 MB 00:00 +(5/31): gcc-plugin-annobin-11.5.0-5.amzn2023.0. 1.3 MB/s | 38 kB 00:00 +(6/31): git-2.50.1-1.amzn2023.0.1.x86_64.rpm 2.3 MB/s | 53 kB 00:00 +(7/31): git-core-2.50.1-1.amzn2023.0.1.x86_64.r 74 MB/s | 4.9 MB 00:00 +(8/31): cargo-1.97.0-1.amzn2023.0.1.x86_64.rpm 27 MB/s | 8.2 MB 00:00 +(9/31): git-core-doc-2.50.1-1.amzn2023.0.1.noar 73 MB/s | 2.8 MB 00:00 +(10/31): glibc-devel-2.34-231.amzn2023.0.5.x86_ 1.4 MB/s | 33 kB 00:00 +(11/31): glibc-headers-x86-2.34-231.amzn2023.0. 16 MB/s | 439 kB 00:00 +(12/31): kernel-headers-6.1.177-224.371.amzn202 49 MB/s | 1.4 MB 00:00 +(13/31): guile22-2.2.7-2.amzn2023.0.3.x86_64.rp 73 MB/s | 6.4 MB 00:00 +(14/31): libmpc-1.2.1-2.amzn2023.0.2.x86_64.rpm 1.7 MB/s | 62 kB 00:00 +(15/31): libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_ 1.7 MB/s | 38 kB 00:00 +(16/31): libxcrypt-devel-4.4.33-7.amzn2023.x86_ 1.4 MB/s | 32 kB 00:00 +(17/31): gcc-11.5.0-5.amzn2023.0.5.x86_64.rpm 79 MB/s | 36 MB 00:00 +(18/31): mpdecimal-2.5.1-3.amzn2023.0.3.x86_64. 1.1 MB/s | 101 kB 00:00 +(19/31): make-4.3-5.amzn2023.0.2.x86_64.rpm 5.5 MB/s | 534 kB 00:00 +(20/31): perl-Error-0.17030-2.amzn2023.0.1.noar 1.4 MB/s | 42 kB 00:00 +(21/31): perl-File-Find-1.37-477.amzn2023.0.9.n 872 kB/s | 26 kB 00:00 +(22/31): perl-Git-2.50.1-1.amzn2023.0.1.noarch. 1.4 MB/s | 41 kB 00:00 +(23/31): perl-TermReadKey-2.38-9.amzn2023.0.3.x 1.4 MB/s | 36 kB 00:00 +(24/31): python3.12-3.12.13-2.amzn2023.0.5.x86_ 1.1 MB/s | 27 kB 00:00 +(25/31): perl-lib-0.65-477.amzn2023.0.9.x86_64. 584 kB/s | 15 kB 00:00 +(26/31): python3.12-pip-wheel-23.2.1-4.amzn2023 46 MB/s | 1.5 MB 00:00 +(27/31): python3.12-pip-23.2.1-4.amzn2023.0.10. 63 MB/s | 2.8 MB 00:00 +(28/31): python3.12-setuptools-68.2.2-4.amzn202 34 MB/s | 1.4 MB 00:00 +(29/31): python3.12-libs-3.12.13-2.amzn2023.0.5 75 MB/s | 9.2 MB 00:00 +(30/31): rust-std-static-1.97.0-1.amzn2023.0.1. 69 MB/s | 39 MB 00:00 +(31/31): rust-1.97.0-1.amzn2023.0.1.x86_64.rpm 52 MB/s | 62 MB 00:01 +-------------------------------------------------------------------------------- +Total 102 MB/s | 189 MB 00:01 +Running transaction check +Transaction check succeeded. +Running transaction test +Transaction test succeeded. +Running transaction + Preparing : 1/1 + Installing : libmpc-1.2.1-2.amzn2023.0.2.x86_64 1/31 + Installing : git-core-2.50.1-1.amzn2023.0.1.x86_64 2/31 + Installing : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 3/31 + Installing : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Installing : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 5/31 + Installing : perl-lib-0.65-477.amzn2023.0.9.x86_64 6/31 + Installing : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 7/31 + Installing : perl-File-Find-1.37-477.amzn2023.0.9.noarch 8/31 + Installing : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 9/31 + Installing : perl-Git-2.50.1-1.amzn2023.0.1.noarch 10/31 + Installing : git-2.50.1-1.amzn2023.0.1.x86_64 11/31 + Installing : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 12/31 + Installing : python3.12-3.12.13-2.amzn2023.0.5.x86_64 13/31 + Installing : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 14/31 + Installing : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 15/31 + Installing : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Installing : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 17/31 + Installing : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 18/31 + Installing : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 19/31 + Installing : glibc-devel-2.34-231.amzn2023.0.5.x86_64 20/31 + Installing : gc-8.0.4-5.amzn2023.0.2.x86_64 21/31 + Installing : guile22-2.2.7-2.amzn2023.0.3.x86_64 22/31 + Installing : make-1:4.3-5.amzn2023.0.2.x86_64 23/31 + Installing : gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Running scriptlet: gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Installing : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 25/31 + Installing : rust-1.97.0-1.amzn2023.0.1.x86_64 26/31 + Installing : annobin-docs-12.69-1.amzn2023.0.1.noarch 27/31 + Installing : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Running scriptlet: annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Installing : cargo-1.97.0-1.amzn2023.0.1.x86_64 29/31 + Installing : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 30/31 + Installing : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Running scriptlet: python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Verifying : annobin-docs-12.69-1.amzn2023.0.1.noarch 1/31 + Verifying : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 2/31 + Verifying : cargo-1.97.0-1.amzn2023.0.1.x86_64 3/31 + Verifying : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Verifying : gc-8.0.4-5.amzn2023.0.2.x86_64 5/31 + Verifying : gcc-11.5.0-5.amzn2023.0.5.x86_64 6/31 + Verifying : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 7/31 + Verifying : git-2.50.1-1.amzn2023.0.1.x86_64 8/31 + Verifying : git-core-2.50.1-1.amzn2023.0.1.x86_64 9/31 + Verifying : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 10/31 + Verifying : glibc-devel-2.34-231.amzn2023.0.5.x86_64 11/31 + Verifying : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 12/31 + Verifying : guile22-2.2.7-2.amzn2023.0.3.x86_64 13/31 + Verifying : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 14/31 + Verifying : libmpc-1.2.1-2.amzn2023.0.2.x86_64 15/31 + Verifying : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Verifying : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 17/31 + Verifying : make-1:4.3-5.amzn2023.0.2.x86_64 18/31 + Verifying : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 19/31 + Verifying : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 20/31 + Verifying : perl-File-Find-1.37-477.amzn2023.0.9.noarch 21/31 + Verifying : perl-Git-2.50.1-1.amzn2023.0.1.noarch 22/31 + Verifying : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 23/31 + Verifying : perl-lib-0.65-477.amzn2023.0.9.x86_64 24/31 + Verifying : python3.12-3.12.13-2.amzn2023.0.5.x86_64 25/31 + Verifying : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 26/31 + Verifying : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 27/31 + Verifying : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 28/31 + Verifying : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 29/31 + Verifying : rust-1.97.0-1.amzn2023.0.1.x86_64 30/31 + Verifying : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 31/31 + +Installed: + annobin-docs-12.69-1.amzn2023.0.1.noarch + annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 + cargo-1.97.0-1.amzn2023.0.1.x86_64 + cpp-11.5.0-5.amzn2023.0.5.x86_64 + gc-8.0.4-5.amzn2023.0.2.x86_64 + gcc-11.5.0-5.amzn2023.0.5.x86_64 + gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 + git-2.50.1-1.amzn2023.0.1.x86_64 + git-core-2.50.1-1.amzn2023.0.1.x86_64 + git-core-doc-2.50.1-1.amzn2023.0.1.noarch + glibc-devel-2.34-231.amzn2023.0.5.x86_64 + glibc-headers-x86-2.34-231.amzn2023.0.5.noarch + guile22-2.2.7-2.amzn2023.0.3.x86_64 + kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 + libmpc-1.2.1-2.amzn2023.0.2.x86_64 + libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 + libxcrypt-devel-4.4.33-7.amzn2023.x86_64 + make-1:4.3-5.amzn2023.0.2.x86_64 + mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 + perl-Error-1:0.17030-2.amzn2023.0.1.noarch + perl-File-Find-1.37-477.amzn2023.0.9.noarch + perl-Git-2.50.1-1.amzn2023.0.1.noarch + perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 + perl-lib-0.65-477.amzn2023.0.9.x86_64 + python3.12-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-pip-23.2.1-4.amzn2023.0.10.noarch + python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch + python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch + rust-1.97.0-1.amzn2023.0.1.x86_64 + rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 + +Complete! +Cloning into '/opt/qwen35-acceptance/lecore'... +Updating files: 0% (2/903) Updating files: 1% (10/903) Updating files: 2% (19/903) Updating files: 3% (28/903) Updating files: 4% (37/903) Updating files: 5% (46/903) Updating files: 6% (55/903) Updating files: 7% (64/903) Updating files: 8% (73/903) Updating files: 9% (82/903) Updating files: 10% (91/903) Updating files: 11% (100/903) Updating files: 12% (109/903) Updating files: 13% (118/903) Updating files: 14% (127/903) Updating files: 15% (136/903) Updating files: 16% (145/903) Updating files: 17% (154/903) Updating files: 18% (163/903) Updating files: 19% (172/903) Updating files: 20% (181/903) Updating files: 21% (190/903) Updating files: 22% (199/903) Updating files: 23% (208/903) Updating files: 24% (217/903) Updating files: 25% (226/903) Updating files: 26% (235/903) Updating files: 27% (244/903) Updating files: 28% (253/903) Updating files: 29% (262/903) Updating files: 30% (271/903) Updating files: 31% (280/903) Updating files: 32% (289/903) Updating files: 33% (298/903) Updating files: 34% (308/903) Updating files: 35% (317/903) Updating files: 36% (326/903) Updating files: 37% (335/903) Updating files: 38% (344/903) Updating files: 39% (353/903) Updating files: 40% (362/903) Updating files: 41% (371/903) Updating files: 42% (380/903) Updating files: 43% (389/903) Updating files: 44% (398/903) Updating files: 45% (407/903) Updating files: 46% (416/903) Updating files: 47% (425/903) Updating files: 48% (434/903) Updating files: 49% (443/903) Updating files: 50% (452/903) Updating files: 51% (461/903) Updating files: 52% (470/903) Updating files: 53% (479/903) Updating files: 54% (488/903) Updating files: 55% (497/903) Updating files: 56% (506/903) Updating files: 57% (515/903) Updating files: 58% (524/903) Updating files: 59% (533/903) Updating files: 60% (542/903) Updating files: 61% (551/903) Updating files: 62% (560/903) Updating files: 63% (569/903) Updating files: 64% (578/903) Updating files: 65% (587/903) Updating files: 66% (596/903) Updating files: 67% (606/903) Updating files: 68% (615/903) Updating files: 69% (624/903) Updating files: 70% (633/903) Updating files: 71% (642/903) Updating files: 72% (651/903) Updating files: 73% (660/903) Updating files: 74% (669/903) Updating files: 75% (678/903) Updating files: 76% (687/903) Updating files: 77% (696/903) Updating files: 78% (705/903) Updating files: 79% (714/903) Updating files: 80% (723/903) Updating files: 81% (732/903) Updating files: 82% (741/903) Updating files: 83% (750/903) Updating files: 84% (759/903) Updating files: 85% (768/903) Updating files: 86% (777/903) Updating files: 87% (786/903) Updating files: 88% (795/903) Updating files: 89% (804/903) Updating files: 90% (813/903) Updating files: 91% (822/903) Updating files: 92% (831/903) Updating files: 93% (840/903) Updating files: 94% (849/903) Updating files: 95% (858/903) Updating files: 96% (867/903) Updating files: 97% (876/903) Updating files: 98% (885/903) Updating files: 99% (894/903) Updating files: 100% (903/903) Updating files: 100% (903/903), done. +HEAD is now at a04ab56 fix: enforce qwen pre-install controls +/opt/qwen35-acceptance/lecore/REFERENCE.md: OK +/opt/qwen35-acceptance/inputs/federalist-papers.txt: OK +Cloning into '/opt/qwen35-acceptance/ilxyr'... +HEAD is now at e92382f Publish EXP-006 experiment evidence + Updating crates.io index + Downloading crates ... + Downloaded der v0.7.10 + Downloaded getrandom v0.2.17 + Downloaded digest v0.10.7 + Downloaded zeroize v1.9.0 + Downloaded cfg-if v1.0.4 + Downloaded serde_core v1.0.229 + Downloaded spki v0.7.3 + Downloaded sha2 v0.10.9 + Downloaded block-buffer v0.10.4 + Downloaded cpufeatures v0.2.17 + Downloaded crypto-common v0.1.7 + Downloaded signature v2.2.0 + Downloaded version_check v0.9.5 + Downloaded rustc_version v0.4.1 + Downloaded ed25519 v2.2.3 + Downloaded pkcs8 v0.10.2 + Downloaded quote v1.0.46 + Downloaded curve25519-dalek-derive v0.1.1 + Downloaded itoa v1.0.18 + Downloaded generic-array v0.14.7 + Downloaded subtle v2.6.1 + Downloaded zmij v1.0.23 + Downloaded rand_core v0.6.4 + Downloaded const-oid v0.9.6 + Downloaded memchr v2.8.3 + Downloaded base64ct v1.8.3 + Downloaded semver v1.0.28 + Downloaded serde_derive v1.0.229 + Downloaded unicode-ident v1.0.24 + Downloaded proc-macro2 v1.0.106 + Downloaded base64 v0.22.1 + Downloaded serde v1.0.229 + Downloaded typenum v1.20.1 + Downloaded ed25519-dalek v2.2.0 + Downloaded serde_json v1.0.150 + Downloaded syn v2.0.119 + Downloaded curve25519-dalek v4.1.3 + Downloaded syn v3.0.0 + Downloaded libc v0.2.186 + Compiling proc-macro2 v1.0.106 + Compiling version_check v0.9.5 + Compiling unicode-ident v1.0.24 + Compiling quote v1.0.46 + Compiling typenum v1.20.1 + Compiling semver v1.0.28 + Compiling serde_core v1.0.229 + Compiling cpufeatures v0.2.17 + Compiling zmij v1.0.23 + Compiling cfg-if v1.0.4 + Compiling serde_json v1.0.150 + Compiling subtle v2.6.1 + Compiling serde v1.0.229 + Compiling signature v2.2.0 + Compiling zeroize v1.9.0 + Compiling memchr v2.8.3 + Compiling itoa v1.0.18 + Compiling base64 v0.22.1 + Compiling ed25519 v2.2.3 + Compiling rustc_version v0.4.1 + Compiling generic-array v0.14.7 + Compiling curve25519-dalek v4.1.3 + Compiling syn v2.0.119 + Compiling syn v3.0.0 + Compiling block-buffer v0.10.4 + Compiling crypto-common v0.1.7 + Compiling digest v0.10.7 + Compiling sha2 v0.10.9 + Compiling serde_derive v1.0.229 + Compiling curve25519-dalek-derive v0.1.1 + Compiling ed25519-dalek v2.2.0 + Compiling ilxyr-core v0.1.0 (/opt/qwen35-acceptance/ilxyr/crates/ilxyr-core) + Compiling ilxyr-cli v0.1.0 (/opt/qwen35-acceptance/ilxyr/crates/ilxyr-cli) + Finished `release` profile [optimized] target(s) in 21.51s +Looking in indexes: https://download.pytorch.org/whl/cpu +Collecting torch==2.11.0 + Obtaining dependency information for torch==2.11.0 from https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata + Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata (29 kB) +Collecting filelock (from torch==2.11.0) + Obtaining dependency information for filelock from https://files.pythonhosted.org/packages/81/47/dd9a212ef6e343a6857485ffe25bba537304f1913bdbed446a23f7f592e1/filelock-3.29.0-py3-none-any.whl.metadata + Downloading filelock-3.29.0-py3-none-any.whl.metadata (2.0 kB) +Collecting typing-extensions>=4.10.0 (from torch==2.11.0) + Obtaining dependency information for typing-extensions>=4.10.0 from https://files.pythonhosted.org/packages/18/67/36e9267722cc04a6b9f15c7f3441c2363321a3ea07da7ae0c0707beb2a9c/typing_extensions-4.15.0-py3-none-any.whl.metadata + Downloading typing_extensions-4.15.0-py3-none-any.whl.metadata (3.3 kB) +Collecting setuptools<82 (from torch==2.11.0) + Obtaining dependency information for setuptools<82 from https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata (6.6 kB) +Collecting sympy>=1.13.3 (from torch==2.11.0) + Obtaining dependency information for sympy>=1.13.3 from https://files.pythonhosted.org/packages/a2/09/77d55d46fd61b4a135c444fc97158ef34a095e5681d0a6c10b75bf356191/sympy-1.14.0-py3-none-any.whl.metadata + Downloading sympy-1.14.0-py3-none-any.whl.metadata (12 kB) +Collecting networkx>=2.5.1 (from torch==2.11.0) + Obtaining dependency information for networkx>=2.5.1 from https://files.pythonhosted.org/packages/9e/c9/b2622292ea83fbb4ec318f5b9ab867d0a28ab43c5717bb85b0a5f6b3b0a4/networkx-3.6.1-py3-none-any.whl.metadata + Downloading networkx-3.6.1-py3-none-any.whl.metadata (6.8 kB) +Collecting jinja2 (from torch==2.11.0) + Obtaining dependency information for jinja2 from https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata (2.9 kB) +Collecting fsspec>=0.8.5 (from torch==2.11.0) + Obtaining dependency information for fsspec>=0.8.5 from https://files.pythonhosted.org/packages/d5/0c/043d5e551459da400957a1395e0febbf771446ff34291afcbe3d8be2a279/fsspec-2026.4.0-py3-none-any.whl.metadata + Downloading fsspec-2026.4.0-py3-none-any.whl.metadata (10 kB) +Collecting mpmath<1.4,>=1.1.0 (from sympy>=1.13.3->torch==2.11.0) + Obtaining dependency information for mpmath<1.4,>=1.1.0 from https://files.pythonhosted.org/packages/43/e3/7d92a15f894aa0c9c4b49b8ee9ac9850d6e63b03c9c32c0367a13ae62209/mpmath-1.3.0-py3-none-any.whl.metadata + Downloading mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) +Collecting MarkupSafe>=2.0 (from jinja2->torch==2.11.0) + Obtaining dependency information for MarkupSafe>=2.0 from https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.7 kB) +Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl (190.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 190.3/190.3 MB 13.8 MB/s eta 0:00:00 +Downloading fsspec-2026.4.0-py3-none-any.whl (203 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 203.4/203.4 kB 24.0 MB/s eta 0:00:00 +Downloading networkx-3.6.1-py3-none-any.whl (2.1 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 2.1/2.1 MB 159.2 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl (1.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.3/1.3 MB 188.0 MB/s eta 0:00:00 +Downloading sympy-1.14.0-py3-none-any.whl (6.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.3/6.3 MB 192.4 MB/s eta 0:00:00 +Downloading typing_extensions-4.15.0-py3-none-any.whl (44 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 44.6/44.6 kB 17.0 MB/s eta 0:00:00 +Downloading filelock-3.29.0-py3-none-any.whl (39 kB) +Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl (134 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 134.9/134.9 kB 48.2 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (22 kB) +Downloading mpmath-1.3.0-py3-none-any.whl (536 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 536.2/536.2 kB 133.6 MB/s eta 0:00:00 +Installing collected packages: mpmath, typing-extensions, sympy, setuptools, networkx, MarkupSafe, fsspec, filelock, jinja2, torch +Successfully installed MarkupSafe-3.0.3 filelock-3.29.0 fsspec-2026.4.0 jinja2-3.1.6 mpmath-1.3.0 networkx-3.6.1 setuptools-78.1.0 sympy-1.14.0 torch-2.11.0+cpu typing-extensions-4.15.0 +Collecting numpy==2.4.6 + Obtaining dependency information for numpy==2.4.6 from https://files.pythonhosted.org/packages/99/60/14115e6364fa676c5397c2ad3004e527e9aa487abf5d0706ec81bbd08529/numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (6.6 kB) +Collecting pillow==12.3.0 + Obtaining dependency information for pillow==12.3.0 from https://files.pythonhosted.org/packages/84/21/a35af28dcc61f37ed850a2d64c65c701321dfbf25085e469d5559360cbbf/pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (9.1 kB) +Collecting safetensors==0.8.0 + Obtaining dependency information for safetensors==0.8.0 from https://files.pythonhosted.org/packages/28/50/f203ff3a3ddfe19308efc83c5a3a29ed02bf786732ec35e68bf9162f3365/safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (4.2 kB) +Collecting huggingface-hub==1.27.0 + Obtaining dependency information for huggingface-hub==1.27.0 from https://files.pythonhosted.org/packages/de/d8/95b735e183957c1f26d94c52977f09d466d55119cbbc1558ea4975e4c216/huggingface_hub-1.27.0-py3-none-any.whl.metadata + Downloading huggingface_hub-1.27.0-py3-none-any.whl.metadata (16 kB) +Collecting transformers==5.14.0 + Obtaining dependency information for transformers==5.14.0 from https://files.pythonhosted.org/packages/19/3b/a0a807b957c5f8f2075a8740f597a021420583bb0aa68ff6432d74a530d4/transformers-5.14.0-py3-none-any.whl.metadata + Downloading transformers-5.14.0-py3-none-any.whl.metadata (32 kB) +Collecting click<9.0.0,>=8.4.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for click<9.0.0,>=8.4.2 from https://files.pythonhosted.org/packages/fb/e2/79c688af8b210d232694e31e59da9f6ec747bae31c3f5946e4e9b98860d5/click-8.4.2-py3-none-any.whl.metadata + Downloading click-8.4.2-py3-none-any.whl.metadata (2.6 kB) +Requirement already satisfied: filelock>=3.10.0 in /opt/qwen35-acceptance/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (3.29.0) +Requirement already satisfied: fsspec>=2023.5.0 in /opt/qwen35-acceptance/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (2026.4.0) +Collecting hf-xet<2.0.0,>=1.5.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for hf-xet<2.0.0,>=1.5.2 from https://files.pythonhosted.org/packages/67/4e/a28359bf1c1ecf11eba22123168c138698f7cb576ac678f5a2e16cd5da08/hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata + Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (4.9 kB) +Collecting httpx<1,>=0.23.0 (from huggingface-hub==1.27.0) + Obtaining dependency information for httpx<1,>=0.23.0 from https://files.pythonhosted.org/packages/2a/39/e50c7c3a983047577ee07d2a9e53faf5a69493943ec3f6a384bdc792deb2/httpx-0.28.1-py3-none-any.whl.metadata + Downloading httpx-0.28.1-py3-none-any.whl.metadata (7.1 kB) +Collecting packaging>=20.9 (from huggingface-hub==1.27.0) + Obtaining dependency information for packaging>=20.9 from https://files.pythonhosted.org/packages/63/34/ba1c580383c9eada3711951fef0795c80b829a078d72188184bcab9dd527/packaging-26.3-py3-none-any.whl.metadata + Downloading packaging-26.3-py3-none-any.whl.metadata (3.5 kB) +Collecting pyyaml>=5.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for pyyaml>=5.1 from https://files.pythonhosted.org/packages/8b/9d/b3589d3877982d4f2329302ef98a8026e7f4443c765c46cfecc8858c6b4b/pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.4 kB) +Collecting tqdm>=4.42.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for tqdm>=4.42.1 from https://files.pythonhosted.org/packages/f9/1c/01bfd571a64e7f270e6bab5e33777debe0edc56759233ce84f27dec92d14/tqdm-4.70.0-py3-none-any.whl.metadata + Downloading tqdm-4.70.0-py3-none-any.whl.metadata (57 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 57.3/57.3 kB 13.8 MB/s eta 0:00:00 +Requirement already satisfied: typing-extensions>=4.1.0 in /opt/qwen35-acceptance/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (4.15.0) +Collecting regex>=2025.10.22 (from transformers==5.14.0) + Obtaining dependency information for regex>=2025.10.22 from https://files.pythonhosted.org/packages/2a/8e/096d00c7c480ef2ff4265349b14e2261d4ab787ba1f74e2e80d1c58079c3/regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (40 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 40.5/40.5 kB 12.7 MB/s eta 0:00:00 +Collecting tokenizers<=0.23.0,>=0.22.0 (from transformers==5.14.0) + Obtaining dependency information for tokenizers<=0.23.0,>=0.22.0 from https://files.pythonhosted.org/packages/2e/76/932be4b50ef6ccedf9d3c6639b056a967a86258c6d9200643f01269211ca/tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (7.3 kB) +Collecting typer (from transformers==5.14.0) + Obtaining dependency information for typer from https://files.pythonhosted.org/packages/43/89/9518bc0c3929bee36b3a4a8e3daddd6e03f92f9961c66d4983b837160543/typer-0.27.1-py3-none-any.whl.metadata + Downloading typer-0.27.1-py3-none-any.whl.metadata (16 kB) +Collecting anyio (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for anyio from https://files.pythonhosted.org/packages/da/35/f2287558c17e29fafc8ef3daf819bb9834061cfa43bff8014f7df7f63bdc/anyio-4.14.2-py3-none-any.whl.metadata + Downloading anyio-4.14.2-py3-none-any.whl.metadata (4.6 kB) +Collecting certifi (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for certifi from https://files.pythonhosted.org/packages/0b/a7/71ac2cff56fec219ed242bb11b8efb69fcc4bec75db06fb7bfe35de520e6/certifi-2026.7.22-py3-none-any.whl.metadata + Downloading certifi-2026.7.22-py3-none-any.whl.metadata (2.5 kB) +Collecting httpcore==1.* (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for httpcore==1.* from https://files.pythonhosted.org/packages/7e/f5/f66802a942d491edb555dd61e3a9961140fd64c90bce1eafd741609d334d/httpcore-1.0.9-py3-none-any.whl.metadata + Downloading httpcore-1.0.9-py3-none-any.whl.metadata (21 kB) +Collecting idna (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for idna from https://files.pythonhosted.org/packages/1e/5e/d4e9f1a599fb8e573b7b87160658329fbf28d19eac2718f51fc3def3aa5a/idna-3.18-py3-none-any.whl.metadata + Downloading idna-3.18-py3-none-any.whl.metadata (6.1 kB) +Collecting h11>=0.16 (from httpcore==1.*->httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for h11>=0.16 from https://files.pythonhosted.org/packages/04/4b/29cac41a4d98d144bf5f6d33995617b185d14b22401f75ca86f384e87ff1/h11-0.16.0-py3-none-any.whl.metadata + Downloading h11-0.16.0-py3-none-any.whl.metadata (8.3 kB) +Collecting shellingham>=1.3.0 (from typer->transformers==5.14.0) + Obtaining dependency information for shellingham>=1.3.0 from https://files.pythonhosted.org/packages/e0/f9/0595336914c5619e5f28a1fb793285925a8cd4b432c9da0a987836c7f822/shellingham-1.5.4-py2.py3-none-any.whl.metadata + Downloading shellingham-1.5.4-py2.py3-none-any.whl.metadata (3.5 kB) +Collecting rich>=13.8.0 (from typer->transformers==5.14.0) + Obtaining dependency information for rich>=13.8.0 from https://files.pythonhosted.org/packages/82/3b/64d4899d73f91ba49a8c18a8ff3f0ea8f1c1d75481760df8c68ef5235bf5/rich-15.0.0-py3-none-any.whl.metadata + Downloading rich-15.0.0-py3-none-any.whl.metadata (18 kB) +Collecting annotated-doc>=0.0.2 (from typer->transformers==5.14.0) + Obtaining dependency information for annotated-doc>=0.0.2 from https://files.pythonhosted.org/packages/3e/30/e900b21425a860e195f32e37657aa1f7c7f2b1bfb26f03ca209b90933c06/annotated_doc-0.0.5-py3-none-any.whl.metadata + Downloading annotated_doc-0.0.5-py3-none-any.whl.metadata (6.5 kB) +Collecting markdown-it-py>=2.2.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for markdown-it-py>=2.2.0 from https://files.pythonhosted.org/packages/b3/81/4da04ced5a082363ecfa159c010d200ecbd959ae410c10c0264a38cac0f5/markdown_it_py-4.2.0-py3-none-any.whl.metadata + Downloading markdown_it_py-4.2.0-py3-none-any.whl.metadata (7.4 kB) +Collecting pygments<3.0.0,>=2.13.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for pygments<3.0.0,>=2.13.0 from https://files.pythonhosted.org/packages/f4/7e/a72dd26f3b0f4f2bf1dd8923c85f7ceb43172af56d63c7383eb62b332364/pygments-2.20.0-py3-none-any.whl.metadata + Downloading pygments-2.20.0-py3-none-any.whl.metadata (2.5 kB) +Collecting mdurl~=0.1 (from markdown-it-py>=2.2.0->rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for mdurl~=0.1 from https://files.pythonhosted.org/packages/b3/38/89ba8ad64ae25be8de66a6d463314cf1eb366222074cfda9ee839c56a4b4/mdurl-0.1.2-py3-none-any.whl.metadata + Downloading mdurl-0.1.2-py3-none-any.whl.metadata (1.6 kB) +Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (16.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 16.6/16.6 MB 100.4 MB/s eta 0:00:00 +Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.9 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.9/6.9 MB 156.3 MB/s eta 0:00:00 +Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (516 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 516.0/516.0 kB 94.8 MB/s eta 0:00:00 +Downloading huggingface_hub-1.27.0-py3-none-any.whl (784 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 784.9/784.9 kB 118.7 MB/s eta 0:00:00 +Downloading transformers-5.14.0-py3-none-any.whl (11.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 11.6/11.6 MB 104.4 MB/s eta 0:00:00 +Downloading click-8.4.2-py3-none-any.whl (119 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 119.2/119.2 kB 29.2 MB/s eta 0:00:00 +Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (4.5 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 4.5/4.5 MB 222.0 MB/s eta 0:00:00 +Downloading httpx-0.28.1-py3-none-any.whl (73 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 73.5/73.5 kB 25.9 MB/s eta 0:00:00 +Downloading httpcore-1.0.9-py3-none-any.whl (78 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 78.8/78.8 kB 19.8 MB/s eta 0:00:00 +Downloading packaging-26.3-py3-none-any.whl (129 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 130.0/130.0 kB 27.7 MB/s eta 0:00:00 +Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (807 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 807.9/807.9 kB 120.9 MB/s eta 0:00:00 +Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (801 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 801.8/801.8 kB 118.4 MB/s eta 0:00:00 +Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (3.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 3.3/3.3 MB 202.6 MB/s eta 0:00:00 +Downloading tqdm-4.70.0-py3-none-any.whl (80 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 80.2/80.2 kB 29.4 MB/s eta 0:00:00 +Downloading typer-0.27.1-py3-none-any.whl (122 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 122.9/122.9 kB 38.6 MB/s eta 0:00:00 +Downloading annotated_doc-0.0.5-py3-none-any.whl (5.3 kB) +Downloading rich-15.0.0-py3-none-any.whl (310 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 310.7/310.7 kB 65.7 MB/s eta 0:00:00 +Downloading shellingham-1.5.4-py2.py3-none-any.whl (9.8 kB) +Downloading anyio-4.14.2-py3-none-any.whl (125 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 125.8/125.8 kB 45.2 MB/s eta 0:00:00 +Downloading idna-3.18-py3-none-any.whl (65 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 65.5/65.5 kB 23.7 MB/s eta 0:00:00 +Downloading certifi-2026.7.22-py3-none-any.whl (136 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 137.0/137.0 kB 41.6 MB/s eta 0:00:00 +Downloading h11-0.16.0-py3-none-any.whl (37 kB) +Downloading markdown_it_py-4.2.0-py3-none-any.whl (91 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 91.7/91.7 kB 33.1 MB/s eta 0:00:00 +Downloading pygments-2.20.0-py3-none-any.whl (1.2 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.2/1.2 MB 169.4 MB/s eta 0:00:00 +Downloading mdurl-0.1.2-py3-none-any.whl (10.0 kB) +Installing collected packages: tqdm, shellingham, safetensors, regex, pyyaml, pygments, pillow, packaging, numpy, mdurl, idna, hf-xet, h11, click, certifi, annotated-doc, markdown-it-py, httpcore, anyio, rich, httpx, typer, huggingface-hub, tokenizers, transformers +Successfully installed annotated-doc-0.0.5 anyio-4.14.2 certifi-2026.7.22 click-8.4.2 h11-0.16.0 hf-xet-1.6.0 httpcore-1.0.9 httpx-0.28.1 huggingface-hub-1.27.0 idna-3.18 markdown-it-py-4.2.0 mdurl-0.1.2 numpy-2.4.6 packaging-26.3 pillow-12.3.0 pygments-2.20.0 pyyaml-6.0.3 regex-2026.7.19 rich-15.0.0 safetensors-0.8.0 shellingham-1.5.4 tokenizers-0.22.2 tqdm-4.70.0 transformers-5.14.0 typer-0.27.1 + +[notice] A new release of pip is available: 23.2.1 -> 26.2.1 +[notice] To update, run: python3.12 -m pip install --upgrade pip +No broken requirements found. +torch 2.11.0+cpu +transformers 5.14.0 +AutoProcessor Qwen3_5ForConditionalGeneration + Fetching 13 files: 0%| | 0/13 [00:00 +ilxyr contribute +ilxyr shared-task-register +ilxyr compile +ilxyr retro --execute +ilxyr forecast +ilxyr fund +ilxyr trust-key +ilxyr budget-payload +ilxyr budget-register +ilxyr allocate ... +ilxyr admit +ilxyr run --execute +ilxyr authorize +ilxyr run-auto +ilxyr loop-cycle +ilxyr sandbox +ilxyr certify +ilxyr calibration +ilxyr status +ilxyr verify diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/execution-summary.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/execution-summary.json new file mode 100644 index 0000000..c8a6187 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/execution-summary.json @@ -0,0 +1,8 @@ +{ + "completed_at": "2026-08-12T21:47:27.702547+00:00", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "run_exit_code": 0, + "schema": "lecore.qwen35-execution-summary.v1", + "status_exit_code": 0, + "verify_exit_code": 0 +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/generation.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/generation.json new file mode 100644 index 0000000..3f0c99e --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/generation.json @@ -0,0 +1 @@ +{"experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", "project": "/opt/qwen35-acceptance/project/project.json", "project_dir": "/opt/qwen35-acceptance/project"} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-admission.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-admission.json new file mode 100644 index 0000000..c40fdc1 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-admission.json @@ -0,0 +1,68 @@ +{ + "schema": "ilxyr.admission.v1", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "accepted": true, + "checks": [ + { + "gate": "methodology", + "passed": true, + "detail": "4/4 required research stages resolved" + }, + { + "gate": "outcome_contract", + "passed": true, + "detail": "3 frozen mutually evaluated outcomes declared" + }, + { + "gate": "forecast_participation", + "passed": true, + "detail": "2/2 distinct forecasters" + }, + { + "gate": "forecast_stake", + "passed": true, + "detail": "10/10 forecast credits staked" + }, + { + "gate": "compute_funding", + "passed": true, + "detail": "100/100 compute credits committed" + }, + { + "gate": "executor_available", + "passed": true, + "detail": "local-command adapter is installed" + }, + { + "gate": "weight_protection", + "passed": true, + "detail": "local execution is limited to public-weight handles" + }, + { + "gate": "local_execution_policy", + "passed": true, + "detail": "local execution requires network=open and an absolute executable path" + }, + { + "gate": "code_policy", + "passed": true, + "detail": "local-command directly executes the declared program" + }, + { + "gate": "export_policy", + "passed": true, + "detail": "local-command records stdout, stderr, and metric artifacts" + }, + { + "gate": "reviewer_separation", + "passed": true, + "detail": "proposer and engineering reviewer use distinct handles" + }, + { + "gate": "forecaster_separation", + "passed": true, + "detail": "proposer and forecasters use distinct handles" + } + ], + "decided_at_ms": 1786571246029 +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt new file mode 100644 index 0000000..c29264e --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt @@ -0,0 +1,7 @@ +commit e92382ff2a5e8714466533a160f6609b4ef9cee8 +Author: atimics +AuthorDate: Sat Aug 8 12:46:56 2026 -0700 +Commit: atimics +CommitDate: Sat Aug 8 12:46:56 2026 -0700 + + Publish EXP-006 experiment evidence diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-compile.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-compile.json new file mode 100644 index 0000000..74db857 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-compile.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json new file mode 100644 index 0000000..aebc77a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json new file mode 100644 index 0000000..740562f --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json new file mode 100644 index 0000000..587d9fd --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json new file mode 100644 index 0000000..b627783 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json new file mode 100644 index 0000000..0b9fbe0 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/52df3c246b4b322591efcd5402e1bbc04ed5df3d32aacd1f490782dc030f47f2" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json new file mode 100644 index 0000000..5cb356b --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/bcf9449a813715c1ff821104588fa7f4169db64da5435b93d698beae6ab09ef2" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-funding.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-funding.json new file mode 100644 index 0000000..dc18f9b --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-funding.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/edbcba5124b657e19d76e7728791cb2a29a9f5d9dfb57247a165e46a06833a47" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-init.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-init.json new file mode 100644 index 0000000..9baebdc --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-init.json @@ -0,0 +1,4 @@ +{ + "initialized": true, + "workspace": "/opt/qwen35-acceptance/workspace" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-run.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-run.json new file mode 100644 index 0000000..3cdf8a0 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-run.json @@ -0,0 +1,94 @@ +{ + "run": { + "schema": "ilxyr.run.v1", + "id": "run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "started_at_ms": 1786571246690, + "completed_at_ms": 1786571246741, + "exit_code": 1, + "timed_out": false, + "stdout": "", + "stderr": "Traceback (most recent call last):\n File \"/opt/qwen35-acceptance/lecore/experiments/qwen35_acceptance/run.py\", line 17, in \n import numpy as np\nModuleNotFoundError: No module named 'numpy'\n", + "output_truncated": false, + "metrics": {} + }, + "evidence": { + "schema": "ilxyr.evidence.v1", + "id": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "run_ref": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731", + "resolved_outcome": "execution_failure", + "metrics": {}, + "recorded_at_ms": 1786571246771, + "authority": { + "level": "corpus_proxy", + "scope": { + "seeds": [ + 0 + ], + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "coverage": 1.0 + }, + "provenance": { + "artifact_hashes": [ + "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e", + "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731" + ], + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "checker": "checker://lecore/qwen35-acceptance/v1" + } + }, + "lane": "promoted" + }, + "settlements": [ + { + "schema": "ilxyr.forecast_settlement.v1", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "resolved_outcome": "execution_failure", + "brier_score": 1.185, + "stake": 5 + }, + { + "schema": "ilxyr.forecast_settlement.v1", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "resolved_outcome": "execution_failure", + "brier_score": 0.815, + "stake": 5 + } + ], + "calibrations": [ + { + "schema": "ilxyr.calibration_record.v1", + "handle": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical", + "forecasts_settled": 1, + "reliability": 1.185, + "resolution": 0.0, + "brier_score": 1.185, + "uncertainty": 0.0, + "probationary": true, + "last_settlement_at_ms": 1786571246828, + "forecast_ids": [ + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast.empirical" + ] + }, + { + "schema": "ilxyr.calibration_record.v1", + "handle": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic", + "forecasts_settled": 1, + "reliability": 0.815, + "resolution": 0.0, + "brier_score": 0.815, + "uncertainty": 0.0, + "probationary": true, + "last_settlement_at_ms": 1786571246900, + "forecast_ids": [ + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.forecast-mechanistic" + ] + } + ] +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-status.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-status.json new file mode 100644 index 0000000..93a8de8 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-status.json @@ -0,0 +1,121 @@ +{ + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "compiled_ref": "artifact://sha256/3d4c5b79be266085e4a3495a83cbfe3c4c8434c230bef81e7b6a19ed7eeb3ff6", + "forecasts": 2, + "total_stake": 10, + "funding_commitments": 1, + "funded_compute_credits": 100, + "latest_admission": { + "schema": "ilxyr.admission.v1", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "accepted": true, + "checks": [ + { + "gate": "methodology", + "passed": true, + "detail": "4/4 required research stages resolved" + }, + { + "gate": "outcome_contract", + "passed": true, + "detail": "3 frozen mutually evaluated outcomes declared" + }, + { + "gate": "forecast_participation", + "passed": true, + "detail": "2/2 distinct forecasters" + }, + { + "gate": "forecast_stake", + "passed": true, + "detail": "10/10 forecast credits staked" + }, + { + "gate": "compute_funding", + "passed": true, + "detail": "100/100 compute credits committed" + }, + { + "gate": "executor_available", + "passed": true, + "detail": "local-command adapter is installed" + }, + { + "gate": "weight_protection", + "passed": true, + "detail": "local execution is limited to public-weight handles" + }, + { + "gate": "local_execution_policy", + "passed": true, + "detail": "local execution requires network=open and an absolute executable path" + }, + { + "gate": "code_policy", + "passed": true, + "detail": "local-command directly executes the declared program" + }, + { + "gate": "export_policy", + "passed": true, + "detail": "local-command records stdout, stderr, and metric artifacts" + }, + { + "gate": "reviewer_separation", + "passed": true, + "detail": "proposer and engineering reviewer use distinct handles" + }, + { + "gate": "forecaster_separation", + "passed": true, + "detail": "proposer and forecasters use distinct handles" + } + ], + "decided_at_ms": 1786571246029 + }, + "execution_started": true, + "latest_run": { + "schema": "ilxyr.run.v1", + "id": "run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "started_at_ms": 1786571246690, + "completed_at_ms": 1786571246741, + "exit_code": 1, + "timed_out": false, + "stdout": "", + "stderr": "Traceback (most recent call last):\n File \"/opt/qwen35-acceptance/lecore/experiments/qwen35_acceptance/run.py\", line 17, in \n import numpy as np\nModuleNotFoundError: No module named 'numpy'\n", + "output_truncated": false, + "metrics": {} + }, + "latest_evidence": { + "schema": "ilxyr.evidence.v1", + "id": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance:1786571246690", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "run_ref": "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731", + "resolved_outcome": "execution_failure", + "metrics": {}, + "recorded_at_ms": 1786571246771, + "authority": { + "level": "corpus_proxy", + "scope": { + "seeds": [ + 0 + ], + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "coverage": 1.0 + }, + "provenance": { + "artifact_hashes": [ + "artifact://sha256/04bed9185d0a9647fea419ac6cb2f671a106ae60bc82c23b172aacff6dbcaf0b", + "artifact://sha256/3b4c72fe06906bda15e22dac5fd12f9d1b1347f687f3656bbaa6ff3c3d11cbd1", + "artifact://sha256/50519e75792a64ca250e673ed7b05be888419d53969be5ec4c074fb50ed8ebb7", + "artifact://sha256/826a9741b544257f3658ecb4014abc4bcaee45a33be7e5e467dc1cb2786fc17e", + "artifact://sha256/cfabbc3d6aaa06b860e8126e0a2a9c2ab9e1151cd1ad8af7d9cfaf804c5eb731" + ], + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "checker": "checker://lecore/qwen35-acceptance/v1" + } + }, + "lane": "promoted" + } +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-verify.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-verify.json new file mode 100644 index 0000000..d3d17dd --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/ilxyr-verify.json @@ -0,0 +1,5 @@ +{ + "objects_checked": 15, + "events_checked": 16, + "valid": true +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/launch-manifest.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/launch-manifest.json new file mode 100644 index 0000000..d77227a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/launch-manifest.json @@ -0,0 +1,98 @@ +{ + "schema": "lecore.qwen35-launch.v1", + "state": "authorized_prelaunch", + "attempt": { + "maximum_formal_runs": 1, + "retry_after_admission": false, + "seed": 0 + }, + "source": { + "repository": "https://github.com/atimics/holostuff.git", + "commit": "a04ab5692be38f06120aba4b0bc5e2a284eb2c79" + }, + "ilxyr": { + "repository": "https://github.com/cenetex/ilXyr.git", + "commit": "e92382ff2a5e8714466533a160f6609b4ef9cee8" + }, + "model": { + "repository": "Qwen/Qwen3.5-0.8B", + "revision": "2fc06364715b967f1860aea9cf38778875588b17", + "license": "Apache-2.0", + "critical_files": [ + { + "path": "model.safetensors-00001-of-00001.safetensors", + "bytes": 1746942600, + "sha256": "04b1c301231dd422b8860db31311ab2721511346a32cb1e079c4c4e5f1fe4696" + }, + { + "path": "tokenizer.json", + "bytes": 12807982, + "sha256": "5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42" + } + ], + "complete_manifest": "generated and frozen before ilxyr admission" + }, + "corpora": { + "installation": { + "handle": "git:a04ab5692be38f06120aba4b0bc5e2a284eb2c79:REFERENCE.md", + "license": "MIT", + "bytes": 2300089, + "sha256": "d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f" + }, + "evaluation": { + "handle": "https://www.gutenberg.org/ebooks/18", + "retrieved_at": "2026-08-12T21:32:31Z", + "last_modified": "2026-08-01T07:33:54Z", + "copyright": "Public domain in the USA", + "bytes": 1213410, + "sha256": "a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + } + }, + "method": { + "minimum_paired_positions": 4096, + "reference_logit_relative_tolerance": 0.001, + "maximum_regression": 0.01, + "confidence": 0.95, + "bootstrap": "paired moving block", + "spectral_filtering": false, + "installer_experimental": true + }, + "dependencies": { + "python": "3.12", + "numpy": "2.4.6", + "pillow": "12.3.0", + "torch": "2.11.0+cpu", + "transformers": "5.14.0", + "resolved_environment": "recorded before ilxyr admission" + }, + "aws": { + "region": "us-east-1", + "availability_zone": "us-east-1a", + "ami": "ami-07a5b367e8dc8bd92", + "ami_name": "al2023-ami-2023.12.20260803.3-kernel-6.1-x86_64", + "instance_type": "r7i.4xlarge", + "vcpu": 16, + "memory_gib": 128, + "market": "on-demand", + "hourly_compute_usd": 1.0584, + "root_volume": { + "type": "gp3", + "size_gib": 100, + "encrypted": true, + "delete_on_termination": true + }, + "inbound_ports": [], + "experiment_timeout_seconds": 21600, + "instance_lifetime_seconds": 28800, + "instance_initiated_shutdown_behavior": "terminate", + "maximum_compute_usd": 8.4672, + "total_ceiling_usd": 10.0 + }, + "publication": { + "temporary_staging": "s3://zero-training-022118847419/qwen35-acceptance/a04ab56-2fc06364-a6c9d113/", + "review_index": "https://github.com/atimics/holostuff/pull/3", + "permanent_transport": "Arweave", + "publish_all_outcomes": true, + "exclude_model_weights": true + } +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/lecore-commit.txt b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/lecore-commit.txt new file mode 100644 index 0000000..ad0a2c0 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/lecore-commit.txt @@ -0,0 +1,7 @@ +commit a04ab5692be38f06120aba4b0bc5e2a284eb2c79 +Author: atimics +AuthorDate: Wed Aug 12 14:31:25 2026 -0700 +Commit: atimics +CommitDate: Wed Aug 12 14:31:25 2026 -0700 + + fix: enforce qwen pre-install controls diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/lecore-status.txt b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/lecore-status.txt new file mode 100644 index 0000000..e69de29 diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/model-manifest.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/model-manifest.json new file mode 100644 index 0000000..65559ac --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/model-manifest.json @@ -0,0 +1,66 @@ +{ + "digest": "20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "files": [ + { + "bytes": 11544, + "path": "LICENSE", + "sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a" + }, + { + "bytes": 61705, + "path": "README.md", + "sha256": "87a163af54f32fa608a0f8d3ac67945c53dd2b4a7c96740b3d7fdc28e8458864" + }, + { + "bytes": 7755, + "path": "chat_template.jinja", + "sha256": "273d8e0e683b885071fb17e08d71e5f2a5ddfb5309756181681de4f5a1822d80" + }, + { + "bytes": 2907, + "path": "config.json", + "sha256": "b90b86f35c8e6925ef74ee04d0e758f0a845c83a42089ad82bbaa948de9b4204" + }, + { + "bytes": 3353259, + "path": "merges.txt", + "sha256": "a9d356d7bdf1ef4949e3e748e95b8e10ad9d4e2e838eddc38a0a7b6b94d1db8d" + }, + { + "bytes": 1746942600, + "path": "model.safetensors-00001-of-00001.safetensors", + "sha256": "04b1c301231dd422b8860db31311ab2721511346a32cb1e079c4c4e5f1fe4696" + }, + { + "bytes": 50900, + "path": "model.safetensors.index.json", + "sha256": "d8a08838a613b025eb7952ed9db11696213e57e76a375661ef5c12f9dd5dcf4e" + }, + { + "bytes": 390, + "path": "preprocessor_config.json", + "sha256": "27225450ac9c6529872ee1924fcb0962ff5634834f817040f444118116f4e516" + }, + { + "bytes": 12807982, + "path": "tokenizer.json", + "sha256": "5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42" + }, + { + "bytes": 16709, + "path": "tokenizer_config.json", + "sha256": "49e2b6e395f959f077f1e992b338919c0d4a9732fc6e613995e06557f843500c" + }, + { + "bytes": 385, + "path": "video_preprocessor_config.json", + "sha256": "7768af27c1fafa9cc9011c1dc20067e03f8915e03b63504550e11d5066986d13" + }, + { + "bytes": 6722759, + "path": "vocab.json", + "sha256": "ce99b4cb2983d118806ce0a8b777a35b093e2000a503ebde25853284c9dfa003" + } + ], + "revision": "2fc06364715b967f1860aea9cf38778875588b17" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/requirements-resolved.txt b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/requirements-resolved.txt new file mode 100644 index 0000000..7bfa4f7 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/requirements-resolved.txt @@ -0,0 +1,36 @@ +Jinja2==3.1.6 +MarkupSafe==3.0.3 +PyYAML==6.0.3 +Pygments==2.20.0 +annotated-doc==0.0.5 +anyio==4.14.2 +certifi==2026.7.22 +click==8.4.2 +filelock==3.29.0 +fsspec==2026.4.0 +h11==0.16.0 +hf-xet==1.6.0 +httpcore==1.0.9 +httpx==0.28.1 +huggingface_hub==1.27.0 +idna==3.18 +markdown-it-py==4.2.0 +mdurl==0.1.2 +mpmath==1.3.0 +networkx==3.6.1 +numpy==2.4.6 +packaging==26.3 +pillow==12.3.0 +pip==23.2.1 +regex==2026.7.19 +rich==15.0.0 +safetensors==0.8.0 +setuptools==78.1.0 +shellingham==1.5.4 +sympy==1.14.0 +tokenizers==0.22.2 +torch==2.11.0+cpu +tqdm==4.70.0 +transformers==5.14.0 +typer==0.27.1 +typing_extensions==4.15.0 diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/runner-status.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/runner-status.json new file mode 100644 index 0000000..ebf3eff --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/runner-status.json @@ -0,0 +1,7 @@ +{ + "exit_code": 0, + "formal_run_started": true, + "schema": "lecore.qwen35-runner-status.v1", + "stage": "complete", + "updated_at": "2026-08-12T21:47:27.721607+00:00" +} diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/user-data.log b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/user-data.log new file mode 100644 index 0000000..8d705db --- /dev/null +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/result/user-data.log @@ -0,0 +1,464 @@ +Shutdown scheduled for Thu 2026-08-13 05:45:38 UTC, use 'shutdown -c' to cancel. +Amazon Linux 2023 repository 80 MB/s | 75 MB 00:00 +Amazon Linux 2023 Kernel Livepatch repository 506 kB/s | 69 kB 00:00 +Last metadata expiration check: 0:00:01 ago on Wed Aug 12 21:45:54 2026. +Package tar-2:1.34-1.amzn2023.0.4.x86_64 is already installed. +Package gzip-1.12-1.amzn2023.0.1.x86_64 is already installed. +Dependencies resolved. +================================================================================ + Package Arch Version Repository Size +================================================================================ +Installing: + cargo x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 8.2 M + git x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 53 k + python3.12 x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 27 k + python3.12-pip noarch 23.2.1-4.amzn2023.0.10 amazonlinux 2.8 M + rust x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 62 M +Installing dependencies: + annobin-docs noarch 12.69-1.amzn2023.0.1 amazonlinux 93 k + annobin-plugin-gcc x86_64 12.69-1.amzn2023.0.1 amazonlinux 974 k + cpp x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 11 M + gc x86_64 8.0.4-5.amzn2023.0.2 amazonlinux 105 k + gcc x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 36 M + gcc-plugin-annobin x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 38 k + git-core x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 4.9 M + git-core-doc noarch 2.50.1-1.amzn2023.0.1 amazonlinux 2.8 M + glibc-devel x86_64 2.34-231.amzn2023.0.5 amazonlinux 33 k + glibc-headers-x86 noarch 2.34-231.amzn2023.0.5 amazonlinux 439 k + guile22 x86_64 2.2.7-2.amzn2023.0.3 amazonlinux 6.4 M + kernel-headers x86_64 1:6.1.177-224.371.amzn2023 amazonlinux 1.4 M + libmpc x86_64 1.2.1-2.amzn2023.0.2 amazonlinux 62 k + libtool-ltdl x86_64 2.4.7-1.amzn2023.0.3 amazonlinux 38 k + libxcrypt-devel x86_64 4.4.33-7.amzn2023 amazonlinux 32 k + make x86_64 1:4.3-5.amzn2023.0.2 amazonlinux 534 k + mpdecimal x86_64 2.5.1-3.amzn2023.0.3 amazonlinux 101 k + perl-Error noarch 1:0.17030-2.amzn2023.0.1 amazonlinux 42 k + perl-File-Find noarch 1.37-477.amzn2023.0.9 amazonlinux 26 k + perl-Git noarch 2.50.1-1.amzn2023.0.1 amazonlinux 41 k + perl-TermReadKey x86_64 2.38-9.amzn2023.0.3 amazonlinux 36 k + perl-lib x86_64 0.65-477.amzn2023.0.9 amazonlinux 15 k + python3.12-libs x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 9.2 M + python3.12-pip-wheel noarch 23.2.1-4.amzn2023.0.10 amazonlinux 1.5 M + rust-std-static x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 39 M +Installing weak dependencies: + python3.12-setuptools noarch 68.2.2-4.amzn2023.0.3 amazonlinux 1.4 M + +Transaction Summary +================================================================================ +Install 31 Packages + +Total download size: 189 M +Installed size: 700 M +Downloading Packages: +(1/31): annobin-docs-12.69-1.amzn2023.0.1.noarc 3.0 MB/s | 93 kB 00:00 +(2/31): annobin-plugin-gcc-12.69-1.amzn2023.0.1 22 MB/s | 974 kB 00:00 +(3/31): gc-8.0.4-5.amzn2023.0.2.x86_64.rpm 3.7 MB/s | 105 kB 00:00 +(4/31): cpp-11.5.0-5.amzn2023.0.5.x86_64.rpm 76 MB/s | 11 MB 00:00 +(5/31): gcc-plugin-annobin-11.5.0-5.amzn2023.0. 1.3 MB/s | 38 kB 00:00 +(6/31): git-2.50.1-1.amzn2023.0.1.x86_64.rpm 2.3 MB/s | 53 kB 00:00 +(7/31): git-core-2.50.1-1.amzn2023.0.1.x86_64.r 74 MB/s | 4.9 MB 00:00 +(8/31): cargo-1.97.0-1.amzn2023.0.1.x86_64.rpm 27 MB/s | 8.2 MB 00:00 +(9/31): git-core-doc-2.50.1-1.amzn2023.0.1.noar 73 MB/s | 2.8 MB 00:00 +(10/31): glibc-devel-2.34-231.amzn2023.0.5.x86_ 1.4 MB/s | 33 kB 00:00 +(11/31): glibc-headers-x86-2.34-231.amzn2023.0. 16 MB/s | 439 kB 00:00 +(12/31): kernel-headers-6.1.177-224.371.amzn202 49 MB/s | 1.4 MB 00:00 +(13/31): guile22-2.2.7-2.amzn2023.0.3.x86_64.rp 73 MB/s | 6.4 MB 00:00 +(14/31): libmpc-1.2.1-2.amzn2023.0.2.x86_64.rpm 1.7 MB/s | 62 kB 00:00 +(15/31): libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_ 1.7 MB/s | 38 kB 00:00 +(16/31): libxcrypt-devel-4.4.33-7.amzn2023.x86_ 1.4 MB/s | 32 kB 00:00 +(17/31): gcc-11.5.0-5.amzn2023.0.5.x86_64.rpm 79 MB/s | 36 MB 00:00 +(18/31): mpdecimal-2.5.1-3.amzn2023.0.3.x86_64. 1.1 MB/s | 101 kB 00:00 +(19/31): make-4.3-5.amzn2023.0.2.x86_64.rpm 5.5 MB/s | 534 kB 00:00 +(20/31): perl-Error-0.17030-2.amzn2023.0.1.noar 1.4 MB/s | 42 kB 00:00 +(21/31): perl-File-Find-1.37-477.amzn2023.0.9.n 872 kB/s | 26 kB 00:00 +(22/31): perl-Git-2.50.1-1.amzn2023.0.1.noarch. 1.4 MB/s | 41 kB 00:00 +(23/31): perl-TermReadKey-2.38-9.amzn2023.0.3.x 1.4 MB/s | 36 kB 00:00 +(24/31): python3.12-3.12.13-2.amzn2023.0.5.x86_ 1.1 MB/s | 27 kB 00:00 +(25/31): perl-lib-0.65-477.amzn2023.0.9.x86_64. 584 kB/s | 15 kB 00:00 +(26/31): python3.12-pip-wheel-23.2.1-4.amzn2023 46 MB/s | 1.5 MB 00:00 +(27/31): python3.12-pip-23.2.1-4.amzn2023.0.10. 63 MB/s | 2.8 MB 00:00 +(28/31): python3.12-setuptools-68.2.2-4.amzn202 34 MB/s | 1.4 MB 00:00 +(29/31): python3.12-libs-3.12.13-2.amzn2023.0.5 75 MB/s | 9.2 MB 00:00 +(30/31): rust-std-static-1.97.0-1.amzn2023.0.1. 69 MB/s | 39 MB 00:00 +(31/31): rust-1.97.0-1.amzn2023.0.1.x86_64.rpm 52 MB/s | 62 MB 00:01 +-------------------------------------------------------------------------------- +Total 102 MB/s | 189 MB 00:01 +Running transaction check +Transaction check succeeded. +Running transaction test +Transaction test succeeded. +Running transaction + Preparing : 1/1 + Installing : libmpc-1.2.1-2.amzn2023.0.2.x86_64 1/31 + Installing : git-core-2.50.1-1.amzn2023.0.1.x86_64 2/31 + Installing : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 3/31 + Installing : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Installing : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 5/31 + Installing : perl-lib-0.65-477.amzn2023.0.9.x86_64 6/31 + Installing : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 7/31 + Installing : perl-File-Find-1.37-477.amzn2023.0.9.noarch 8/31 + Installing : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 9/31 + Installing : perl-Git-2.50.1-1.amzn2023.0.1.noarch 10/31 + Installing : git-2.50.1-1.amzn2023.0.1.x86_64 11/31 + Installing : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 12/31 + Installing : python3.12-3.12.13-2.amzn2023.0.5.x86_64 13/31 + Installing : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 14/31 + Installing : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 15/31 + Installing : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Installing : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 17/31 + Installing : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 18/31 + Installing : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 19/31 + Installing : glibc-devel-2.34-231.amzn2023.0.5.x86_64 20/31 + Installing : gc-8.0.4-5.amzn2023.0.2.x86_64 21/31 + Installing : guile22-2.2.7-2.amzn2023.0.3.x86_64 22/31 + Installing : make-1:4.3-5.amzn2023.0.2.x86_64 23/31 + Installing : gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Running scriptlet: gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Installing : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 25/31 + Installing : rust-1.97.0-1.amzn2023.0.1.x86_64 26/31 + Installing : annobin-docs-12.69-1.amzn2023.0.1.noarch 27/31 + Installing : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Running scriptlet: annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Installing : cargo-1.97.0-1.amzn2023.0.1.x86_64 29/31 + Installing : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 30/31 + Installing : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Running scriptlet: python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Verifying : annobin-docs-12.69-1.amzn2023.0.1.noarch 1/31 + Verifying : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 2/31 + Verifying : cargo-1.97.0-1.amzn2023.0.1.x86_64 3/31 + Verifying : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Verifying : gc-8.0.4-5.amzn2023.0.2.x86_64 5/31 + Verifying : gcc-11.5.0-5.amzn2023.0.5.x86_64 6/31 + Verifying : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 7/31 + Verifying : git-2.50.1-1.amzn2023.0.1.x86_64 8/31 + Verifying : git-core-2.50.1-1.amzn2023.0.1.x86_64 9/31 + Verifying : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 10/31 + Verifying : glibc-devel-2.34-231.amzn2023.0.5.x86_64 11/31 + Verifying : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 12/31 + Verifying : guile22-2.2.7-2.amzn2023.0.3.x86_64 13/31 + Verifying : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 14/31 + Verifying : libmpc-1.2.1-2.amzn2023.0.2.x86_64 15/31 + Verifying : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Verifying : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 17/31 + Verifying : make-1:4.3-5.amzn2023.0.2.x86_64 18/31 + Verifying : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 19/31 + Verifying : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 20/31 + Verifying : perl-File-Find-1.37-477.amzn2023.0.9.noarch 21/31 + Verifying : perl-Git-2.50.1-1.amzn2023.0.1.noarch 22/31 + Verifying : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 23/31 + Verifying : perl-lib-0.65-477.amzn2023.0.9.x86_64 24/31 + Verifying : python3.12-3.12.13-2.amzn2023.0.5.x86_64 25/31 + Verifying : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 26/31 + Verifying : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 27/31 + Verifying : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 28/31 + Verifying : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 29/31 + Verifying : rust-1.97.0-1.amzn2023.0.1.x86_64 30/31 + Verifying : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 31/31 + +Installed: + annobin-docs-12.69-1.amzn2023.0.1.noarch + annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 + cargo-1.97.0-1.amzn2023.0.1.x86_64 + cpp-11.5.0-5.amzn2023.0.5.x86_64 + gc-8.0.4-5.amzn2023.0.2.x86_64 + gcc-11.5.0-5.amzn2023.0.5.x86_64 + gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 + git-2.50.1-1.amzn2023.0.1.x86_64 + git-core-2.50.1-1.amzn2023.0.1.x86_64 + git-core-doc-2.50.1-1.amzn2023.0.1.noarch + glibc-devel-2.34-231.amzn2023.0.5.x86_64 + glibc-headers-x86-2.34-231.amzn2023.0.5.noarch + guile22-2.2.7-2.amzn2023.0.3.x86_64 + kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 + libmpc-1.2.1-2.amzn2023.0.2.x86_64 + libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 + libxcrypt-devel-4.4.33-7.amzn2023.x86_64 + make-1:4.3-5.amzn2023.0.2.x86_64 + mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 + perl-Error-1:0.17030-2.amzn2023.0.1.noarch + perl-File-Find-1.37-477.amzn2023.0.9.noarch + perl-Git-2.50.1-1.amzn2023.0.1.noarch + perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 + perl-lib-0.65-477.amzn2023.0.9.x86_64 + python3.12-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-pip-23.2.1-4.amzn2023.0.10.noarch + python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch + python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch + rust-1.97.0-1.amzn2023.0.1.x86_64 + rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 + +Complete! +Cloning into '/opt/qwen35-acceptance/lecore'... +Updating files: 0% (2/903) Updating files: 1% (10/903) Updating files: 2% (19/903) Updating files: 3% (28/903) Updating files: 4% (37/903) Updating files: 5% (46/903) Updating files: 6% (55/903) Updating files: 7% (64/903) Updating files: 8% (73/903) Updating files: 9% (82/903) Updating files: 10% (91/903) Updating files: 11% (100/903) Updating files: 12% (109/903) Updating files: 13% (118/903) Updating files: 14% (127/903) Updating files: 15% (136/903) Updating files: 16% (145/903) Updating files: 17% (154/903) Updating files: 18% (163/903) Updating files: 19% (172/903) Updating files: 20% (181/903) Updating files: 21% (190/903) Updating files: 22% (199/903) Updating files: 23% (208/903) Updating files: 24% (217/903) Updating files: 25% (226/903) Updating files: 26% (235/903) Updating files: 27% (244/903) Updating files: 28% (253/903) Updating files: 29% (262/903) Updating files: 30% (271/903) Updating files: 31% (280/903) Updating files: 32% (289/903) Updating files: 33% (298/903) Updating files: 34% (308/903) Updating files: 35% (317/903) Updating files: 36% (326/903) Updating files: 37% (335/903) Updating files: 38% (344/903) Updating files: 39% (353/903) Updating files: 40% (362/903) Updating files: 41% (371/903) Updating files: 42% (380/903) Updating files: 43% (389/903) Updating files: 44% (398/903) Updating files: 45% (407/903) Updating files: 46% (416/903) Updating files: 47% (425/903) Updating files: 48% (434/903) Updating files: 49% (443/903) Updating files: 50% (452/903) Updating files: 51% (461/903) Updating files: 52% (470/903) Updating files: 53% (479/903) Updating files: 54% (488/903) Updating files: 55% (497/903) Updating files: 56% (506/903) Updating files: 57% (515/903) Updating files: 58% (524/903) Updating files: 59% (533/903) Updating files: 60% (542/903) Updating files: 61% (551/903) Updating files: 62% (560/903) Updating files: 63% (569/903) Updating files: 64% (578/903) Updating files: 65% (587/903) Updating files: 66% (596/903) Updating files: 67% (606/903) Updating files: 68% (615/903) Updating files: 69% (624/903) Updating files: 70% (633/903) Updating files: 71% (642/903) Updating files: 72% (651/903) Updating files: 73% (660/903) Updating files: 74% (669/903) Updating files: 75% (678/903) Updating files: 76% (687/903) Updating files: 77% (696/903) Updating files: 78% (705/903) Updating files: 79% (714/903) Updating files: 80% (723/903) Updating files: 81% (732/903) Updating files: 82% (741/903) Updating files: 83% (750/903) Updating files: 84% (759/903) Updating files: 85% (768/903) Updating files: 86% (777/903) Updating files: 87% (786/903) Updating files: 88% (795/903) Updating files: 89% (804/903) Updating files: 90% (813/903) Updating files: 91% (822/903) Updating files: 92% (831/903) Updating files: 93% (840/903) Updating files: 94% (849/903) Updating files: 95% (858/903) Updating files: 96% (867/903) Updating files: 97% (876/903) Updating files: 98% (885/903) Updating files: 99% (894/903) Updating files: 100% (903/903) Updating files: 100% (903/903), done. +HEAD is now at a04ab56 fix: enforce qwen pre-install controls +/opt/qwen35-acceptance/lecore/REFERENCE.md: OK +/opt/qwen35-acceptance/inputs/federalist-papers.txt: OK +Cloning into '/opt/qwen35-acceptance/ilxyr'... +HEAD is now at e92382f Publish EXP-006 experiment evidence + Updating crates.io index + Downloading crates ... + Downloaded der v0.7.10 + Downloaded getrandom v0.2.17 + Downloaded digest v0.10.7 + Downloaded zeroize v1.9.0 + Downloaded cfg-if v1.0.4 + Downloaded serde_core v1.0.229 + Downloaded spki v0.7.3 + Downloaded sha2 v0.10.9 + Downloaded block-buffer v0.10.4 + Downloaded cpufeatures v0.2.17 + Downloaded crypto-common v0.1.7 + Downloaded signature v2.2.0 + Downloaded version_check v0.9.5 + Downloaded rustc_version v0.4.1 + Downloaded ed25519 v2.2.3 + Downloaded pkcs8 v0.10.2 + Downloaded quote v1.0.46 + Downloaded curve25519-dalek-derive v0.1.1 + Downloaded itoa v1.0.18 + Downloaded generic-array v0.14.7 + Downloaded subtle v2.6.1 + Downloaded zmij v1.0.23 + Downloaded rand_core v0.6.4 + Downloaded const-oid v0.9.6 + Downloaded memchr v2.8.3 + Downloaded base64ct v1.8.3 + Downloaded semver v1.0.28 + Downloaded serde_derive v1.0.229 + Downloaded unicode-ident v1.0.24 + Downloaded proc-macro2 v1.0.106 + Downloaded base64 v0.22.1 + Downloaded serde v1.0.229 + Downloaded typenum v1.20.1 + Downloaded ed25519-dalek v2.2.0 + Downloaded serde_json v1.0.150 + Downloaded syn v2.0.119 + Downloaded curve25519-dalek v4.1.3 + Downloaded syn v3.0.0 + Downloaded libc v0.2.186 + Compiling proc-macro2 v1.0.106 + Compiling version_check v0.9.5 + Compiling unicode-ident v1.0.24 + Compiling quote v1.0.46 + Compiling typenum v1.20.1 + Compiling semver v1.0.28 + Compiling serde_core v1.0.229 + Compiling cpufeatures v0.2.17 + Compiling zmij v1.0.23 + Compiling cfg-if v1.0.4 + Compiling serde_json v1.0.150 + Compiling subtle v2.6.1 + Compiling serde v1.0.229 + Compiling signature v2.2.0 + Compiling zeroize v1.9.0 + Compiling memchr v2.8.3 + Compiling itoa v1.0.18 + Compiling base64 v0.22.1 + Compiling ed25519 v2.2.3 + Compiling rustc_version v0.4.1 + Compiling generic-array v0.14.7 + Compiling curve25519-dalek v4.1.3 + Compiling syn v2.0.119 + Compiling syn v3.0.0 + Compiling block-buffer v0.10.4 + Compiling crypto-common v0.1.7 + Compiling digest v0.10.7 + Compiling sha2 v0.10.9 + Compiling serde_derive v1.0.229 + Compiling curve25519-dalek-derive v0.1.1 + Compiling ed25519-dalek v2.2.0 + Compiling ilxyr-core v0.1.0 (/opt/qwen35-acceptance/ilxyr/crates/ilxyr-core) + Compiling ilxyr-cli v0.1.0 (/opt/qwen35-acceptance/ilxyr/crates/ilxyr-cli) + Finished `release` profile [optimized] target(s) in 21.51s +Looking in indexes: https://download.pytorch.org/whl/cpu +Collecting torch==2.11.0 + Obtaining dependency information for torch==2.11.0 from https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata + Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata (29 kB) +Collecting filelock (from torch==2.11.0) + Obtaining dependency information for filelock from https://files.pythonhosted.org/packages/81/47/dd9a212ef6e343a6857485ffe25bba537304f1913bdbed446a23f7f592e1/filelock-3.29.0-py3-none-any.whl.metadata + Downloading filelock-3.29.0-py3-none-any.whl.metadata (2.0 kB) +Collecting typing-extensions>=4.10.0 (from torch==2.11.0) + Obtaining dependency information for typing-extensions>=4.10.0 from https://files.pythonhosted.org/packages/18/67/36e9267722cc04a6b9f15c7f3441c2363321a3ea07da7ae0c0707beb2a9c/typing_extensions-4.15.0-py3-none-any.whl.metadata + Downloading typing_extensions-4.15.0-py3-none-any.whl.metadata (3.3 kB) +Collecting setuptools<82 (from torch==2.11.0) + Obtaining dependency information for setuptools<82 from https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata (6.6 kB) +Collecting sympy>=1.13.3 (from torch==2.11.0) + Obtaining dependency information for sympy>=1.13.3 from https://files.pythonhosted.org/packages/a2/09/77d55d46fd61b4a135c444fc97158ef34a095e5681d0a6c10b75bf356191/sympy-1.14.0-py3-none-any.whl.metadata + Downloading sympy-1.14.0-py3-none-any.whl.metadata (12 kB) +Collecting networkx>=2.5.1 (from torch==2.11.0) + Obtaining dependency information for networkx>=2.5.1 from https://files.pythonhosted.org/packages/9e/c9/b2622292ea83fbb4ec318f5b9ab867d0a28ab43c5717bb85b0a5f6b3b0a4/networkx-3.6.1-py3-none-any.whl.metadata + Downloading networkx-3.6.1-py3-none-any.whl.metadata (6.8 kB) +Collecting jinja2 (from torch==2.11.0) + Obtaining dependency information for jinja2 from https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata (2.9 kB) +Collecting fsspec>=0.8.5 (from torch==2.11.0) + Obtaining dependency information for fsspec>=0.8.5 from https://files.pythonhosted.org/packages/d5/0c/043d5e551459da400957a1395e0febbf771446ff34291afcbe3d8be2a279/fsspec-2026.4.0-py3-none-any.whl.metadata + Downloading fsspec-2026.4.0-py3-none-any.whl.metadata (10 kB) +Collecting mpmath<1.4,>=1.1.0 (from sympy>=1.13.3->torch==2.11.0) + Obtaining dependency information for mpmath<1.4,>=1.1.0 from https://files.pythonhosted.org/packages/43/e3/7d92a15f894aa0c9c4b49b8ee9ac9850d6e63b03c9c32c0367a13ae62209/mpmath-1.3.0-py3-none-any.whl.metadata + Downloading mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) +Collecting MarkupSafe>=2.0 (from jinja2->torch==2.11.0) + Obtaining dependency information for MarkupSafe>=2.0 from https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.7 kB) +Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl (190.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 190.3/190.3 MB 13.8 MB/s eta 0:00:00 +Downloading fsspec-2026.4.0-py3-none-any.whl (203 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 203.4/203.4 kB 24.0 MB/s eta 0:00:00 +Downloading networkx-3.6.1-py3-none-any.whl (2.1 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 2.1/2.1 MB 159.2 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl (1.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.3/1.3 MB 188.0 MB/s eta 0:00:00 +Downloading sympy-1.14.0-py3-none-any.whl (6.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.3/6.3 MB 192.4 MB/s eta 0:00:00 +Downloading typing_extensions-4.15.0-py3-none-any.whl (44 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 44.6/44.6 kB 17.0 MB/s eta 0:00:00 +Downloading filelock-3.29.0-py3-none-any.whl (39 kB) +Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl (134 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 134.9/134.9 kB 48.2 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (22 kB) +Downloading mpmath-1.3.0-py3-none-any.whl (536 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 536.2/536.2 kB 133.6 MB/s eta 0:00:00 +Installing collected packages: mpmath, typing-extensions, sympy, setuptools, networkx, MarkupSafe, fsspec, filelock, jinja2, torch +Successfully installed MarkupSafe-3.0.3 filelock-3.29.0 fsspec-2026.4.0 jinja2-3.1.6 mpmath-1.3.0 networkx-3.6.1 setuptools-78.1.0 sympy-1.14.0 torch-2.11.0+cpu typing-extensions-4.15.0 +Collecting numpy==2.4.6 + Obtaining dependency information for numpy==2.4.6 from https://files.pythonhosted.org/packages/99/60/14115e6364fa676c5397c2ad3004e527e9aa487abf5d0706ec81bbd08529/numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (6.6 kB) +Collecting pillow==12.3.0 + Obtaining dependency information for pillow==12.3.0 from https://files.pythonhosted.org/packages/84/21/a35af28dcc61f37ed850a2d64c65c701321dfbf25085e469d5559360cbbf/pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (9.1 kB) +Collecting safetensors==0.8.0 + Obtaining dependency information for safetensors==0.8.0 from https://files.pythonhosted.org/packages/28/50/f203ff3a3ddfe19308efc83c5a3a29ed02bf786732ec35e68bf9162f3365/safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (4.2 kB) +Collecting huggingface-hub==1.27.0 + Obtaining dependency information for huggingface-hub==1.27.0 from https://files.pythonhosted.org/packages/de/d8/95b735e183957c1f26d94c52977f09d466d55119cbbc1558ea4975e4c216/huggingface_hub-1.27.0-py3-none-any.whl.metadata + Downloading huggingface_hub-1.27.0-py3-none-any.whl.metadata (16 kB) +Collecting transformers==5.14.0 + Obtaining dependency information for transformers==5.14.0 from https://files.pythonhosted.org/packages/19/3b/a0a807b957c5f8f2075a8740f597a021420583bb0aa68ff6432d74a530d4/transformers-5.14.0-py3-none-any.whl.metadata + Downloading transformers-5.14.0-py3-none-any.whl.metadata (32 kB) +Collecting click<9.0.0,>=8.4.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for click<9.0.0,>=8.4.2 from https://files.pythonhosted.org/packages/fb/e2/79c688af8b210d232694e31e59da9f6ec747bae31c3f5946e4e9b98860d5/click-8.4.2-py3-none-any.whl.metadata + Downloading click-8.4.2-py3-none-any.whl.metadata (2.6 kB) +Requirement already satisfied: filelock>=3.10.0 in /opt/qwen35-acceptance/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (3.29.0) +Requirement already satisfied: fsspec>=2023.5.0 in /opt/qwen35-acceptance/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (2026.4.0) +Collecting hf-xet<2.0.0,>=1.5.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for hf-xet<2.0.0,>=1.5.2 from https://files.pythonhosted.org/packages/67/4e/a28359bf1c1ecf11eba22123168c138698f7cb576ac678f5a2e16cd5da08/hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata + Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (4.9 kB) +Collecting httpx<1,>=0.23.0 (from huggingface-hub==1.27.0) + Obtaining dependency information for httpx<1,>=0.23.0 from https://files.pythonhosted.org/packages/2a/39/e50c7c3a983047577ee07d2a9e53faf5a69493943ec3f6a384bdc792deb2/httpx-0.28.1-py3-none-any.whl.metadata + Downloading httpx-0.28.1-py3-none-any.whl.metadata (7.1 kB) +Collecting packaging>=20.9 (from huggingface-hub==1.27.0) + Obtaining dependency information for packaging>=20.9 from https://files.pythonhosted.org/packages/63/34/ba1c580383c9eada3711951fef0795c80b829a078d72188184bcab9dd527/packaging-26.3-py3-none-any.whl.metadata + Downloading packaging-26.3-py3-none-any.whl.metadata (3.5 kB) +Collecting pyyaml>=5.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for pyyaml>=5.1 from https://files.pythonhosted.org/packages/8b/9d/b3589d3877982d4f2329302ef98a8026e7f4443c765c46cfecc8858c6b4b/pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.4 kB) +Collecting tqdm>=4.42.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for tqdm>=4.42.1 from https://files.pythonhosted.org/packages/f9/1c/01bfd571a64e7f270e6bab5e33777debe0edc56759233ce84f27dec92d14/tqdm-4.70.0-py3-none-any.whl.metadata + Downloading tqdm-4.70.0-py3-none-any.whl.metadata (57 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 57.3/57.3 kB 13.8 MB/s eta 0:00:00 +Requirement already satisfied: typing-extensions>=4.1.0 in /opt/qwen35-acceptance/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (4.15.0) +Collecting regex>=2025.10.22 (from transformers==5.14.0) + Obtaining dependency information for regex>=2025.10.22 from https://files.pythonhosted.org/packages/2a/8e/096d00c7c480ef2ff4265349b14e2261d4ab787ba1f74e2e80d1c58079c3/regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (40 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 40.5/40.5 kB 12.7 MB/s eta 0:00:00 +Collecting tokenizers<=0.23.0,>=0.22.0 (from transformers==5.14.0) + Obtaining dependency information for tokenizers<=0.23.0,>=0.22.0 from https://files.pythonhosted.org/packages/2e/76/932be4b50ef6ccedf9d3c6639b056a967a86258c6d9200643f01269211ca/tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (7.3 kB) +Collecting typer (from transformers==5.14.0) + Obtaining dependency information for typer from https://files.pythonhosted.org/packages/43/89/9518bc0c3929bee36b3a4a8e3daddd6e03f92f9961c66d4983b837160543/typer-0.27.1-py3-none-any.whl.metadata + Downloading typer-0.27.1-py3-none-any.whl.metadata (16 kB) +Collecting anyio (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for anyio from https://files.pythonhosted.org/packages/da/35/f2287558c17e29fafc8ef3daf819bb9834061cfa43bff8014f7df7f63bdc/anyio-4.14.2-py3-none-any.whl.metadata + Downloading anyio-4.14.2-py3-none-any.whl.metadata (4.6 kB) +Collecting certifi (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for certifi from https://files.pythonhosted.org/packages/0b/a7/71ac2cff56fec219ed242bb11b8efb69fcc4bec75db06fb7bfe35de520e6/certifi-2026.7.22-py3-none-any.whl.metadata + Downloading certifi-2026.7.22-py3-none-any.whl.metadata (2.5 kB) +Collecting httpcore==1.* (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for httpcore==1.* from https://files.pythonhosted.org/packages/7e/f5/f66802a942d491edb555dd61e3a9961140fd64c90bce1eafd741609d334d/httpcore-1.0.9-py3-none-any.whl.metadata + Downloading httpcore-1.0.9-py3-none-any.whl.metadata (21 kB) +Collecting idna (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for idna from https://files.pythonhosted.org/packages/1e/5e/d4e9f1a599fb8e573b7b87160658329fbf28d19eac2718f51fc3def3aa5a/idna-3.18-py3-none-any.whl.metadata + Downloading idna-3.18-py3-none-any.whl.metadata (6.1 kB) +Collecting h11>=0.16 (from httpcore==1.*->httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for h11>=0.16 from https://files.pythonhosted.org/packages/04/4b/29cac41a4d98d144bf5f6d33995617b185d14b22401f75ca86f384e87ff1/h11-0.16.0-py3-none-any.whl.metadata + Downloading h11-0.16.0-py3-none-any.whl.metadata (8.3 kB) +Collecting shellingham>=1.3.0 (from typer->transformers==5.14.0) + Obtaining dependency information for shellingham>=1.3.0 from https://files.pythonhosted.org/packages/e0/f9/0595336914c5619e5f28a1fb793285925a8cd4b432c9da0a987836c7f822/shellingham-1.5.4-py2.py3-none-any.whl.metadata + Downloading shellingham-1.5.4-py2.py3-none-any.whl.metadata (3.5 kB) +Collecting rich>=13.8.0 (from typer->transformers==5.14.0) + Obtaining dependency information for rich>=13.8.0 from https://files.pythonhosted.org/packages/82/3b/64d4899d73f91ba49a8c18a8ff3f0ea8f1c1d75481760df8c68ef5235bf5/rich-15.0.0-py3-none-any.whl.metadata + Downloading rich-15.0.0-py3-none-any.whl.metadata (18 kB) +Collecting annotated-doc>=0.0.2 (from typer->transformers==5.14.0) + Obtaining dependency information for annotated-doc>=0.0.2 from https://files.pythonhosted.org/packages/3e/30/e900b21425a860e195f32e37657aa1f7c7f2b1bfb26f03ca209b90933c06/annotated_doc-0.0.5-py3-none-any.whl.metadata + Downloading annotated_doc-0.0.5-py3-none-any.whl.metadata (6.5 kB) +Collecting markdown-it-py>=2.2.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for markdown-it-py>=2.2.0 from https://files.pythonhosted.org/packages/b3/81/4da04ced5a082363ecfa159c010d200ecbd959ae410c10c0264a38cac0f5/markdown_it_py-4.2.0-py3-none-any.whl.metadata + Downloading markdown_it_py-4.2.0-py3-none-any.whl.metadata (7.4 kB) +Collecting pygments<3.0.0,>=2.13.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for pygments<3.0.0,>=2.13.0 from https://files.pythonhosted.org/packages/f4/7e/a72dd26f3b0f4f2bf1dd8923c85f7ceb43172af56d63c7383eb62b332364/pygments-2.20.0-py3-none-any.whl.metadata + Downloading pygments-2.20.0-py3-none-any.whl.metadata (2.5 kB) +Collecting mdurl~=0.1 (from markdown-it-py>=2.2.0->rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for mdurl~=0.1 from https://files.pythonhosted.org/packages/b3/38/89ba8ad64ae25be8de66a6d463314cf1eb366222074cfda9ee839c56a4b4/mdurl-0.1.2-py3-none-any.whl.metadata + Downloading mdurl-0.1.2-py3-none-any.whl.metadata (1.6 kB) +Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (16.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 16.6/16.6 MB 100.4 MB/s eta 0:00:00 +Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.9 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.9/6.9 MB 156.3 MB/s eta 0:00:00 +Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (516 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 516.0/516.0 kB 94.8 MB/s eta 0:00:00 +Downloading huggingface_hub-1.27.0-py3-none-any.whl (784 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 784.9/784.9 kB 118.7 MB/s eta 0:00:00 +Downloading transformers-5.14.0-py3-none-any.whl (11.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 11.6/11.6 MB 104.4 MB/s eta 0:00:00 +Downloading click-8.4.2-py3-none-any.whl (119 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 119.2/119.2 kB 29.2 MB/s eta 0:00:00 +Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (4.5 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 4.5/4.5 MB 222.0 MB/s eta 0:00:00 +Downloading httpx-0.28.1-py3-none-any.whl (73 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 73.5/73.5 kB 25.9 MB/s eta 0:00:00 +Downloading httpcore-1.0.9-py3-none-any.whl (78 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 78.8/78.8 kB 19.8 MB/s eta 0:00:00 +Downloading packaging-26.3-py3-none-any.whl (129 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 130.0/130.0 kB 27.7 MB/s eta 0:00:00 +Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (807 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 807.9/807.9 kB 120.9 MB/s eta 0:00:00 +Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (801 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 801.8/801.8 kB 118.4 MB/s eta 0:00:00 +Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (3.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 3.3/3.3 MB 202.6 MB/s eta 0:00:00 +Downloading tqdm-4.70.0-py3-none-any.whl (80 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 80.2/80.2 kB 29.4 MB/s eta 0:00:00 +Downloading typer-0.27.1-py3-none-any.whl (122 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 122.9/122.9 kB 38.6 MB/s eta 0:00:00 +Downloading annotated_doc-0.0.5-py3-none-any.whl (5.3 kB) +Downloading rich-15.0.0-py3-none-any.whl (310 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 310.7/310.7 kB 65.7 MB/s eta 0:00:00 +Downloading shellingham-1.5.4-py2.py3-none-any.whl (9.8 kB) +Downloading anyio-4.14.2-py3-none-any.whl (125 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 125.8/125.8 kB 45.2 MB/s eta 0:00:00 +Downloading idna-3.18-py3-none-any.whl (65 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 65.5/65.5 kB 23.7 MB/s eta 0:00:00 +Downloading certifi-2026.7.22-py3-none-any.whl (136 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 137.0/137.0 kB 41.6 MB/s eta 0:00:00 +Downloading h11-0.16.0-py3-none-any.whl (37 kB) +Downloading markdown_it_py-4.2.0-py3-none-any.whl (91 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 91.7/91.7 kB 33.1 MB/s eta 0:00:00 +Downloading pygments-2.20.0-py3-none-any.whl (1.2 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.2/1.2 MB 169.4 MB/s eta 0:00:00 +Downloading mdurl-0.1.2-py3-none-any.whl (10.0 kB) +Installing collected packages: tqdm, shellingham, safetensors, regex, pyyaml, pygments, pillow, packaging, numpy, mdurl, idna, hf-xet, h11, click, certifi, annotated-doc, markdown-it-py, httpcore, anyio, rich, httpx, typer, huggingface-hub, tokenizers, transformers +Successfully installed annotated-doc-0.0.5 anyio-4.14.2 certifi-2026.7.22 click-8.4.2 h11-0.16.0 hf-xet-1.6.0 httpcore-1.0.9 httpx-0.28.1 huggingface-hub-1.27.0 idna-3.18 markdown-it-py-4.2.0 mdurl-0.1.2 numpy-2.4.6 packaging-26.3 pillow-12.3.0 pygments-2.20.0 pyyaml-6.0.3 regex-2026.7.19 rich-15.0.0 safetensors-0.8.0 shellingham-1.5.4 tokenizers-0.22.2 tqdm-4.70.0 transformers-5.14.0 typer-0.27.1 + +[notice] A new release of pip is available: 23.2.1 -> 26.2.1 +[notice] To update, run: python3.12 -m pip install --upgrade pip +No broken requirements found. +torch 2.11.0+cpu +transformers 5.14.0 +AutoProcessor Qwen3_5ForConditionalGeneration + Fetching 13 files: 0%| | 0/13 [00:00 Date: Wed, 12 Aug 2026 15:11:49 -0700 Subject: [PATCH 06/17] docs: record arweave download verification --- .../publication-receipt.json | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-receipt.json b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-receipt.json index 3c0b77e..ddb4787 100644 --- a/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-receipt.json +++ b/experiments/qwen35_acceptance/results/v1-20c3330d0b3e-execution-failure/publication-receipt.json @@ -40,7 +40,12 @@ "confirmation": { "status_service": "https://upload.ardrive.io/v1/tx/{transaction_id}/status", "transactions_confirmed": 63, - "total_winc_charged": 0 + "total_winc_charged": 0, + "public_gateway_verified_at": "2026-08-12T22:09:00Z", + "verified_downloads": { + "publication-manifest.json": "80ebc69fc40c4bbab9e3692a1713217fba007d1005adf2f23be79be739d1b836", + "ledger/events.jsonl": "0254b4dc60bf2903cf308104bf3f8c34d64fb050fb1e4e014590fe0c1847dc35" + } }, "superseded_path_manifest": { "transaction_id": "lPXKEJON5Xa4wXswwRwFv2GWqrTPo7yuOhMtVKyOpVI", From 94eb569aa62cce4dae3d10a868b0f22bb04d962e Mon Sep 17 00:00:00 2001 From: docs-bot Date: Thu, 13 Aug 2026 00:04:47 +0000 Subject: [PATCH 07/17] docs: refresh generated docs (REFERENCE, CAPABILITIES, API_QUICKREF, PIPELINE_MAP, FACULTY_MAP, DOC_MAP) [skip ci] --- REFERENCE.md | 278 +++++++++++++++++++++++++-------------------------- 1 file changed, 139 insertions(+), 139 deletions(-) diff --git a/REFERENCE.md b/REFERENCE.md index e06473d..164a4cd 100644 --- a/REFERENCE.md +++ b/REFERENCE.md @@ -76,8 +76,8 @@ | module | what it is | lines | |---|---|---| | [`holographic_session.py`](#holographic-session) | holographic_session.py -- ONE render session that ties the disconnected rendering threads together. | 227 | -| [`holographic_session.py`](#holographic-session) | SESSION -- never compute the same conversation prefix twice. | 253 | | [`holographic_session.py`](#holographic-session) | SESSION STORE -- contexts that outlive the process. | 261 | +| [`holographic_session.py`](#holographic-session) | SESSION -- never compute the same conversation prefix twice. | 253 | ### `splat*` family (6) @@ -423,8 +423,8 @@ | [`holographic_materialio.py`](#holographic-materialio) | PBR material interchange (G2-IO): the standard-format bridge for materials, and a VSA-native carrier. | 307 | | [`holographic_matlib.py`](#holographic-matlib) | holographic_matlib.py -- a comprehensive RENDER material library: plain diffuse -> a fractal planet. | 801 | | [`holographic_meaning_predict.py`](#holographic-meaning-predict) | Generation with structure: predict a next-MEANING vector and settle it, rather | 198 | -| [`holographic_measure.py`](#holographic-measure) | The variance harness: every headline number gets a mean, a spread, and a confidence | 187 | | [`holographic_measure.py`](#holographic-measure) | MEASURE -- perplexity with error bars, and decisions that respect them. | 239 | +| [`holographic_measure.py`](#holographic-measure) | The variance harness: every headline number gets a mean, a spread, and a confidence | 187 | | [`holographic_memory.py`](#holographic-memory) | MEMORY -- the Galvatron's own store, built on leCore's holographic database. | 241 | | [`holographic_memoryhome.py`](#holographic-memoryhome) | holographic_memoryhome.py -- the MEMORY home (consolidation backlog H6): keep the hot working set where the CP | 165 | | [`holographic_memsearch.py`](#holographic-memsearch) | MEMSEARCH -- searchable memory that lives in the weights and runs in the pass. | 212 | @@ -577,8 +577,8 @@ | [`holographic_rolling.py`](#holographic-rolling) | holographic_rolling.py -- the CAUSAL rolling / streaming statistics kit: trailing mean, std, min, max, | 378 | | [`holographic_roomacoustic.py`](#holographic-roomacoustic) | holographic_roomacoustic.py -- A6: GEOMETRIC ROOM ACOUSTICS. How a room echoes -- reflections and reverb. | 161 | | [`holographic_route.py`](#holographic-route) | Representation routing (ARCH-7): route each operation to the representation that supports it -- the policy lay | 183 | -| [`holographic_router.py`](#holographic-router) | ROUTER -- the model DECIDING, inside one forward pass. | 229 | | [`holographic_router.py`](#holographic-router) | Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. | 173 | +| [`holographic_router.py`](#holographic-router) | ROUTER -- the model DECIDING, inside one forward pass. | 229 | | [`holographic_sampler.py`](#holographic-sampler) | holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). | 223 | | [`holographic_sampling.py`](#holographic-sampling) | Blue-noise / Poisson-disk point sampling -- the EXCLUSION principle, done right. | 134 | | [`holographic_samplinghome.py`](#holographic-samplinghome) | holographic_samplinghome.py -- the SAMPLING home (consolidation backlog R4): one place for the Monte-Carlo | 127 | @@ -683,8 +683,8 @@ | [`holographic_toolclient.py`](#holographic-toolclient) | holographic_toolclient.py -- call another node the same way leCore is called. | 98 | | [`holographic_topology.py`](#holographic-topology) | Principled topology by persistent homology (EXP-7). | 288 | | [`holographic_transfer.py`](#holographic-transfer) | holographic_transfer.py -- KERNEL SCATTER / GATHER: the ONE bundle/readout under every particle<->grid transfe | 224 | -| [`holographic_transform.py`](#holographic-transform) | holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). | 297 | | [`holographic_transform.py`](#holographic-transform) | TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. | 189 | +| [`holographic_transform.py`](#holographic-transform) | holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). | 297 | | [`holographic_transform_space.py`](#holographic-transform-space) | holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it | 176 | | [`holographic_transformbank.py`](#holographic-transformbank) | holographic_transformbank.py -- a prebuilt map of hypervector transforms, and what it can and cannot hold. | 358 | | [`holographic_transformhome.py`](#holographic-transformhome) | holographic_transformhome.py -- the TRANSFORM home (consolidation backlog H5): one facade over "move / rotate | 223 | @@ -14848,39 +14848,6 @@ ### holographic_measure.py -> The variance harness: every headline number gets a mean, a spread, and a confidence -> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. -> -> WHY this exists -> --------------- -> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's -> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A -> single-seed score is therefore a sample from a distribution, and reporting it alone hides -> how wide that distribution is. For an engine whose entire pitch is "measured, not -> promised," reporting a number without its noise is the sharpest blind spot -- so this -> points that same discipline at the numbers themselves. -> -> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the -> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no -> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI -> bound clears the floor -- which is what stops a single fortunate seed from passing a -> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". -> -> USE REAL DATA. The point of the harness is to characterise the real distribution of a -> real claim; running it on a toy makes the spread meaningless. The measurements wired -> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). - -**Public API:** - -- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. -- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. -- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. -- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is -- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above -- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a - -### holographic_measure.py - > MEASURE -- perplexity with error bars, and decisions that respect them. > > Moose asked what assimilation is actually doing. The answer, from his own run: @@ -14913,6 +14880,39 @@ - `def better_than(a, b, alpha, seed, resamples)` -- Is model A better than model B, or is the difference undecidable? - `def tokens_needed(reference, effect_pct, alpha)` -- How many tokens would be needed to RESOLVE an effect of this size. +### holographic_measure.py + +> The variance harness: every headline number gets a mean, a spread, and a confidence +> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. +> +> WHY this exists +> --------------- +> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's +> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A +> single-seed score is therefore a sample from a distribution, and reporting it alone hides +> how wide that distribution is. For an engine whose entire pitch is "measured, not +> promised," reporting a number without its noise is the sharpest blind spot -- so this +> points that same discipline at the numbers themselves. +> +> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the +> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no +> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI +> bound clears the floor -- which is what stops a single fortunate seed from passing a +> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". +> +> USE REAL DATA. The point of the harness is to characterise the real distribution of a +> real claim; running it on a toy makes the spread meaningless. The measurements wired +> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). + +**Public API:** + +- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. +- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. +- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. +- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is +- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above +- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a + ### holographic_memory.py > MEMORY -- the Galvatron's own store, built on leCore's holographic database. @@ -22656,6 +22656,33 @@ ### holographic_router.py +> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. +> +> WHY THIS EXISTS (measured, backlog N9/N28): +> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, +> but it has no notion of MEANING: "squish a big array down for storage" shares no token with +> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics +> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median +> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. +> +> WHAT SHIPS (and what does NOT): +> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted +> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a +> query it already has a VECTOR for: +> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) +> * a caller that supplies its own query vector (an app that ran the encoder itself) +> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this +> router says so and returns None, so the caller falls back to the token router rather than guessing. That +> honesty is the whole point: silence beats a confident wrong route. +> +> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. + +**Public API:** + +- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. + +### holographic_router.py + > ROUTER -- the model DECIDING, inside one forward pass. > > Moose raised the architecture that dissolves the wall this project kept hitting: @@ -22695,33 +22722,6 @@ - `def route(runtime, router, text, tokenize)` -- Would this model choose to use the capability on this prompt? - `def install_routed(weights, cfg, operator, router, layer, gain, temperature)` -- Install a circuit whose GATE is the router, not a constant. -### holographic_router.py - -> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. -> -> WHY THIS EXISTS (measured, backlog N9/N28): -> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, -> but it has no notion of MEANING: "squish a big array down for storage" shares no token with -> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics -> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median -> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. -> -> WHAT SHIPS (and what does NOT): -> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted -> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a -> query it already has a VECTOR for: -> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) -> * a caller that supplies its own query vector (an app that ran the encoder itself) -> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this -> router says so and returns None, so the caller falls back to the token router rather than guessing. That -> honesty is the whole point: silence beats a confident wrong route. -> -> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. - -**Public API:** - -- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. - ### holographic_sampler.py > holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). @@ -24352,39 +24352,6 @@ ### holographic_session.py -> SESSION -- never compute the same conversation prefix twice. -> -> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a -> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS -> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens -> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS -> REPEATED, and the fraction grows with every turn. -> -> leCore already had the pieces and never joined them: the runtime exposes -> prefill/step over an InferenceState with copy(), galvacache memoises pure -> functions at a measured 75% hit rate with bit-identical output, and sessions -> exist. What was missing is the RADIX TREE -- the structure that answers "what is -> the longest prefix of this prompt that I have already computed?" -> -> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three -> turns with turn 3, and a dictionary keyed on the whole prompt misses that -> completely. vLLM and SGLang call this RadixAttention; the idea is the same -> whatever the model: index by prefix, resume from the deepest match, compute only -> the tail. -> -> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no -> cache: resuming from a cached state reproduces a full recompute TO FLOAT -> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the -> difference is real rather than pedantic: resuming STEPS the tail while a fresh -> call PREFILLS it, and the two associate their sums differently. The selftest -> asserts the measured bound against the live runtime rather than assuming it. - -**Public API:** - -- `class PrefixCache` -- A radix tree over token sequences, holding inference states. - -### holographic_session.py - > SESSION STORE -- contexts that outlive the process. > > A Galvatron's context is not a transcript, it is STATE: the GDN recurrent @@ -24417,6 +24384,39 @@ - `class SessionStore` -- Named, persistent, independent contexts on disk. - `def runtime_fingerprint(runtime)` -- A cheap, deterministic id for the checkpoint behind a runtime, so a +### holographic_session.py + +> SESSION -- never compute the same conversation prefix twice. +> +> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a +> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS +> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens +> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS +> REPEATED, and the fraction grows with every turn. +> +> leCore already had the pieces and never joined them: the runtime exposes +> prefill/step over an InferenceState with copy(), galvacache memoises pure +> functions at a measured 75% hit rate with bit-identical output, and sessions +> exist. What was missing is the RADIX TREE -- the structure that answers "what is +> the longest prefix of this prompt that I have already computed?" +> +> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three +> turns with turn 3, and a dictionary keyed on the whole prompt misses that +> completely. vLLM and SGLang call this RadixAttention; the idea is the same +> whatever the model: index by prefix, resume from the deepest match, compute only +> the tail. +> +> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no +> cache: resuming from a cached state reproduces a full recompute TO FLOAT +> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the +> difference is real rather than pedantic: resuming STEPS the tail while a fresh +> call PREFILLS it, and the two associate their sums differently. The selftest +> asserts the measured bound against the live runtime rather than assuming it. + +**Public API:** + +- `class PrefixCache` -- A radix tree over token sequences, holding inference states. + ### holographic_shader.py > holographic_shader.py -- N filter passes in ONE evaluation. Two things a GPU structurally cannot do. @@ -27945,48 +27945,6 @@ ### holographic_transform.py -> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). -> -> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's -> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need -> in one place. This gathers the standard, well-known math: -> -> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, -> and what a property panel shows as the T/R/S fields; -> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); -> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector -> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); -> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. -> -> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN -> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit -> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL -> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is -> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. -> Deterministic; NumPy + stdlib only. - -**Public API:** - -- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. -- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). -- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). -- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. -- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear -- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. -- `def quat_normalize(q)` -- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". -- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. -- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. -- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. -- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for -- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). -- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock -- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an -- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. -- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks - -### holographic_transform.py - > TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. > > Everything before this applied leCore's levers uniformly: grow a memory channel @@ -28029,6 +27987,48 @@ - `def plan(weights, cfg, target_tokens, kv_rank, grow_gain)` -- Decide what to do to each layer, from the analysis rather than by rule. - `def apply_plan(weights, cfg, the_plan, progress)` -- Carry out the growth actions. KV compression is a RUNTIME setting and is +### holographic_transform.py + +> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). +> +> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's +> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need +> in one place. This gathers the standard, well-known math: +> +> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, +> and what a property panel shows as the T/R/S fields; +> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); +> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector +> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); +> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. +> +> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN +> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit +> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL +> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is +> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. +> Deterministic; NumPy + stdlib only. + +**Public API:** + +- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. +- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). +- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). +- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. +- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear +- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. +- `def quat_normalize(q)` +- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". +- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. +- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. +- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. +- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for +- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). +- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock +- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an +- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. +- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks + ### holographic_transform_space.py > holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it From 3ed8d53839128c478196139a88407066570f571f Mon Sep 17 00:00:00 2001 From: atimics Date: Wed, 12 Aug 2026 17:13:41 -0700 Subject: [PATCH 08/17] experiments: authorize bounded qwen acceptance v2 --- experiments/qwen35_acceptance/README.md | 13 +++++++++---- experiments/qwen35_acceptance/launch-manifest.json | 14 +++++++++++--- 2 files changed, 20 insertions(+), 7 deletions(-) diff --git a/experiments/qwen35_acceptance/README.md b/experiments/qwen35_acceptance/README.md index 72549c1..a509b08 100644 --- a/experiments/qwen35_acceptance/README.md +++ b/experiments/qwen35_acceptance/README.md @@ -1,8 +1,12 @@ # Qwen3.5 acceptance experiment -The committee-facing design and open review questions are in -[`PROPOSAL.md`](PROPOSAL.md). The proposal must be approved, and its corpus -separation requirement implemented, before a formal real-model run. +The committee-facing design and review questions are in +[`PROPOSAL.md`](PROPOSAL.md). The first authorized run is preserved as an +execution failure: ilxyr invoked the system interpreter instead of the selected +virtual environment, so it produced no scientific result. The second and final +currently authorized attempt is frozen in [`launch-manifest.json`](launch-manifest.json). +It changes only the interpreter-path handling and the attempt identity; the +model revision, corpora, thresholds, seed, and spending ceiling remain fixed. This directory turns the open Qwen integration questions into a frozen ilxyr project for one owner-authorized, bounded attempt. It does not claim that the @@ -26,7 +30,8 @@ python experiments/qwen35_acceptance/generate.py \ /absolute/path/to/Qwen3.5-0.8B \ /absolute/path/to/installation-corpus.txt \ /absolute/path/to/evaluation-corpus.txt \ - /absolute/path/to/ilxyr-project + /absolute/path/to/ilxyr-project \ + --experiment-version 2 ``` The two corpora must have different contents. diff --git a/experiments/qwen35_acceptance/launch-manifest.json b/experiments/qwen35_acceptance/launch-manifest.json index d77227a..d5e6e2e 100644 --- a/experiments/qwen35_acceptance/launch-manifest.json +++ b/experiments/qwen35_acceptance/launch-manifest.json @@ -2,13 +2,21 @@ "schema": "lecore.qwen35-launch.v1", "state": "authorized_prelaunch", "attempt": { + "experiment_version": 2, "maximum_formal_runs": 1, "retry_after_admission": false, - "seed": 0 + "seed": 0, + "authorized_at": "2026-08-13T00:09:30Z", + "predecessor": { + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "outcome": "execution_failure", + "scientific_result": false, + "corrective_change": "preserve the selected virtual-environment interpreter path" + } }, "source": { "repository": "https://github.com/atimics/holostuff.git", - "commit": "a04ab5692be38f06120aba4b0bc5e2a284eb2c79" + "commit": "2f32151c6d44c80a7e2ff98c0696650cd1d9c6f1" }, "ilxyr": { "repository": "https://github.com/cenetex/ilXyr.git", @@ -89,7 +97,7 @@ "total_ceiling_usd": 10.0 }, "publication": { - "temporary_staging": "s3://zero-training-022118847419/qwen35-acceptance/a04ab56-2fc06364-a6c9d113/", + "temporary_staging": "s3://zero-training-022118847419/qwen35-acceptance/2f32151-2fc06364-a6c9d113/v2-formal/", "review_index": "https://github.com/atimics/holostuff/pull/3", "permanent_transport": "Arweave", "publish_all_outcomes": true, From cc97e6ee8cab5fafe515cb1273fa18523a97a88a Mon Sep 17 00:00:00 2001 From: atimics Date: Wed, 12 Aug 2026 22:05:22 -0700 Subject: [PATCH 09/17] results: publish qwen acceptance v2 evidence --- experiments/qwen35_acceptance/README.md | 24 +- .../qwen35_acceptance/launch-manifest.json | 23 +- .../README.md | 80 +++ .../evidence/evidence.in-toto.json | 395 +++++++++++++ .../evidence/evidence.native.json | 383 +++++++++++++ .../evidence/evidence.ro-crate.json | 286 ++++++++++ .../evidence/exporter-provenance.json | 23 + .../ledger/config.json | 5 + .../ledger/events.jsonl | 16 + .../project/engineering-review.json | 21 + .../project/experiment-design.json | 22 + .../project/experiment.json | 241 ++++++++ .../project/forecast-empirical.json | 17 + .../project/forecast-mechanistic.json | 17 + .../project/foundation.json | 20 + .../project/funding.json | 11 + .../project/hypothesis.json | 19 + .../project/project.json | 156 ++++++ .../publication-manifest.json | 284 ++++++++++ .../publication-receipt.json | 52 ++ .../result/acceptance-artifacts/install.log | 29 + .../result/admission-preflight.json | 23 + .../result/aws-instance-identity.json | 17 + .../result/cloud-init-output.log | 526 ++++++++++++++++++ .../result/corpus-manifest.json | 12 + .../result/environment.txt | 72 +++ .../result/execution-summary.json | 8 + .../result/generation.json | 1 + .../result/ilxyr-admission.json | 68 +++ .../result/ilxyr-commit.txt | 7 + .../result/ilxyr-compile.json | 3 + .../ilxyr-contribute-engineering-review.json | 3 + .../ilxyr-contribute-experiment-design.json | 3 + .../result/ilxyr-contribute-foundation.json | 3 + .../result/ilxyr-contribute-hypothesis.json | 3 + .../result/ilxyr-forecast-empirical.json | 3 + .../result/ilxyr-forecast-mechanistic.json | 3 + .../result/ilxyr-funding.json | 3 + .../result/ilxyr-init.json | 4 + .../result/ilxyr-run.json | 94 ++++ .../result/ilxyr-status.json | 121 ++++ .../result/ilxyr-verify.json | 5 + .../result/launch-manifest.json | 106 ++++ .../result/lecore-commit.txt | 7 + .../result/lecore-status.txt | 0 .../result/model-manifest.json | 66 +++ .../result/requirements-resolved.txt | 36 ++ .../result/runner-status.json | 7 + .../result/user-data.log | 463 +++++++++++++++ 49 files changed, 3782 insertions(+), 9 deletions(-) create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/README.md create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.in-toto.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.native.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.ro-crate.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/exporter-provenance.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/ledger/config.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/ledger/events.jsonl create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/engineering-review.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/experiment-design.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/experiment.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/forecast-empirical.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/forecast-mechanistic.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/foundation.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/funding.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/hypothesis.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/project.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/publication-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/publication-receipt.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/acceptance-artifacts/install.log create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/admission-preflight.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/aws-instance-identity.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/cloud-init-output.log create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/corpus-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/environment.txt create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/execution-summary.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/generation.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-admission.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-compile.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-funding.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-init.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-run.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-status.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-verify.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/launch-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/lecore-commit.txt create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/lecore-status.txt create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/model-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/requirements-resolved.txt create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/runner-status.json create mode 100644 experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/user-data.log diff --git a/experiments/qwen35_acceptance/README.md b/experiments/qwen35_acceptance/README.md index a509b08..118ad3a 100644 --- a/experiments/qwen35_acceptance/README.md +++ b/experiments/qwen35_acceptance/README.md @@ -1,16 +1,24 @@ # Qwen3.5 acceptance experiment The committee-facing design and review questions are in -[`PROPOSAL.md`](PROPOSAL.md). The first authorized run is preserved as an -execution failure: ilxyr invoked the system interpreter instead of the selected -virtual environment, so it produced no scientific result. The second and final -currently authorized attempt is frozen in [`launch-manifest.json`](launch-manifest.json). -It changes only the interpreter-path handling and the attempt identity; the -model revision, corpora, thresholds, seed, and spending ceiling remain fixed. +[`PROPOSAL.md`](PROPOSAL.md). Both authorized runs are preserved as execution +failures. V1 exposed an interpreter-path bug before scientific execution. V2 +fixed that problem and completed both 4,097-token evaluations, but the emitted +checkpoint failed the mandatory official Transformers reload because the two +prepended layers' tensor names and MLP shapes do not match the saved Qwen +architecture. The terminal v2 evidence is in +[`results/v2-20c3330d0b3e-execution-failure/`](results/v2-20c3330d0b3e-execution-failure/). +The permanent evidence index is +[`f35OgXu…ARZTU`](https://arweave.net/f35OgXuOoulk9EcndJBNWIgZ5tmaQ6DRzeINP7ARZTU/). + +The layer-prepending installer therefore remains experimental. No v3 run is +authorized. The model revision, corpora, thresholds, seed, and spending ceiling +for the completed v2 attempt remain frozen in +[`launch-manifest.json`](launch-manifest.json). This directory turns the open Qwen integration questions into a frozen ilxyr -project for one owner-authorized, bounded attempt. It does not claim that the -full run has passed. It generates the +project and preserves its bounded executions. It does not claim that the full +run passed. It generates the hypothesis, methodology contributions, experiment contract, two explicit model forecasts, funding record, and ordered ilxyr commands for one real checkpoint. diff --git a/experiments/qwen35_acceptance/launch-manifest.json b/experiments/qwen35_acceptance/launch-manifest.json index d5e6e2e..466aeca 100644 --- a/experiments/qwen35_acceptance/launch-manifest.json +++ b/experiments/qwen35_acceptance/launch-manifest.json @@ -1,6 +1,6 @@ { "schema": "lecore.qwen35-launch.v1", - "state": "authorized_prelaunch", + "state": "terminal_execution_failure", "attempt": { "experiment_version": 2, "maximum_formal_runs": 1, @@ -100,7 +100,28 @@ "temporary_staging": "s3://zero-training-022118847419/qwen35-acceptance/2f32151-2fc06364-a6c9d113/v2-formal/", "review_index": "https://github.com/atimics/holostuff/pull/3", "permanent_transport": "Arweave", + "permanent_root": "https://arweave.net/f35OgXuOoulk9EcndJBNWIgZ5tmaQ6DRzeINP7ARZTU/", + "permanent_root_transaction": "f35OgXuOoulk9EcndJBNWIgZ5tmaQ6DRzeINP7ARZTU", + "publication_manifest_sha256": "21874555c7f9e73d20b9d11599c79236f8f40267ee975f703ffc053eb6b1f9ab", "publish_all_outcomes": true, "exclude_model_weights": true + }, + "terminal": { + "completed_at": "2026-08-13T04:51:28.293Z", + "formal_duration_seconds": 15891.556, + "runner_exit_code": 1, + "timed_out": false, + "resolved_outcome": "execution_failure", + "scientific_result": false, + "failure_gate": "official_transformers_checkpoint_reload", + "aws_instance": "i-059aab8bf1d3d87c9", + "estimated_compute_usd": 4.71, + "evidence_bundle_sha256": "df2ba31608f9039e9528a43a93d4766c8e8feb023668a1b9ffc1c85553210c1c", + "ledger_verification": { + "valid": true, + "objects_checked": 15, + "events_checked": 16 + }, + "retry_authorized": false } } diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/README.md b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/README.md new file mode 100644 index 0000000..f03fafd --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/README.md @@ -0,0 +1,80 @@ +# Qwen3.5 acceptance experiment: v2 terminal record + +- Experiment: `lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance` +- ilxyr outcome: `execution_failure` +- AWS instance: `i-059aab8bf1d3d87c9` +- Formal execution: 2026-08-13 00:26:36.737Z–04:51:28.293Z + (`4h 24m 51.556s`), exited `1`, did not time out +- AWS lifecycle: launched 00:24:43Z, self-terminated after evidence upload; + estimated on-demand compute charge `$4.71` plus a few cents of storage +- Ledger verification: valid (`15` objects and `16` events) +- Scientific conclusion: no acceptance or rejection result; the official + checkpoint reload gate failed before the runner could emit its exact metric + contract + +## What happened + +The AWS preflight verified the exact source, model, tokenizer, corpora, minimum +token contract, timeout, virtual-environment interpreter, and disabled spectral +path. ilxyr accepted the project after all twelve admission gates passed. + +The runner then passed the pre-install tokenizer and reference-logit gates, +evaluated `4,097` input tokens from the original checkpoint, ran the experimental +installer, reloaded the emitted checkpoint through leCore's NumPy runtime, and +evaluated the same `4,097` input tokens again. The installer wrote a 2,104.2 MB +checkpoint and reported bit-identical prepend drift in its internal check. + +The required official Transformers reload rejected the two prepended layers: + +```text +linear_attn.in_proj_qkvz.weight | UNEXPECTED +linear_attn.in_proj_ba.weight | UNEXPECTED +linear_attn.in_proj_a.weight | MISSING +linear_attn.in_proj_z.weight | MISSING +linear_attn.in_proj_qkv.weight | MISSING +linear_attn.in_proj_b.weight | MISSING +mlp.up_proj.weight | MISMATCH: [128, 1024] vs [3584, 1024] +mlp.down_proj.weight | MISMATCH: [1024, 128] vs [1024, 3584] +mlp.gate_proj.weight | MISMATCH: [128, 1024] vs [3584, 1024] +``` + +The prepender emitted leCore's packed linear-attention tensors and a 128-wide +MLP, while the saved Qwen configuration causes Transformers to construct split +linear-attention projections and a 3,584-wide MLP for those layers. The leCore +runtime accepts its own heterogeneous representation; the official model does +not. This is precisely the incompatibility the mandatory reload smoke was +designed to expose. + +Because the exception occurred before the runner's single atomic metric +envelope was emitted, ilxyr correctly recorded `execution_failure` with an empty +metric set. Partial internal observations are not promoted to preregistered +scientific metrics. Text generation and the official vision-input smoke did not +run. + +## Decision + +The layer-prepending installer remains experimental and must not be enabled by +default. This run does not invalidate the independently useful Qwen loader, +tokenizer, configuration parsing, text runtime, or diagnostics. A future +implementation should emit the exact official per-layer tensor names and shapes +or encode a genuinely supported heterogeneous layer architecture, then prove +reloadability in a portable fixture before requesting another real-model run. + +No retry is authorized or performed under the v2 identity. + +## Evidence map + +- `project/` contains the exact ilxyr project admitted on AWS. +- `result/` contains the preflight, installer transcript, runtime, status, and + environment records. The redistributable corpora and model checkpoint are + omitted from Git; their exact hashes remain in the manifests. +- `ledger/` is the independently reverified `.ilxyr` event store with the + directory renamed for Git publication. +- `evidence/` contains native ilxyr, RO-Crate, and in-toto exports. +- `publication-manifest.json` is the canonical hash-and-size index generated by + `../../build_publication.py`. + +The original AWS evidence bundle is retained at +`s3://zero-training-022118847419/qwen35-acceptance/2f32151-2fc06364-a6c9d113/v2-formal/qwen35-evidence.tgz` +with SHA-256 +`df2ba31608f9039e9528a43a93d4766c8e8feb023668a1b9ffc1c85553210c1c`. diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.in-toto.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.in-toto.json new file mode 100644 index 0000000..bc0552a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.in-toto.json @@ -0,0 +1,395 @@ +{ + "_type": "https://in-toto.io/Statement/v1", + "predicate": { + "certificates": [], + "cold_replayable": false, + "compiled": { + "evidence_authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [ + "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6", + "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91", + "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c", + "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" + ], + "checker": "checker://lecore/qwen35-acceptance/v2", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "resolved_lineage": { + "engineering_review": "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c", + "experiment_design": "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91", + "hypothesis": "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0", + "mathematical_foundation": "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6" + }, + "schema": "ilxyr.compiled_experiment.v1", + "source_digest": "f6ab72f512e5dcd4acbbd4121456585b594c5c446b131cc0fc12d7ec9d29d037", + "spec": { + "baseline": "baseline://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "datasets": [ + "dataset://lecore/qwen-installation/d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f", + "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + ], + "evidence_authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [], + "checker": "checker://lecore/qwen35-acceptance/v2", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "execution": { + "args": [ + "/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py", + "/opt/qwen35-acceptance-v2/model", + "/opt/qwen35-acceptance-v2/installed-checkpoint", + "/opt/qwen35-acceptance-v2/lecore/REFERENCE.md", + "/opt/qwen35-acceptance-v2/inputs/federalist-papers.txt", + "--min-tokens", + "4096" + ], + "executor": "local-command", + "max_cost_credits": 100, + "network": "open", + "program": "/opt/qwen35-acceptance-v2/venv/bin/python", + "timeout_seconds": 21600 + }, + "expected_outputs": [ + "metrics.acceptance_pass", + "metrics.source_clean", + "metrics.spectral_filtering_enabled", + "metrics.experimental_installer_used", + "metrics.tokenizer_parity_pass", + "metrics.reference_logit_parity_pass", + "metrics.reference_logit_relative_error", + "metrics.eval_tokens", + "metrics.original_perplexity", + "metrics.installed_perplexity", + "metrics.perplexity_delta_pct", + "metrics.paired_ci_lo_nats", + "metrics.paired_ci_hi_nats", + "metrics.statistical_gate_pass", + "metrics.paired_block_length", + "metrics.paired_effective_tokens", + "metrics.peak_rss_mb", + "metrics.peak_gpu_mb", + "metrics.emitted_checkpoint_mb", + "metrics.reload_pass", + "metrics.text_generation_pass", + "metrics.vision_smoke_pass", + "resolved_outcome", + "forecast_settlements" + ], + "funding": { + "minimum_forecasters": 2, + "minimum_total_stake": 10, + "required_compute_credits": 100 + }, + "hypothesis": "The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.", + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "lineage": { + "engineering_review": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.engineering-review", + "experiment_design": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.experiment-design", + "hypothesis": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.hypothesis", + "mathematical_foundation": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.foundation" + }, + "metrics": [ + { + "description": "All mandatory Qwen installation acceptance gates passed, encoded as 0 or 1.", + "name": "acceptance_pass", + "unit": "boolean" + }, + { + "description": "The tracked leCore checkout was clean when the run began, encoded as 0 or 1.", + "name": "source_clean", + "unit": "boolean" + }, + { + "description": "Whether research-only spectral filtering was used; the acceptance path requires 0.", + "name": "spectral_filtering_enabled", + "unit": "boolean" + }, + { + "description": "The explicitly acknowledged layer-prepending installer ran, encoded as 0 or 1.", + "name": "experimental_installer_used", + "unit": "boolean" + }, + { + "description": "leCore and the official Transformers tokenizer produced identical reference token IDs.", + "name": "tokenizer_parity_pass", + "unit": "boolean" + }, + { + "description": "Pre-install leCore logits matched the official Transformers text model within the frozen tolerance.", + "name": "reference_logit_parity_pass", + "unit": "boolean" + }, + { + "description": "Maximum absolute pre-install logit error divided by the maximum absolute reference logit.", + "name": "reference_logit_relative_error", + "unit": "ratio" + }, + { + "description": "Paired token positions included in the streamed evaluation.", + "name": "eval_tokens", + "unit": "tokens" + }, + { + "description": "Original checkpoint perplexity on the frozen corpus and chunking procedure.", + "name": "original_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed checkpoint perplexity on the same token positions.", + "name": "installed_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed minus original perplexity as a percentage of original.", + "name": "perplexity_delta_pct", + "unit": "percent" + }, + { + "description": "Lower 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_lo_nats", + "unit": "nats_per_token" + }, + { + "description": "Upper 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_hi_nats", + "unit": "nats_per_token" + }, + { + "description": "The paired upper confidence bound stayed within the preregistered maximum regression.", + "name": "statistical_gate_pass", + "unit": "boolean" + }, + { + "description": "Moving-block length inferred from autocorrelation in paired token loss differences.", + "name": "paired_block_length", + "unit": "tokens" + }, + { + "description": "Effective paired sample size after serial-correlation adjustment.", + "name": "paired_effective_tokens", + "unit": "tokens" + }, + { + "description": "Peak resident memory observed across the runner and installer child process.", + "name": "peak_rss_mb", + "unit": "megabytes" + }, + { + "description": "Peak accelerator allocation reported by PyTorch, or 0 when no accelerator was used.", + "name": "peak_gpu_mb", + "unit": "megabytes" + }, + { + "description": "Total safetensors size of the emitted installed checkpoint.", + "name": "emitted_checkpoint_mb", + "unit": "megabytes" + }, + { + "description": "The emitted checkpoint reloaded from disk and produced finite logits.", + "name": "reload_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint generated text through the official Transformers model.", + "name": "text_generation_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint accepted a synthetic image through the official Qwen vision-language processor and generated a token.", + "name": "vision_smoke_pass", + "unit": "boolean" + } + ], + "models": [ + "weight://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + ], + "outcome_contract": { + "outcomes": [ + { + "description": "Every frozen Qwen acceptance gate passed.", + "id": "accepted", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "gte", + "threshold": 1.0 + } + }, + { + "description": "The run completed but at least one acceptance gate failed.", + "id": "rejected", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "lt", + "threshold": 1.0 + } + }, + { + "description": "The runner failed or did not emit its exact metric contract.", + "id": "execution_failure", + "predicate": { + "kind": "execution_failure" + } + } + ], + "primary_metric": "acceptance_pass", + "success_outcome": "accepted" + }, + "proposer": { + "id": "model://codex/lecore-qwen/research-director", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director" + }, + "rationale": "This converts the unresolved Qwen integration claims into one frozen, replayable, statistically gated run. Spectral filtering is a separate research control and is forbidden in this contract.", + "schema": "ilxyr.experiment.v1", + "security": { + "code_policy": "arbitrary", + "export_policy": "artifacts", + "weight_class": "public" + }, + "seeds": [ + 0 + ], + "title": "Qwen3.5 leCore layer-prepending acceptance" + } + }, + "compiled_ref": "artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d", + "evidence": { + "authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [ + "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6", + "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91", + "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c", + "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" + ], + "checker": "checker://lecore/qwen35-acceptance/v2", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "id": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737", + "lane": "promoted", + "metrics": {}, + "recorded_at_ms": 1786596688298, + "resolved_outcome": "execution_failure", + "run_ref": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "schema": "ilxyr.evidence.v1" + }, + "evidence_event_hash": "65b4d4afb9170be4496b9fbc168d5e48acca7a24b0c3f5ecb76b274e29cca882", + "evidence_ref": "artifact://sha256/15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc", + "executor_attestations": [], + "executor_attested": false, + "forecast_risked": true, + "forecasts": [ + { + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-mechanistic", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic", + "probabilities": { + "accepted": 0.15, + "execution_failure": 0.3, + "rejected": 0.55 + }, + "rationale": "The blank prepend has a strong identity construction, while full-checkpoint memory pressure and official multimodal reload remain material execution risks.", + "schema": "ilxyr.forecast.v1", + "stake": 5 + }, + { + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-empirical", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical", + "probabilities": { + "accepted": 0.2, + "execution_failure": 0.15, + "rejected": 0.65 + }, + "rationale": "Structural rehearsal passes, but no complete post-memory-fix real-Qwen install and powered acceptance run has succeeded yet.", + "schema": "ilxyr.forecast.v1", + "stake": 5 + } + ], + "ledger_head": "2bfa3e37875427fc823b948a6d84497a8ddc73664f6da032cf23e6bfdff0816a", + "run": { + "completed_at_ms": 1786596688293, + "exit_code": 1, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "id": "run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737", + "metrics": {}, + "output_truncated": false, + "schema": "ilxyr.run.v1", + "started_at_ms": 1786580796737, + "stderr": "[transformers] `torch_dtype` is deprecated! Use `dtype` instead!\n[transformers] The fast path is not available because one of the required library is not installed. Falling back to torch implementation. To install follow https://github.com/fla-org/flash-linear-attention#installation and https://github.com/Dao-AILab/causal-conv1d\n\rLoading weights: 0%| | 0/320 [00:00\n raise SystemExit(main())\n ^^^^^^\n File \"/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py\", line 296, in main\n text_pass, vision_pass, peak_gpu = official_output_smokes(installed_dir)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py\", line 158, in official_output_smokes\n model = Qwen3_5ForConditionalGeneration.from_pretrained(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/modeling_utils.py\", line 4457, in from_pretrained\n loading_info = cls._finalize_model_loading(model, load_config, loading_info)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/modeling_utils.py\", line 4633, in _finalize_model_loading\n log_state_dict_report(\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/utils/loading_report.py\", line 278, in log_state_dict_report\n raise RuntimeError(\nRuntimeError: You set `ignore_mismatched_sizes` to `False`, thus raising an error. For details look at the above report!\n", + "stdout": " qkv layout: flat (ppl 20.34 vs 201031.79 for grouped -- 9881.2x better)\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n", + "timed_out": false + }, + "run_ref": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "schema": "ilxyr.evidence_bundle.v1", + "settlements": [ + { + "brier_score": 0.815, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic", + "resolved_outcome": "execution_failure", + "schema": "ilxyr.forecast_settlement.v1", + "stake": 5 + }, + { + "brier_score": 1.185, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical", + "resolved_outcome": "execution_failure", + "schema": "ilxyr.forecast_settlement.v1", + "stake": 5 + } + ], + "source_attested": false + }, + "predicateType": "https://ilxyr.dev/attestations/evidence/v1", + "subject": [ + { + "digest": { + "sha256": "15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc" + }, + "name": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737" + } + ] +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.native.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.native.json new file mode 100644 index 0000000..ec74b99 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.native.json @@ -0,0 +1,383 @@ +{ + "certificates": [], + "cold_replayable": false, + "compiled": { + "evidence_authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [ + "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6", + "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91", + "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c", + "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" + ], + "checker": "checker://lecore/qwen35-acceptance/v2", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "resolved_lineage": { + "engineering_review": "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c", + "experiment_design": "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91", + "hypothesis": "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0", + "mathematical_foundation": "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6" + }, + "schema": "ilxyr.compiled_experiment.v1", + "source_digest": "f6ab72f512e5dcd4acbbd4121456585b594c5c446b131cc0fc12d7ec9d29d037", + "spec": { + "baseline": "baseline://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "datasets": [ + "dataset://lecore/qwen-installation/d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f", + "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + ], + "evidence_authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [], + "checker": "checker://lecore/qwen35-acceptance/v2", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "execution": { + "args": [ + "/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py", + "/opt/qwen35-acceptance-v2/model", + "/opt/qwen35-acceptance-v2/installed-checkpoint", + "/opt/qwen35-acceptance-v2/lecore/REFERENCE.md", + "/opt/qwen35-acceptance-v2/inputs/federalist-papers.txt", + "--min-tokens", + "4096" + ], + "executor": "local-command", + "max_cost_credits": 100, + "network": "open", + "program": "/opt/qwen35-acceptance-v2/venv/bin/python", + "timeout_seconds": 21600 + }, + "expected_outputs": [ + "metrics.acceptance_pass", + "metrics.source_clean", + "metrics.spectral_filtering_enabled", + "metrics.experimental_installer_used", + "metrics.tokenizer_parity_pass", + "metrics.reference_logit_parity_pass", + "metrics.reference_logit_relative_error", + "metrics.eval_tokens", + "metrics.original_perplexity", + "metrics.installed_perplexity", + "metrics.perplexity_delta_pct", + "metrics.paired_ci_lo_nats", + "metrics.paired_ci_hi_nats", + "metrics.statistical_gate_pass", + "metrics.paired_block_length", + "metrics.paired_effective_tokens", + "metrics.peak_rss_mb", + "metrics.peak_gpu_mb", + "metrics.emitted_checkpoint_mb", + "metrics.reload_pass", + "metrics.text_generation_pass", + "metrics.vision_smoke_pass", + "resolved_outcome", + "forecast_settlements" + ], + "funding": { + "minimum_forecasters": 2, + "minimum_total_stake": 10, + "required_compute_credits": 100 + }, + "hypothesis": "The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.", + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "lineage": { + "engineering_review": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.engineering-review", + "experiment_design": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.experiment-design", + "hypothesis": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.hypothesis", + "mathematical_foundation": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.foundation" + }, + "metrics": [ + { + "description": "All mandatory Qwen installation acceptance gates passed, encoded as 0 or 1.", + "name": "acceptance_pass", + "unit": "boolean" + }, + { + "description": "The tracked leCore checkout was clean when the run began, encoded as 0 or 1.", + "name": "source_clean", + "unit": "boolean" + }, + { + "description": "Whether research-only spectral filtering was used; the acceptance path requires 0.", + "name": "spectral_filtering_enabled", + "unit": "boolean" + }, + { + "description": "The explicitly acknowledged layer-prepending installer ran, encoded as 0 or 1.", + "name": "experimental_installer_used", + "unit": "boolean" + }, + { + "description": "leCore and the official Transformers tokenizer produced identical reference token IDs.", + "name": "tokenizer_parity_pass", + "unit": "boolean" + }, + { + "description": "Pre-install leCore logits matched the official Transformers text model within the frozen tolerance.", + "name": "reference_logit_parity_pass", + "unit": "boolean" + }, + { + "description": "Maximum absolute pre-install logit error divided by the maximum absolute reference logit.", + "name": "reference_logit_relative_error", + "unit": "ratio" + }, + { + "description": "Paired token positions included in the streamed evaluation.", + "name": "eval_tokens", + "unit": "tokens" + }, + { + "description": "Original checkpoint perplexity on the frozen corpus and chunking procedure.", + "name": "original_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed checkpoint perplexity on the same token positions.", + "name": "installed_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed minus original perplexity as a percentage of original.", + "name": "perplexity_delta_pct", + "unit": "percent" + }, + { + "description": "Lower 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_lo_nats", + "unit": "nats_per_token" + }, + { + "description": "Upper 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_hi_nats", + "unit": "nats_per_token" + }, + { + "description": "The paired upper confidence bound stayed within the preregistered maximum regression.", + "name": "statistical_gate_pass", + "unit": "boolean" + }, + { + "description": "Moving-block length inferred from autocorrelation in paired token loss differences.", + "name": "paired_block_length", + "unit": "tokens" + }, + { + "description": "Effective paired sample size after serial-correlation adjustment.", + "name": "paired_effective_tokens", + "unit": "tokens" + }, + { + "description": "Peak resident memory observed across the runner and installer child process.", + "name": "peak_rss_mb", + "unit": "megabytes" + }, + { + "description": "Peak accelerator allocation reported by PyTorch, or 0 when no accelerator was used.", + "name": "peak_gpu_mb", + "unit": "megabytes" + }, + { + "description": "Total safetensors size of the emitted installed checkpoint.", + "name": "emitted_checkpoint_mb", + "unit": "megabytes" + }, + { + "description": "The emitted checkpoint reloaded from disk and produced finite logits.", + "name": "reload_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint generated text through the official Transformers model.", + "name": "text_generation_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint accepted a synthetic image through the official Qwen vision-language processor and generated a token.", + "name": "vision_smoke_pass", + "unit": "boolean" + } + ], + "models": [ + "weight://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + ], + "outcome_contract": { + "outcomes": [ + { + "description": "Every frozen Qwen acceptance gate passed.", + "id": "accepted", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "gte", + "threshold": 1.0 + } + }, + { + "description": "The run completed but at least one acceptance gate failed.", + "id": "rejected", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "lt", + "threshold": 1.0 + } + }, + { + "description": "The runner failed or did not emit its exact metric contract.", + "id": "execution_failure", + "predicate": { + "kind": "execution_failure" + } + } + ], + "primary_metric": "acceptance_pass", + "success_outcome": "accepted" + }, + "proposer": { + "id": "model://codex/lecore-qwen/research-director", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director" + }, + "rationale": "This converts the unresolved Qwen integration claims into one frozen, replayable, statistically gated run. Spectral filtering is a separate research control and is forbidden in this contract.", + "schema": "ilxyr.experiment.v1", + "security": { + "code_policy": "arbitrary", + "export_policy": "artifacts", + "weight_class": "public" + }, + "seeds": [ + 0 + ], + "title": "Qwen3.5 leCore layer-prepending acceptance" + } + }, + "compiled_ref": "artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d", + "evidence": { + "authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [ + "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6", + "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91", + "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c", + "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" + ], + "checker": "checker://lecore/qwen35-acceptance/v2", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "id": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737", + "lane": "promoted", + "metrics": {}, + "recorded_at_ms": 1786596688298, + "resolved_outcome": "execution_failure", + "run_ref": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "schema": "ilxyr.evidence.v1" + }, + "evidence_event_hash": "65b4d4afb9170be4496b9fbc168d5e48acca7a24b0c3f5ecb76b274e29cca882", + "evidence_ref": "artifact://sha256/15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc", + "executor_attestations": [], + "executor_attested": false, + "forecast_risked": true, + "forecasts": [ + { + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-mechanistic", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic", + "probabilities": { + "accepted": 0.15, + "execution_failure": 0.3, + "rejected": 0.55 + }, + "rationale": "The blank prepend has a strong identity construction, while full-checkpoint memory pressure and official multimodal reload remain material execution risks.", + "schema": "ilxyr.forecast.v1", + "stake": 5 + }, + { + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-empirical", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical", + "probabilities": { + "accepted": 0.2, + "execution_failure": 0.15, + "rejected": 0.65 + }, + "rationale": "Structural rehearsal passes, but no complete post-memory-fix real-Qwen install and powered acceptance run has succeeded yet.", + "schema": "ilxyr.forecast.v1", + "stake": 5 + } + ], + "ledger_head": "2bfa3e37875427fc823b948a6d84497a8ddc73664f6da032cf23e6bfdff0816a", + "run": { + "completed_at_ms": 1786596688293, + "exit_code": 1, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "id": "run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737", + "metrics": {}, + "output_truncated": false, + "schema": "ilxyr.run.v1", + "started_at_ms": 1786580796737, + "stderr": "[transformers] `torch_dtype` is deprecated! Use `dtype` instead!\n[transformers] The fast path is not available because one of the required library is not installed. Falling back to torch implementation. To install follow https://github.com/fla-org/flash-linear-attention#installation and https://github.com/Dao-AILab/causal-conv1d\n\rLoading weights: 0%| | 0/320 [00:00\n raise SystemExit(main())\n ^^^^^^\n File \"/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py\", line 296, in main\n text_pass, vision_pass, peak_gpu = official_output_smokes(installed_dir)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py\", line 158, in official_output_smokes\n model = Qwen3_5ForConditionalGeneration.from_pretrained(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/modeling_utils.py\", line 4457, in from_pretrained\n loading_info = cls._finalize_model_loading(model, load_config, loading_info)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/modeling_utils.py\", line 4633, in _finalize_model_loading\n log_state_dict_report(\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/utils/loading_report.py\", line 278, in log_state_dict_report\n raise RuntimeError(\nRuntimeError: You set `ignore_mismatched_sizes` to `False`, thus raising an error. For details look at the above report!\n", + "stdout": " qkv layout: flat (ppl 20.34 vs 201031.79 for grouped -- 9881.2x better)\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n", + "timed_out": false + }, + "run_ref": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "schema": "ilxyr.evidence_bundle.v1", + "settlements": [ + { + "brier_score": 0.815, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic", + "resolved_outcome": "execution_failure", + "schema": "ilxyr.forecast_settlement.v1", + "stake": 5 + }, + { + "brier_score": 1.185, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical", + "resolved_outcome": "execution_failure", + "schema": "ilxyr.forecast_settlement.v1", + "stake": 5 + } + ], + "source_attested": false +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.ro-crate.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.ro-crate.json new file mode 100644 index 0000000..5221d3c --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/evidence.ro-crate.json @@ -0,0 +1,286 @@ +{ + "@context": [ + "https://w3id.org/ro/crate/1.3/context", + { + "ilxyr": "https://ilxyr.dev/ns#", + "prov": "http://www.w3.org/ns/prov#" + } + ], + "@graph": [ + { + "@id": "ro-crate-metadata.json", + "@type": "CreativeWork", + "about": { + "@id": "./" + }, + "conformsTo": { + "@id": "https://w3id.org/ro/crate/1.3" + } + }, + { + "@id": "./", + "@type": [ + "Dataset", + "prov:Entity" + ], + "description": "A ledger-verified ilXyr evidence bundle exported for discovery, preservation, and reuse.", + "hasPart": [ + { + "@id": "artifact://sha256/15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc" + }, + { + "@id": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f" + }, + { + "@id": "artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d" + }, + { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic" + }, + { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical" + }, + { + "@id": "urn:ilxyr:settlement:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic" + }, + { + "@id": "urn:ilxyr:settlement:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical" + }, + { + "@id": "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6" + }, + { + "@id": "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91" + }, + { + "@id": "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c" + }, + { + "@id": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f" + }, + { + "@id": "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" + } + ], + "ilxyr:ledgerHead": "2bfa3e37875427fc823b948a6d84497a8ddc73664f6da032cf23e6bfdff0816a", + "mainEntity": { + "@id": "artifact://sha256/15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc" + }, + "name": "ilXyr evidence bundle — Qwen3.5 leCore layer-prepending acceptance" + }, + { + "@id": "artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d", + "@type": [ + "CreativeWork", + "prov:Plan" + ], + "author": { + "@id": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director", + "@type": "SoftwareApplication", + "identifier": "model://codex/lecore-qwen/research-director" + }, + "description": "The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.", + "identifier": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "ilxyr:sharedTask": null, + "ilxyr:sourceDigest": "f6ab72f512e5dcd4acbbd4121456585b594c5c446b131cc0fc12d7ec9d29d037", + "isBasedOn": [ + { + "@id": "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c" + }, + { + "@id": "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91" + }, + { + "@id": "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" + }, + { + "@id": "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6" + } + ], + "name": "Qwen3.5 leCore layer-prepending acceptance" + }, + { + "@id": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "@type": [ + "CreateAction", + "prov:Activity" + ], + "actionStatus": "CompletedActionStatus", + "ilxyr:run": { + "completed_at_ms": 1786596688293, + "exit_code": 1, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "id": "run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737", + "metrics": {}, + "output_truncated": false, + "schema": "ilxyr.run.v1", + "started_at_ms": 1786580796737, + "stderr": "[transformers] `torch_dtype` is deprecated! Use `dtype` instead!\n[transformers] The fast path is not available because one of the required library is not installed. Falling back to torch implementation. To install follow https://github.com/fla-org/flash-linear-attention#installation and https://github.com/Dao-AILab/causal-conv1d\n\rLoading weights: 0%| | 0/320 [00:00\n raise SystemExit(main())\n ^^^^^^\n File \"/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py\", line 296, in main\n text_pass, vision_pass, peak_gpu = official_output_smokes(installed_dir)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py\", line 158, in official_output_smokes\n model = Qwen3_5ForConditionalGeneration.from_pretrained(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/modeling_utils.py\", line 4457, in from_pretrained\n loading_info = cls._finalize_model_loading(model, load_config, loading_info)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/modeling_utils.py\", line 4633, in _finalize_model_loading\n log_state_dict_report(\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/utils/loading_report.py\", line 278, in log_state_dict_report\n raise RuntimeError(\nRuntimeError: You set `ignore_mismatched_sizes` to `False`, thus raising an error. For details look at the above report!\n", + "stdout": " qkv layout: flat (ppl 20.34 vs 201031.79 for grouped -- 9881.2x better)\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n", + "timed_out": false + }, + "ilxyr:sourceAttested": false, + "name": "Execution for lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "object": { + "@id": "artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d" + }, + "result": { + "@id": "artifact://sha256/15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc" + } + }, + { + "@id": "artifact://sha256/15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc", + "@type": [ + "Dataset", + "prov:Entity" + ], + "additionalProperty": [], + "description": "Resolved outcome: execution_failure", + "identifier": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737", + "ilxyr:authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [ + "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6", + "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91", + "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c", + "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" + ], + "checker": "checker://lecore/qwen35-acceptance/v2", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "ilxyr:coldReplayable": false, + "ilxyr:evidenceEventHash": "65b4d4afb9170be4496b9fbc168d5e48acca7a24b0c3f5ecb76b274e29cca882", + "ilxyr:evidenceLane": "promoted", + "ilxyr:executorAttested": false, + "ilxyr:externallyPreregistered": false, + "ilxyr:forecastRisked": true, + "isBasedOn": { + "@id": "artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d" + }, + "name": "Evidence for Qwen3.5 leCore layer-prepending acceptance", + "prov:wasGeneratedBy": { + "@id": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f" + } + }, + { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic", + "@type": [ + "CreativeWork", + "prov:Entity" + ], + "about": { + "@id": "artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d" + }, + "author": { + "@id": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic", + "@type": "SoftwareApplication", + "identifier": "model://codex/lecore-qwen/forecaster-mechanistic" + }, + "identifier": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic", + "ilxyr:probabilities": { + "accepted": 0.15, + "execution_failure": 0.3, + "rejected": 0.55 + }, + "ilxyr:stake": 5, + "name": "Forecast for lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance" + }, + { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical", + "@type": [ + "CreativeWork", + "prov:Entity" + ], + "about": { + "@id": "artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d" + }, + "author": { + "@id": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical", + "@type": "SoftwareApplication", + "identifier": "model://codex/lecore-qwen/forecaster-empirical" + }, + "identifier": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical", + "ilxyr:probabilities": { + "accepted": 0.2, + "execution_failure": 0.15, + "rejected": 0.65 + }, + "ilxyr:stake": 5, + "name": "Forecast for lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance" + }, + { + "@id": "urn:ilxyr:settlement:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic", + "@type": [ + "CreateAction", + "prov:Activity" + ], + "ilxyr:brierScore": 0.815, + "ilxyr:resolvedOutcome": "execution_failure", + "name": "Forecast settlement for lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic", + "object": { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic" + }, + "result": { + "@id": "artifact://sha256/15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc" + } + }, + { + "@id": "urn:ilxyr:settlement:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical", + "@type": [ + "CreateAction", + "prov:Activity" + ], + "ilxyr:brierScore": 1.185, + "ilxyr:resolvedOutcome": "execution_failure", + "name": "Forecast settlement for lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical", + "object": { + "@id": "urn:ilxyr:forecast:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical" + }, + "result": { + "@id": "artifact://sha256/15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc" + } + }, + { + "@id": "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6", + "@type": [ + "MediaObject", + "prov:Entity" + ], + "sha256": "949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6" + }, + { + "@id": "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91", + "@type": [ + "MediaObject", + "prov:Entity" + ], + "sha256": "9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91" + }, + { + "@id": "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c", + "@type": [ + "MediaObject", + "prov:Entity" + ], + "sha256": "9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c" + }, + { + "@id": "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0", + "@type": [ + "MediaObject", + "prov:Entity" + ], + "sha256": "ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" + } + ] +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/exporter-provenance.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/exporter-provenance.json new file mode 100644 index 0000000..01333e5 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/evidence/exporter-provenance.json @@ -0,0 +1,23 @@ +{ + "schema": "lecore.ilxyr-exporter-provenance.v1", + "export_mode": "read_only_isolated_snapshot", + "source_repository": "/Users/ratimics/develop/ilxyr", + "source_base_commit": "e92382ff2a5e8714466533a160f6609b4ef9cee8", + "source_worktree_clean": false, + "source_snapshot": "/private/tmp/ilxyr-exporter-snapshot", + "source_snapshot_note": "The exporter was compiled from an isolated copy of the user's existing dirty ilxyr worktree. The original worktree and the verified AWS ledger were not modified.", + "exporter_binary": { + "path": "/private/tmp/ilxyr-exporter-snapshot/target/debug/ilxyr", + "sha256": "628c153cfe64a1b916551194f7cc41fb5dbc13954a7d2b9a956641b71e97bd29" + }, + "source_files": { + "Cargo.lock": "4d5a64a9ce286006325c912c8e3e716c427e50e01cc28d72d72d9978d1a7b42b", + "crates/ilxyr-core/src/interop.rs": "b3db6eb51891abecb241bc38751b78d8eb67a4245aed7d28e514c858223d9982", + "crates/ilxyr-cli/src/main.rs": "e3c3d8d7364fbe22af4b927f7b01c58b57f68480e242382c03ffd7854f9d71cb" + }, + "exports": { + "evidence.native.json": "11d8c155cce37e70e2eb8241e437b6c7190b6a7cbbbc3ffe329bfbe33bed4d95", + "evidence.ro-crate.json": "0aa5e25a67a0e0f6de62574097e05102bd1871ca4c9bfddf2335b2ceff38b50c", + "evidence.in-toto.json": "327ef5cf6c2f37e9b8afdabc0b6ca5546493292717a1abb09a9bf8618d320d93" + } +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/ledger/config.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/ledger/config.json new file mode 100644 index 0000000..c7becc4 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/ledger/config.json @@ -0,0 +1,5 @@ +{ + "ledger_mode": "single_writer", + "object_hash": "sha256", + "schema": "ilxyr.workspace.v1" +} \ No newline at end of file diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/ledger/events.jsonl b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/ledger/events.jsonl new file mode 100644 index 0000000..387ab8c --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/ledger/events.jsonl @@ -0,0 +1,16 @@ +{"actor":{"id":"model://codex/lecore-qwen/research-director","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.hypothesis","artifact_ref":"artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0","event_hash":"7804433f3d521b50ae1ce8104fe5cedb893c6d35bf849bb8672cb22cb3128ba3","event_type":"ContributionSubmitted","occurred_at_ms":1786580796199,"schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/statistical-reviewer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/statistical-reviewer"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.foundation","artifact_ref":"artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6","event_hash":"856e681c586728c649369d1a051a3079d6e8177374f237c21d907a718aa60041","event_type":"ContributionSubmitted","occurred_at_ms":1786580796205,"previous_event":"7804433f3d521b50ae1ce8104fe5cedb893c6d35bf849bb8672cb22cb3128ba3","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/engineering-reviewer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/engineering-reviewer"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.engineering-review","artifact_ref":"artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c","event_hash":"8be939a12cc3e71d8326e273ae0682f85a36d4e8b83d423421cd33936fbd5ba9","event_type":"ContributionSubmitted","occurred_at_ms":1786580796211,"previous_event":"856e681c586728c649369d1a051a3079d6e8177374f237c21d907a718aa60041","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/experiment-designer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/experiment-designer"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.experiment-design","artifact_ref":"artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91","event_hash":"f3d086a2f95c4316fa5294cfb0af605d725ff56baf0e67558e6017ec473092f8","event_type":"ContributionSubmitted","occurred_at_ms":1786580796218,"previous_event":"8be939a12cc3e71d8326e273ae0682f85a36d4e8b83d423421cd33936fbd5ba9","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/research-director","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance","artifact_ref":"artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d","event_hash":"185bb369209a9017032bf944e351047542ae9054c23592282b9918b8806edc4d","event_type":"ExperimentCompiled","occurred_at_ms":1786580796225,"previous_event":"f3d086a2f95c4316fa5294cfb0af605d725ff56baf0e67558e6017ec473092f8","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/forecaster-empirical","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical","artifact_ref":"artifact://sha256/32f36305742072ad04d02059b866f824532c2afaefe72151197de23d8967bec6","event_hash":"caa223d7c9a9f2139683268ec7d70a2f32436742d65ecf8f21f8c7053ff295f7","event_type":"ForecastSubmitted","occurred_at_ms":1786580796232,"previous_event":"185bb369209a9017032bf944e351047542ae9054c23592282b9918b8806edc4d","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/forecaster-mechanistic","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic","artifact_ref":"artifact://sha256/b285b2b9ee58772ca6059775c41bb01f2cc75e3e097ca538dea75040c997e451","event_hash":"de657735b05813df2f21dad9313017e25f8d27111c82aa29f866e3077db28b9f","event_type":"ForecastSubmitted","occurred_at_ms":1786580796238,"previous_event":"caa223d7c9a9f2139683268ec7d70a2f32436742d65ecf8f21f8c7053ff295f7","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://lecore/qwen-experiment-generator","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.funding","artifact_ref":"artifact://sha256/a69efa569914df801764e85672dcdad748c424792de726bcbdd5afd9fa13fdb7","event_hash":"150ec21632d5fe17394f490f41f957bbbebcc63b31bf0441be5600ffea132d1b","event_type":"FundingCommitted","occurred_at_ms":1786580796245,"previous_event":"de657735b05813df2f21dad9313017e25f8d27111c82aa29f866e3077db28b9f","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/admission-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance","artifact_ref":"artifact://sha256/acd9e2ef1808f9d8465a3464b7637154c7a40575761fc67e46693b688a5b904d","event_hash":"427f58d84fa304fb74cd19de4cc7cc3090ce6e06ee2920f00bb9243aa9a707ac","event_type":"AdmissionDecided","occurred_at_ms":1786580796251,"previous_event":"150ec21632d5fe17394f490f41f957bbbebcc63b31bf0441be5600ffea132d1b","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/local-executor-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance","event_hash":"5cff2a864d9e4ee0545a1fa44d8517fcb69a9d6f9d7c7385f3c108f85d351157","event_type":"ExecutionStarted","occurred_at_ms":1786580796735,"previous_event":"427f58d84fa304fb74cd19de4cc7cc3090ce6e06ee2920f00bb9243aa9a707ac","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/local-executor-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance","artifact_ref":"artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f","event_hash":"502db6acaa0b6b1de6c0150c3d583536ee6aacbea1111192c4815cf0b8667b0a","event_type":"ExperimentCompleted","occurred_at_ms":1786596688296,"previous_event":"5cff2a864d9e4ee0545a1fa44d8517fcb69a9d6f9d7c7385f3c108f85d351157","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/resolver-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance","artifact_ref":"artifact://sha256/15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc","event_hash":"65b4d4afb9170be4496b9fbc168d5e48acca7a24b0c3f5ecb76b274e29cca882","event_type":"EvidenceRecorded","occurred_at_ms":1786596688300,"previous_event":"502db6acaa0b6b1de6c0150c3d583536ee6aacbea1111192c4815cf0b8667b0a","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/scoring-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical","artifact_ref":"artifact://sha256/3b1d817febf70072b1b926cdcf5a7a8fd388bd9708531b2b2e21cf74c2c2c523","event_hash":"ea22dfe495019b4d226fc8e337b4c6564baa3ef6b381fb360ca2ed48090acd27","event_type":"ForecastSettled","occurred_at_ms":1786596688305,"previous_event":"65b4d4afb9170be4496b9fbc168d5e48acca7a24b0c3f5ecb76b274e29cca882","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/scoring-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic","artifact_ref":"artifact://sha256/16179101f5c235f4d6b1e437c3edc0775b4542057053ebb0df36da357fa7fc0a","event_hash":"eba7e31d7ff90e2487d0f4f5d31c85d53084088d15a4a34ab8f6a1623b0d4764","event_type":"ForecastSettled","occurred_at_ms":1786596688309,"previous_event":"ea22dfe495019b4d226fc8e337b4c6564baa3ef6b381fb360ca2ed48090acd27","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/calibration-v1","kind":"service"},"aggregate_id":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical","artifact_ref":"artifact://sha256/a1dc5e92d5ff354a80a296f223ba1bcf02f1b586400ebb642c22ca365bdc13a0","event_hash":"7405116c71578d9754838e36033b919e8bbd8282115d47b785c342079d92dfba","event_type":"CalibrationUpdated","occurred_at_ms":1786596688314,"previous_event":"eba7e31d7ff90e2487d0f4f5d31c85d53084088d15a4a34ab8f6a1623b0d4764","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/calibration-v1","kind":"service"},"aggregate_id":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic","artifact_ref":"artifact://sha256/d65eb4ca74cd5d04fadd30ed19e0b4808267f3308d4c99437b0857a375ad121e","event_hash":"2bfa3e37875427fc823b948a6d84497a8ddc73664f6da032cf23e6bfdff0816a","event_type":"CalibrationUpdated","occurred_at_ms":1786596688319,"previous_event":"7405116c71578d9754838e36033b919e8bbd8282115d47b785c342079d92dfba","schema":"ilxyr.event.v1"} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/engineering-review.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/engineering-review.json new file mode 100644 index 0000000..a493cd1 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/engineering-review.json @@ -0,0 +1,21 @@ +{ + "actor": { + "id": "model://codex/lecore-qwen/engineering-reviewer", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/engineering-reviewer" + }, + "body": "The shell-free runner uses absolute paths, records the exact leCore commit and checker hashes, keeps spectral filtering disabled, invokes the layer-prepending path only with its experimental acknowledgement, records peak memory, reloads the emitted artifact, and exercises the official Transformers text and image-text interfaces.", + "claims": [ + "The emitted stdout is exactly the ilxyr metrics/source envelope.", + "Installer logs and a human-readable metrics artifact are retained beside the output checkpoint." + ], + "confidence": 0.85, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.engineering-review", + "input_refs": [ + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.hypothesis", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.foundation" + ], + "schema": "ilxyr.contribution.v1", + "stage": "engineering_review", + "title": "Qwen acceptance runner and provenance boundary" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/experiment-design.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/experiment-design.json new file mode 100644 index 0000000..be2cbe5 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/experiment-design.json @@ -0,0 +1,22 @@ +{ + "actor": { + "id": "model://codex/lecore-qwen/experiment-designer", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/experiment-designer" + }, + "body": "Execute once against the content-bound public checkpoint, installation corpus, and separate held-out evaluation corpus. Do not tune thresholds or replace either corpus after admission. Resolve accepted only when source cleanliness, tokenizer parity, reference-logit parity, the paired statistical gate, disk reload, official text generation, and official vision smoke all pass. A cleanly executed no-go is preserved as rejected evidence.", + "claims": [ + "Accepted and rejected are exhaustive for a valid metrics envelope.", + "Runtime or dependency failure resolves separately as execution_failure." + ], + "confidence": 0.95, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.experiment-design", + "input_refs": [ + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.hypothesis", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.foundation", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.engineering-review" + ], + "schema": "ilxyr.contribution.v1", + "stage": "experiment_design", + "title": "One-shot Qwen3.5 installation acceptance run" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/experiment.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/experiment.json new file mode 100644 index 0000000..934b1a5 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/experiment.json @@ -0,0 +1,241 @@ +{ + "baseline": "baseline://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "datasets": [ + "dataset://lecore/qwen-installation/d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f", + "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + ], + "evidence_authority": { + "level": "corpus_proxy", + "provenance": { + "artifact_hashes": [], + "checker": "checker://lecore/qwen35-acceptance/v2", + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + }, + "scope": { + "coverage": 1.0, + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "seeds": [ + 0 + ] + } + }, + "execution": { + "args": [ + "/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py", + "/opt/qwen35-acceptance-v2/model", + "/opt/qwen35-acceptance-v2/installed-checkpoint", + "/opt/qwen35-acceptance-v2/lecore/REFERENCE.md", + "/opt/qwen35-acceptance-v2/inputs/federalist-papers.txt", + "--min-tokens", + "4096" + ], + "executor": "local-command", + "max_cost_credits": 100, + "network": "open", + "program": "/opt/qwen35-acceptance-v2/venv/bin/python", + "timeout_seconds": 21600 + }, + "expected_outputs": [ + "metrics.acceptance_pass", + "metrics.source_clean", + "metrics.spectral_filtering_enabled", + "metrics.experimental_installer_used", + "metrics.tokenizer_parity_pass", + "metrics.reference_logit_parity_pass", + "metrics.reference_logit_relative_error", + "metrics.eval_tokens", + "metrics.original_perplexity", + "metrics.installed_perplexity", + "metrics.perplexity_delta_pct", + "metrics.paired_ci_lo_nats", + "metrics.paired_ci_hi_nats", + "metrics.statistical_gate_pass", + "metrics.paired_block_length", + "metrics.paired_effective_tokens", + "metrics.peak_rss_mb", + "metrics.peak_gpu_mb", + "metrics.emitted_checkpoint_mb", + "metrics.reload_pass", + "metrics.text_generation_pass", + "metrics.vision_smoke_pass", + "resolved_outcome", + "forecast_settlements" + ], + "funding": { + "minimum_forecasters": 2, + "minimum_total_stake": 10, + "required_compute_credits": 100 + }, + "hypothesis": "The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.", + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "lineage": { + "engineering_review": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.engineering-review", + "experiment_design": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.experiment-design", + "hypothesis": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.hypothesis", + "mathematical_foundation": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.foundation" + }, + "metrics": [ + { + "description": "All mandatory Qwen installation acceptance gates passed, encoded as 0 or 1.", + "name": "acceptance_pass", + "unit": "boolean" + }, + { + "description": "The tracked leCore checkout was clean when the run began, encoded as 0 or 1.", + "name": "source_clean", + "unit": "boolean" + }, + { + "description": "Whether research-only spectral filtering was used; the acceptance path requires 0.", + "name": "spectral_filtering_enabled", + "unit": "boolean" + }, + { + "description": "The explicitly acknowledged layer-prepending installer ran, encoded as 0 or 1.", + "name": "experimental_installer_used", + "unit": "boolean" + }, + { + "description": "leCore and the official Transformers tokenizer produced identical reference token IDs.", + "name": "tokenizer_parity_pass", + "unit": "boolean" + }, + { + "description": "Pre-install leCore logits matched the official Transformers text model within the frozen tolerance.", + "name": "reference_logit_parity_pass", + "unit": "boolean" + }, + { + "description": "Maximum absolute pre-install logit error divided by the maximum absolute reference logit.", + "name": "reference_logit_relative_error", + "unit": "ratio" + }, + { + "description": "Paired token positions included in the streamed evaluation.", + "name": "eval_tokens", + "unit": "tokens" + }, + { + "description": "Original checkpoint perplexity on the frozen corpus and chunking procedure.", + "name": "original_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed checkpoint perplexity on the same token positions.", + "name": "installed_perplexity", + "unit": "perplexity" + }, + { + "description": "Installed minus original perplexity as a percentage of original.", + "name": "perplexity_delta_pct", + "unit": "percent" + }, + { + "description": "Lower 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_lo_nats", + "unit": "nats_per_token" + }, + { + "description": "Upper 95 percent paired moving-block-bootstrap bound for installed minus original NLL.", + "name": "paired_ci_hi_nats", + "unit": "nats_per_token" + }, + { + "description": "The paired upper confidence bound stayed within the preregistered maximum regression.", + "name": "statistical_gate_pass", + "unit": "boolean" + }, + { + "description": "Moving-block length inferred from autocorrelation in paired token loss differences.", + "name": "paired_block_length", + "unit": "tokens" + }, + { + "description": "Effective paired sample size after serial-correlation adjustment.", + "name": "paired_effective_tokens", + "unit": "tokens" + }, + { + "description": "Peak resident memory observed across the runner and installer child process.", + "name": "peak_rss_mb", + "unit": "megabytes" + }, + { + "description": "Peak accelerator allocation reported by PyTorch, or 0 when no accelerator was used.", + "name": "peak_gpu_mb", + "unit": "megabytes" + }, + { + "description": "Total safetensors size of the emitted installed checkpoint.", + "name": "emitted_checkpoint_mb", + "unit": "megabytes" + }, + { + "description": "The emitted checkpoint reloaded from disk and produced finite logits.", + "name": "reload_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint generated text through the official Transformers model.", + "name": "text_generation_pass", + "unit": "boolean" + }, + { + "description": "The emitted checkpoint accepted a synthetic image through the official Qwen vision-language processor and generated a token.", + "name": "vision_smoke_pass", + "unit": "boolean" + } + ], + "models": [ + "weight://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680" + ], + "outcome_contract": { + "outcomes": [ + { + "description": "Every frozen Qwen acceptance gate passed.", + "id": "accepted", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "gte", + "threshold": 1 + } + }, + { + "description": "The run completed but at least one acceptance gate failed.", + "id": "rejected", + "predicate": { + "kind": "metric", + "metric": "acceptance_pass", + "operator": "lt", + "threshold": 1 + } + }, + { + "description": "The runner failed or did not emit its exact metric contract.", + "id": "execution_failure", + "predicate": { + "kind": "execution_failure" + } + } + ], + "primary_metric": "acceptance_pass", + "success_outcome": "accepted" + }, + "proposer": { + "id": "model://codex/lecore-qwen/research-director", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director" + }, + "rationale": "This converts the unresolved Qwen integration claims into one frozen, replayable, statistically gated run. Spectral filtering is a separate research control and is forbidden in this contract.", + "schema": "ilxyr.experiment.v1", + "security": { + "code_policy": "arbitrary", + "export_policy": "artifacts", + "weight_class": "public" + }, + "seeds": [ + 0 + ], + "title": "Qwen3.5 leCore layer-prepending acceptance" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/forecast-empirical.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/forecast-empirical.json new file mode 100644 index 0000000..81844bf --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/forecast-empirical.json @@ -0,0 +1,17 @@ +{ + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-empirical", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical", + "probabilities": { + "accepted": 0.2, + "execution_failure": 0.15, + "rejected": 0.65 + }, + "rationale": "Structural rehearsal passes, but no complete post-memory-fix real-Qwen install and powered acceptance run has succeeded yet.", + "schema": "ilxyr.forecast.v1", + "stake": 5 +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/forecast-mechanistic.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/forecast-mechanistic.json new file mode 100644 index 0000000..ff2b4de --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/forecast-mechanistic.json @@ -0,0 +1,17 @@ +{ + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "forecaster": { + "id": "model://codex/lecore-qwen/forecaster-mechanistic", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic", + "probabilities": { + "accepted": 0.15, + "execution_failure": 0.3, + "rejected": 0.55 + }, + "rationale": "The blank prepend has a strong identity construction, while full-checkpoint memory pressure and official multimodal reload remain material execution risks.", + "schema": "ilxyr.forecast.v1", + "stake": 5 +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/foundation.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/foundation.json new file mode 100644 index 0000000..2abde98 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/foundation.json @@ -0,0 +1,20 @@ +{ + "actor": { + "id": "model://codex/lecore-qwen/statistical-reviewer", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/statistical-reviewer" + }, + "body": "Per-token losses are serially correlated. The acceptance decision therefore uses paired installed-minus-original token NLLs, estimates an autocorrelation-aware moving-block length, and requires the upper 95 percent confidence bound to stay within a one-percent perplexity regression budget. Point estimates and independent-token bootstrap intervals are not acceptance evidence.", + "claims": [ + "At least the preregistered minimum number of paired token positions is measured.", + "The statistical gate is decided by the paired block interval rather than the point perplexity delta." + ], + "confidence": 0.95, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.foundation", + "input_refs": [ + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.hypothesis" + ], + "schema": "ilxyr.contribution.v1", + "stage": "mathematical_foundation", + "title": "Paired moving-block inference is required for token-loss comparisons" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/funding.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/funding.json new file mode 100644 index 0000000..6821a40 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/funding.json @@ -0,0 +1,11 @@ +{ + "compute_credits": 100, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "funder": { + "id": "service://lecore/qwen-experiment-generator", + "kind": "service" + }, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.funding", + "rationale": "Reserve one complete public-weight Qwen acceptance run under the frozen contract.", + "schema": "ilxyr.funding.v1" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/hypothesis.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/hypothesis.json new file mode 100644 index 0000000..19b7d22 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/hypothesis.json @@ -0,0 +1,19 @@ +{ + "actor": { + "id": "model://codex/lecore-qwen/research-director", + "kind": "model", + "model_ref": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director" + }, + "body": "The experimental layer-prepending installer can emit an ordinary Qwen3.5 checkpoint while preserving reference behavior within a preregistered paired confidence bound and retaining official text and vision-language execution.", + "claims": [ + "Pre-install leCore logits match the official reference implementation.", + "The installed checkpoint reloads and remains within the paired regression budget.", + "Official Transformers text generation and vision input both remain operational." + ], + "confidence": 0.35, + "id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.hypothesis", + "input_refs": [], + "schema": "ilxyr.contribution.v1", + "stage": "hypothesis", + "title": "Qwen3.5 can carry a reloadable leCore layer-prepending installation" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/project.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/project.json new file mode 100644 index 0000000..535876b --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/project/project.json @@ -0,0 +1,156 @@ +{ + "commands": [ + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance-v2/project/hypothesis.json" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance-v2/project/foundation.json" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance-v2/project/engineering-review.json" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance-v2/project/experiment-design.json" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "compile", + "", + "/opt/qwen35-acceptance-v2/project/experiment.json" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "forecast", + "", + "/opt/qwen35-acceptance-v2/project/forecast-empirical.json" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "forecast", + "", + "/opt/qwen35-acceptance-v2/project/forecast-mechanistic.json" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "fund", + "", + "/opt/qwen35-acceptance-v2/project/funding.json" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "admit", + "", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "run", + "", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "--execute" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "status", + "", + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance" + ], + [ + "/opt/qwen35-acceptance-v2/ilxyr/target/release/ilxyr", + "verify", + "" + ] + ], + "corpora": { + "evaluation": { + "bytes": 1213410, + "path": "/opt/qwen35-acceptance-v2/inputs/federalist-papers.txt", + "sha256": "a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + }, + "installation": { + "bytes": 2300089, + "path": "/opt/qwen35-acceptance-v2/lecore/REFERENCE.md", + "sha256": "d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f" + } + }, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "experiment_version": 2, + "installed_dir": "/opt/qwen35-acceptance-v2/installed-checkpoint", + "model_digest": "20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "model_files": [ + { + "bytes": 11544, + "path": "LICENSE", + "sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a" + }, + { + "bytes": 61705, + "path": "README.md", + "sha256": "87a163af54f32fa608a0f8d3ac67945c53dd2b4a7c96740b3d7fdc28e8458864" + }, + { + "bytes": 7755, + "path": "chat_template.jinja", + "sha256": "273d8e0e683b885071fb17e08d71e5f2a5ddfb5309756181681de4f5a1822d80" + }, + { + "bytes": 2907, + "path": "config.json", + "sha256": "b90b86f35c8e6925ef74ee04d0e758f0a845c83a42089ad82bbaa948de9b4204" + }, + { + "bytes": 3353259, + "path": "merges.txt", + "sha256": "a9d356d7bdf1ef4949e3e748e95b8e10ad9d4e2e838eddc38a0a7b6b94d1db8d" + }, + { + "bytes": 1746942600, + "path": "model.safetensors-00001-of-00001.safetensors", + "sha256": "04b1c301231dd422b8860db31311ab2721511346a32cb1e079c4c4e5f1fe4696" + }, + { + "bytes": 50900, + "path": "model.safetensors.index.json", + "sha256": "d8a08838a613b025eb7952ed9db11696213e57e76a375661ef5c12f9dd5dcf4e" + }, + { + "bytes": 390, + "path": "preprocessor_config.json", + "sha256": "27225450ac9c6529872ee1924fcb0962ff5634834f817040f444118116f4e516" + }, + { + "bytes": 12807982, + "path": "tokenizer.json", + "sha256": "5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42" + }, + { + "bytes": 16709, + "path": "tokenizer_config.json", + "sha256": "49e2b6e395f959f077f1e992b338919c0d4a9732fc6e613995e06557f843500c" + }, + { + "bytes": 385, + "path": "video_preprocessor_config.json", + "sha256": "7768af27c1fafa9cc9011c1dc20067e03f8915e03b63504550e11d5066986d13" + }, + { + "bytes": 6722759, + "path": "vocab.json", + "sha256": "ce99b4cb2983d118806ce0a8b777a35b093e2000a503ebde25853284c9dfa003" + } + ], + "schema": "lecore.ilxyr-project.v1", + "source_commit": "2f32151c6d44c80a7e2ff98c0696650cd1d9c6f1" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/publication-manifest.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/publication-manifest.json new file mode 100644 index 0000000..9f71b2b --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/publication-manifest.json @@ -0,0 +1,284 @@ +{ + "evidence_ref": "artifact://sha256/15885a3e5fbb01a504473e3076c25c9d5d938c13f96faa340ed9600a245845cc", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "files": [ + { + "bytes": 4020, + "media_type": "text/markdown", + "path": "README.md", + "sha256": "954c94e877052af688b1a3b40dc9a1c51a1835776578e9d639f35810d1fcbd4d" + }, + { + "bytes": 29381, + "media_type": "application/json", + "path": "evidence/evidence.in-toto.json", + "sha256": "327ef5cf6c2f37e9b8afdabc0b6ca5546493292717a1abb09a9bf8618d320d93" + }, + { + "bytes": 28212, + "media_type": "application/json", + "path": "evidence/evidence.native.json", + "sha256": "11d8c155cce37e70e2eb8241e437b6c7190b6a7cbbbc3ffe329bfbe33bed4d95" + }, + { + "bytes": 23559, + "media_type": "application/json", + "path": "evidence/evidence.ro-crate.json", + "sha256": "0aa5e25a67a0e0f6de62574097e05102bd1871ca4c9bfddf2335b2ceff38b50c" + }, + { + "bytes": 1324, + "media_type": "application/json", + "path": "evidence/exporter-provenance.json", + "sha256": "476e27c893316f3ad19552b8df987b4af734154a312ba9a4d8636e30fcb22895" + }, + { + "bytes": 97, + "media_type": "application/json", + "path": "ledger/config.json", + "sha256": "2301b853e1e01b8844970b90e72036bed57c580ded81a857ec7faf27f8deca66" + }, + { + "bytes": 9048, + "media_type": "application/octet-stream", + "path": "ledger/events.jsonl", + "sha256": "87808b4a4c1baa64e4388ea0b3c5e893d6f448324a74cb41c92a16cd8b52eb1e" + }, + { + "bytes": 1212, + "media_type": "application/json", + "path": "project/engineering-review.json", + "sha256": "64e07a438050586d7fe72c8f759c392dbc449be0da809b3cb169d52644c586d2" + }, + { + "bytes": 1394, + "media_type": "application/json", + "path": "project/experiment-design.json", + "sha256": "fe6c0dc84158519c298ecef5890a1aff44e5eae5be56915f656c1d7d1e8f47a3" + }, + { + "bytes": 8781, + "media_type": "application/json", + "path": "project/experiment.json", + "sha256": "e2a3e732d22442d43f4d76afbb83c224621afe4f8d206ed3baf8851bfee723c6" + }, + { + "bytes": 714, + "media_type": "application/json", + "path": "project/forecast-empirical.json", + "sha256": "e24492a9bf01f2b7dc790dc3474c4f2a53908f4bf0300241ae6c2707aa6315d5" + }, + { + "bytes": 751, + "media_type": "application/json", + "path": "project/forecast-mechanistic.json", + "sha256": "8519dcd3c068bd28de7c245438ad1b7dad2abecd7bb9e7c66c0010a0448c5909" + }, + { + "bytes": 1204, + "media_type": "application/json", + "path": "project/foundation.json", + "sha256": "17688757c9f1163fb374aabc0dac8da742750bcab997305f5e5d970d246d316c" + }, + { + "bytes": 464, + "media_type": "application/json", + "path": "project/funding.json", + "sha256": "60b60d1c86999e6e56a2895aef4f08449babceaa21559aab03feffbc02b675ed" + }, + { + "bytes": 993, + "media_type": "application/json", + "path": "project/hypothesis.json", + "sha256": "1a8a92cdb8f714328580194f1c559bd8487974f8aa22175e1afe5d9c83795a40" + }, + { + "bytes": 4888, + "media_type": "application/json", + "path": "project/project.json", + "sha256": "ed0d30744b9009d31523883ecfd6adf7dd9be8fd7b38e4a3f50ac1fdac7aa674" + }, + { + "bytes": 1958, + "media_type": "text/plain", + "path": "result/acceptance-artifacts/install.log", + "sha256": "c03b7b0e61a05ed2bedef887a1eedbe90b8d03320051d6edd019bcc5a440f229" + }, + { + "bytes": 650, + "media_type": "application/json", + "path": "result/admission-preflight.json", + "sha256": "a51912a4ad4689142813260d4dc506a8ee57735eb327c9bdbc135dfd41aebbd0" + }, + { + "bytes": 480, + "media_type": "application/json", + "path": "result/aws-instance-identity.json", + "sha256": "e635f72a56b251167f5c6634ee2e0770af8c0d9a7aa8a613ce8f3b1736cb8a47" + }, + { + "bytes": 45353, + "media_type": "text/plain", + "path": "result/cloud-init-output.log", + "sha256": "d9675cc57a7bd7aa4fa7d6247815deb2977448b73487f98e941ae22687667d6b" + }, + { + "bytes": 388, + "media_type": "application/json", + "path": "result/corpus-manifest.json", + "sha256": "2e3bbc256f95a2b6c68e3986e9e9d612a2eee6f16716387e94bb573edf274c54" + }, + { + "bytes": 4830, + "media_type": "text/plain", + "path": "result/environment.txt", + "sha256": "dff45a28f0c4cc5d55f326d04647610e8a90e239c400ff472bffb82aec7c7d4a" + }, + { + "bytes": 289, + "media_type": "application/json", + "path": "result/execution-summary.json", + "sha256": "322071f7abf3adb80b708207113101cd610ce00bb08f1e4187480b1def304940" + }, + { + "bytes": 222, + "media_type": "application/json", + "path": "result/generation.json", + "sha256": "14a11085c04d6acb9fad4e873ecec958c6ba865820ee6a964fe6daf377e34cb8" + }, + { + "bytes": 1813, + "media_type": "application/json", + "path": "result/ilxyr-admission.json", + "sha256": "03fa388faa765e3e7881a2cacb0ce01ec9e098563f4124864d92a6777ea52451" + }, + { + "bytes": 259, + "media_type": "text/plain", + "path": "result/ilxyr-commit.txt", + "sha256": "32873b519f7170a47a099930a325f4bb29d04021b822fa7f5387c94a9976add4" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-compile.json", + "sha256": "3f096ce40aa2a9db04acac937482fe11717f8d726df1aee0c3e6cea0df52c4cf" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-engineering-review.json", + "sha256": "4dbaa1a5f6b9886a3dbf270d58c6340d6b7a7fe192a345e7af83c9bad5ed7650" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-experiment-design.json", + "sha256": "658beda3484acb7136273b955d8377fdc255e389f049cdc8c73087dd9b4679c1" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-foundation.json", + "sha256": "d52ecd321c77ce82a55f5bc0add453b6c2cf1b74fb8bf53c8ec35d2d978d41ef" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-hypothesis.json", + "sha256": "2f43b8db274b224aff44c030db7391810e80a87893138225e8ef3a8fd3805443" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-forecast-empirical.json", + "sha256": "d7a4284a01c451fa91aa0990820a81352f77e905b7a798eaa3bac43297d14ead" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-forecast-mechanistic.json", + "sha256": "2ca73881259fd8e64597517b58b0b7115e1eb02772355f4cf525ba6f1cc7f8b8" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-funding.json", + "sha256": "849f195c6f0d60b873d512c143f1ceb9c26c06fbe9506de10fc318aaacf6d730" + }, + { + "bytes": 80, + "media_type": "application/json", + "path": "result/ilxyr-init.json", + "sha256": "2b53102e3ab266b14be6b330e7815ce35476f898845a92cbbb716b02c6be3b11" + }, + { + "bytes": 15638, + "media_type": "application/json", + "path": "result/ilxyr-run.json", + "sha256": "c20f0c2e63ba78855017293f58350742eac8056ba699a4370d3152fba2459c22" + }, + { + "bytes": 16118, + "media_type": "application/json", + "path": "result/ilxyr-status.json", + "sha256": "151d3d75656a2ed5c06e4a890fb7386e894208849c5fb4ead61d79cb4fc232b9" + }, + { + "bytes": 69, + "media_type": "application/json", + "path": "result/ilxyr-verify.json", + "sha256": "5e03d79a86e57f015be7b79666349b5813a1a702421883e8b58f3ed3a6a73663" + }, + { + "bytes": 3490, + "media_type": "application/json", + "path": "result/launch-manifest.json", + "sha256": "d1a1f255d2c12643f3e12336d85a6fce7441b5f39240cd5ee1554eef88c6e51d" + }, + { + "bytes": 267, + "media_type": "text/plain", + "path": "result/lecore-commit.txt", + "sha256": "33322be861cc97fdeca7f8acdd994cf2388169cf92a3c4c31f97240f79984ee1" + }, + { + "bytes": 0, + "media_type": "text/plain", + "path": "result/lecore-status.txt", + "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" + }, + { + "bytes": 2023, + "media_type": "application/json", + "path": "result/model-manifest.json", + "sha256": "f352ac2c17ee1bea38f70d7ab1975e3297954423cb74f5330383134e4b6c493a" + }, + { + "bytes": 588, + "media_type": "text/plain", + "path": "result/requirements-resolved.txt", + "sha256": "ffc5f6316f74e8e6c465b3845f6b4feb9e3b534160434299de9b05c809d7f647" + }, + { + "bytes": 172, + "media_type": "application/json", + "path": "result/runner-status.json", + "sha256": "3300be84544ed76b749a74228591664b5a18cb4131aba6fd52c775bb0843d583" + }, + { + "bytes": 41815, + "media_type": "text/plain", + "path": "result/user-data.log", + "sha256": "0bfd74c02c09a3dfef2ebae6abd3d8354e56be07c692bd275db5a02781ce6046" + } + ], + "ledger_verification": { + "events_checked": 16, + "objects_checked": 15, + "valid": true + }, + "resolved_outcome": "execution_failure", + "run_ref": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "schema": "lecore.qwen35-publication-manifest.v1" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/publication-receipt.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/publication-receipt.json new file mode 100644 index 0000000..435dcf0 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/publication-receipt.json @@ -0,0 +1,52 @@ +{ + "schema": "lecore.qwen35-arweave-publication-receipt.v1", + "published_at": "2026-08-13T05:00:28Z", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "resolved_outcome": "execution_failure", + "publisher": { + "service": "ArDrive Turbo signed free upload", + "sdk": "@ardrive/turbo-sdk@1.42.0", + "ownership": "The uploads used one-use in-memory Arweave identities. Their private keys were never persisted." + }, + "publication_manifest": { + "path": "publication-manifest.json", + "bytes": 9589, + "sha256": "21874555c7f9e73d20b9d11599c79236f8f40267ee975f703ffc053eb6b1f9ab", + "transaction_id": "l2Y6akCBi7igJoIQGyGCiDD3RGriYwoM_DvLIBkRiUE" + }, + "arweave_path_manifest": { + "transaction_id": "f35OgXuOoulk9EcndJBNWIgZ5tmaQ6DRzeINP7ARZTU", + "bytes": 4854, + "sha256": "7a18bb60bc9c75022705be94f80abb6ea7ae90508545e3971a7bf0e20b032a34", + "item_count": 46, + "gateway_url": "https://arweave.net/f35OgXuOoulk9EcndJBNWIgZ5tmaQ6DRzeINP7ARZTU/", + "index_url": "https://arweave.net/f35OgXuOoulk9EcndJBNWIgZ5tmaQ6DRzeINP7ARZTU/publication-manifest.json", + "confirmed_cache_url": "https://turbo-gateway.com/f35OgXuOoulk9EcndJBNWIgZ5tmaQ6DRzeINP7ARZTU/", + "owner": "CYVtYCPST9t7ziybtEpHSxncOEgCiofFatuiHcTFSjA" + }, + "selected_items": { + "native_evidence": { + "transaction_id": "G0rTDs3o1sbMHCCY2riLzLg-o6mkb2ByUcJ4PTMPilc", + "sha256": "11d8c155cce37e70e2eb8241e437b6c7190b6a7cbbbc3ffe329bfbe33bed4d95" + }, + "ledger_events": { + "transaction_id": "99RUKhUlaM6Nags0ev35rICJ4o3WO6uLJzy4bIS7Yzs", + "sha256": "87808b4a4c1baa64e4388ea0b3c5e893d6f448324a74cb41c92a16cd8b52eb1e" + }, + "ledger_verification": { + "transaction_id": "Cqv57F936hf4TQ3l35MZXQu7vX324hDQSEfshxZO_lA", + "sha256": "5e03d79a86e57f015be7b79666349b5813a1a702421883e8b58f3ed3a6a73663" + } + }, + "confirmation": { + "status_service": "https://upload.ardrive.io/v1/tx/{transaction_id}/status", + "transactions_confirmed": 47, + "total_winc_charged": 0, + "public_cache_verified_at": "2026-08-13T05:00:28Z", + "public_gateway_verified_at": "2026-08-13T05:04:19Z", + "verified_downloads": { + "publication-manifest.json": "21874555c7f9e73d20b9d11599c79236f8f40267ee975f703ffc053eb6b1f9ab", + "ledger/events.jsonl": "87808b4a4c1baa64e4388ea0b3c5e893d6f448324a74cb41c92a16cd8b52eb1e" + } + } +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/acceptance-artifacts/install.log b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/acceptance-artifacts/install.log new file mode 100644 index 0000000..49d688f --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/acceptance-artifacts/install.log @@ -0,0 +1,29 @@ +[load] /opt/qwen35-acceptance-v2/model + sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct + hidden 1024 | 24 layers | vocab 248320 | BF16, F32 + inferred from tensors alone: 24 layers, hidden 1024, tied=True (confidence 1.00) + tokenizer: model tokenizer + corpus: /opt/qwen35-acceptance-v2/lecore/REFERENCE.md (2300 KB) + memory: 128 registers (of 1024 dimensions) and 0 searchable passages + NOTE: this tokenizer uses every vocabulary row, so there is nowhere to put a search index -- skipping it. Registers and everything else still install. + +[install] leCore into the weights + hardware: cpu (disabled) + prepend ok 2 layers added, drift 0.000e+00 (relative 0.000e+00, first at token 0) -- bit-identical + registers ok 128 reserved slots, 896 dims left to the model + hrnn_channel FAIL ValueError: cannot reshape array of size 524288 into shape (64,20,512) [heads k=16 v=16, kdim=128 vdim=128, hidden=1024, 26 layers -- the ladder is optional, continuing without it] + nullspace_guard ok 994 of 1024 dims safe to write (true null space) + router ok layer 0, held-out accuracy 92% + self_write FAIL novelty readout r=0.726, finds 40% of the top decile + state_track ok 4 of 128 registers reserved as no-decay state slots + improvement FAIL no step improved perplexity without making generation more repetitive + boot_record ok row 248319, perplexity +0.000% + +[wrote] /opt/qwen35-acceptance-v2/installed-checkpoint (2104.2 MB, BF16, F32) + +[verify] reloading from disk + sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct + 26 layers | perplexity 20.3023 (was 20.3023) | boots as 'leCore' + verdict WORSE (+0.000%) | repetition 0.96 -> 0.96 + +Next: assess.bat (or: python assimilation/galvatron.py /opt/qwen35-acceptance-v2/installed-checkpoint --assess out.npz) diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/admission-preflight.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/admission-preflight.json new file mode 100644 index 0000000..709ee8d --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/admission-preflight.json @@ -0,0 +1,23 @@ +{ + "checks": { + "checker_version": true, + "complete_top_level_model_manifest": true, + "corpora_distinct": true, + "evaluation_corpus": true, + "experiment_version": true, + "installation_corpus": true, + "launch_attempt": true, + "launch_corpora": true, + "launch_method": true, + "launch_model": true, + "launch_source": true, + "minimum_tokens": true, + "source_commit": true, + "spectral_disabled": true, + "timeout": true, + "tokenizer": true, + "venv_interpreter": true, + "weights": true + }, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/aws-instance-identity.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/aws-instance-identity.json new file mode 100644 index 0000000..7930faa --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/aws-instance-identity.json @@ -0,0 +1,17 @@ +{ + "accountId" : "022118847419", + "architecture" : "x86_64", + "availabilityZone" : "us-east-1a", + "billingProducts" : null, + "devpayProductCodes" : null, + "marketplaceProductCodes" : null, + "imageId" : "ami-07a5b367e8dc8bd92", + "instanceId" : "i-059aab8bf1d3d87c9", + "instanceType" : "r7i.4xlarge", + "kernelId" : null, + "pendingTime" : "2026-08-13T00:24:43Z", + "privateIp" : "172.31.11.240", + "ramdiskId" : null, + "region" : "us-east-1", + "version" : "2017-09-30" +} \ No newline at end of file diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/cloud-init-output.log b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/cloud-init-output.log new file mode 100644 index 0000000..e5330f0 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/cloud-init-output.log @@ -0,0 +1,526 @@ +Cloud-init v. 22.2.2 running 'init' at Thu, 13 Aug 2026 00:24:54 +0000. Up 6.89 seconds. +ci-info: ++++++++++++++++++++++++++++++++++++++Net device info+++++++++++++++++++++++++++++++++++++++ +ci-info: +---------+------+----------------------------+---------------+--------+-------------------+ +ci-info: | Device | Up | Address | Mask | Scope | Hw-Address | +ci-info: +---------+------+----------------------------+---------------+--------+-------------------+ +ci-info: | enp39s0 | True | 172.31.11.240 | 255.255.240.0 | global | 02:2c:9d:36:d7:db | +ci-info: | enp39s0 | True | fe80::2c:9dff:fe36:d7db/64 | . | link | 02:2c:9d:36:d7:db | +ci-info: | lo | True | 127.0.0.1 | 255.0.0.0 | host | . | +ci-info: | lo | True | ::1/128 | . | host | . | +ci-info: +---------+------+----------------------------+---------------+--------+-------------------+ +ci-info: +++++++++++++++++++++++++++++Route IPv4 info++++++++++++++++++++++++++++++ +ci-info: +-------+-------------+------------+-----------------+-----------+-------+ +ci-info: | Route | Destination | Gateway | Genmask | Interface | Flags | +ci-info: +-------+-------------+------------+-----------------+-----------+-------+ +ci-info: | 0 | 0.0.0.0 | 172.31.0.1 | 0.0.0.0 | enp39s0 | UG | +ci-info: | 1 | 172.31.0.0 | 0.0.0.0 | 255.255.240.0 | enp39s0 | U | +ci-info: | 2 | 172.31.0.1 | 0.0.0.0 | 255.255.255.255 | enp39s0 | UH | +ci-info: | 3 | 172.31.0.2 | 0.0.0.0 | 255.255.255.255 | enp39s0 | UH | +ci-info: +-------+-------------+------------+-----------------+-----------+-------+ +ci-info: +++++++++++++++++++Route IPv6 info+++++++++++++++++++ +ci-info: +-------+-------------+---------+-----------+-------+ +ci-info: | Route | Destination | Gateway | Interface | Flags | +ci-info: +-------+-------------+---------+-----------+-------+ +ci-info: | 0 | fe80::/64 | :: | enp39s0 | U | +ci-info: | 2 | local | :: | enp39s0 | U | +ci-info: | 3 | multicast | :: | enp39s0 | U | +ci-info: +-------+-------------+---------+-----------+-------+ +Generating public/private ed25519 key pair. +Your identification has been saved in /etc/ssh/ssh_host_ed25519_key +Your public key has been saved in /etc/ssh/ssh_host_ed25519_key.pub +The key fingerprint is: +SHA256:QqbwUo68oPPPuPBoPOyS52R4uYGik8nX35m+Qaw0BwQ root@ip-172-31-11-240.ec2.internal +The key's randomart image is: ++--[ED25519 256]--+ +| E.. | +| . | +| . . o. | +| . * + o | +|. + + .oS+ | +|.+ + ..= | +|@+B . . . | +|O/.* . . + | +|*=Xoo ...*. | ++----[SHA256]-----+ +Generating public/private ecdsa key pair. +Your identification has been saved in /etc/ssh/ssh_host_ecdsa_key +Your public key has been saved in /etc/ssh/ssh_host_ecdsa_key.pub +The key fingerprint is: +SHA256:fJjI0Olu+1j3h8J9r+XB+c/MLPdhXC2TO4HJukpFwqw root@ip-172-31-11-240.ec2.internal +The key's randomart image is: ++---[ECDSA 256]---+ +| | +| . + | +| . o + . | +| + + = . o ..| +| E S o + = o| +| . o . o=o| +| o o.o. .oB.| +| . = .ooo =O=| +| o.o....o.=&| ++----[SHA256]-----+ +Cloud-init v. 22.2.2 running 'modules:config' at Thu, 13 Aug 2026 00:24:55 +0000. Up 8.29 seconds. +Cloud-init v. 22.2.2 running 'modules:final' at Thu, 13 Aug 2026 00:24:55 +0000. Up 8.78 seconds. +Shutdown scheduled for Thu 2026-08-13 08:24:56 UTC, use 'shutdown -c' to cancel. +Amazon Linux 2023 repository 80 MB/s | 75 MB 00:00 +Amazon Linux 2023 Kernel Livepatch repository 651 kB/s | 69 kB 00:00 +Package tar-2:1.34-1.amzn2023.0.4.x86_64 is already installed. +Package gzip-1.12-1.amzn2023.0.1.x86_64 is already installed. +Dependencies resolved. +================================================================================ + Package Arch Version Repository Size +================================================================================ +Installing: + cargo x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 8.2 M + git x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 53 k + python3.12 x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 27 k + python3.12-pip noarch 23.2.1-4.amzn2023.0.10 amazonlinux 2.8 M + rust x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 62 M +Installing dependencies: + annobin-docs noarch 12.69-1.amzn2023.0.1 amazonlinux 93 k + annobin-plugin-gcc x86_64 12.69-1.amzn2023.0.1 amazonlinux 974 k + cpp x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 11 M + gc x86_64 8.0.4-5.amzn2023.0.2 amazonlinux 105 k + gcc x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 36 M + gcc-plugin-annobin x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 38 k + git-core x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 4.9 M + git-core-doc noarch 2.50.1-1.amzn2023.0.1 amazonlinux 2.8 M + glibc-devel x86_64 2.34-231.amzn2023.0.5 amazonlinux 33 k + glibc-headers-x86 noarch 2.34-231.amzn2023.0.5 amazonlinux 439 k + guile22 x86_64 2.2.7-2.amzn2023.0.3 amazonlinux 6.4 M + kernel-headers x86_64 1:6.1.177-224.371.amzn2023 amazonlinux 1.4 M + libmpc x86_64 1.2.1-2.amzn2023.0.2 amazonlinux 62 k + libtool-ltdl x86_64 2.4.7-1.amzn2023.0.3 amazonlinux 38 k + libxcrypt-devel x86_64 4.4.33-7.amzn2023 amazonlinux 32 k + make x86_64 1:4.3-5.amzn2023.0.2 amazonlinux 534 k + mpdecimal x86_64 2.5.1-3.amzn2023.0.3 amazonlinux 101 k + perl-Error noarch 1:0.17030-2.amzn2023.0.1 amazonlinux 42 k + perl-File-Find noarch 1.37-477.amzn2023.0.9 amazonlinux 26 k + perl-Git noarch 2.50.1-1.amzn2023.0.1 amazonlinux 41 k + perl-TermReadKey x86_64 2.38-9.amzn2023.0.3 amazonlinux 36 k + perl-lib x86_64 0.65-477.amzn2023.0.9 amazonlinux 15 k + python3.12-libs x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 9.2 M + python3.12-pip-wheel noarch 23.2.1-4.amzn2023.0.10 amazonlinux 1.5 M + rust-std-static x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 39 M +Installing weak dependencies: + python3.12-setuptools noarch 68.2.2-4.amzn2023.0.3 amazonlinux 1.4 M + +Transaction Summary +================================================================================ +Install 31 Packages + +Total download size: 189 M +Installed size: 700 M +Downloading Packages: +(1/31): annobin-docs-12.69-1.amzn2023.0.1.noarc 2.5 MB/s | 93 kB 00:00 +(2/31): annobin-plugin-gcc-12.69-1.amzn2023.0.1 24 MB/s | 974 kB 00:00 +(3/31): gc-8.0.4-5.amzn2023.0.2.x86_64.rpm 4.0 MB/s | 105 kB 00:00 +(4/31): cpp-11.5.0-5.amzn2023.0.5.x86_64.rpm 79 MB/s | 11 MB 00:00 +(5/31): gcc-plugin-annobin-11.5.0-5.amzn2023.0. 1.2 MB/s | 38 kB 00:00 +(6/31): git-2.50.1-1.amzn2023.0.1.x86_64.rpm 1.7 MB/s | 53 kB 00:00 +(7/31): git-core-2.50.1-1.amzn2023.0.1.x86_64.r 74 MB/s | 4.9 MB 00:00 +(8/31): git-core-doc-2.50.1-1.amzn2023.0.1.noar 56 MB/s | 2.8 MB 00:00 +(9/31): glibc-devel-2.34-231.amzn2023.0.5.x86_6 941 kB/s | 33 kB 00:00 +(10/31): cargo-1.97.0-1.amzn2023.0.1.x86_64.rpm 19 MB/s | 8.2 MB 00:00 +(11/31): glibc-headers-x86-2.34-231.amzn2023.0. 12 MB/s | 439 kB 00:00 +(12/31): gcc-11.5.0-5.amzn2023.0.5.x86_64.rpm 78 MB/s | 36 MB 00:00 +(13/31): kernel-headers-6.1.177-224.371.amzn202 13 MB/s | 1.4 MB 00:00 +(14/31): guile22-2.2.7-2.amzn2023.0.3.x86_64.rp 50 MB/s | 6.4 MB 00:00 +(15/31): libmpc-1.2.1-2.amzn2023.0.2.x86_64.rpm 2.2 MB/s | 62 kB 00:00 +(16/31): libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_ 1.6 MB/s | 38 kB 00:00 +(17/31): libxcrypt-devel-4.4.33-7.amzn2023.x86_ 1.2 MB/s | 32 kB 00:00 +(18/31): mpdecimal-2.5.1-3.amzn2023.0.3.x86_64. 4.3 MB/s | 101 kB 00:00 +(19/31): make-4.3-5.amzn2023.0.2.x86_64.rpm 19 MB/s | 534 kB 00:00 +(20/31): perl-Error-0.17030-2.amzn2023.0.1.noar 1.6 MB/s | 42 kB 00:00 +(21/31): perl-File-Find-1.37-477.amzn2023.0.9.n 1.0 MB/s | 26 kB 00:00 +(22/31): perl-Git-2.50.1-1.amzn2023.0.1.noarch. 1.5 MB/s | 41 kB 00:00 +(23/31): perl-lib-0.65-477.amzn2023.0.9.x86_64. 536 kB/s | 15 kB 00:00 +(24/31): python3.12-3.12.13-2.amzn2023.0.5.x86_ 1.0 MB/s | 27 kB 00:00 +(25/31): perl-TermReadKey-2.38-9.amzn2023.0.3.x 1.0 MB/s | 36 kB 00:00 +(26/31): python3.12-pip-23.2.1-4.amzn2023.0.10. 70 MB/s | 2.8 MB 00:00 +(27/31): python3.12-pip-wheel-23.2.1-4.amzn2023 35 MB/s | 1.5 MB 00:00 +(28/31): python3.12-setuptools-68.2.2-4.amzn202 27 MB/s | 1.4 MB 00:00 +(29/31): python3.12-libs-3.12.13-2.amzn2023.0.5 77 MB/s | 9.2 MB 00:00 +(30/31): rust-std-static-1.97.0-1.amzn2023.0.1. 72 MB/s | 39 MB 00:00 +(31/31): rust-1.97.0-1.amzn2023.0.1.x86_64.rpm 73 MB/s | 62 MB 00:00 +-------------------------------------------------------------------------------- +Total 121 MB/s | 189 MB 00:01 +Running transaction check +Transaction check succeeded. +Running transaction test +Transaction test succeeded. +Running transaction + Preparing : 1/1 + Installing : libmpc-1.2.1-2.amzn2023.0.2.x86_64 1/31 + Installing : git-core-2.50.1-1.amzn2023.0.1.x86_64 2/31 + Installing : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 3/31 + Installing : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Installing : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 5/31 + Installing : perl-lib-0.65-477.amzn2023.0.9.x86_64 6/31 + Installing : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 7/31 + Installing : perl-File-Find-1.37-477.amzn2023.0.9.noarch 8/31 + Installing : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 9/31 + Installing : perl-Git-2.50.1-1.amzn2023.0.1.noarch 10/31 + Installing : git-2.50.1-1.amzn2023.0.1.x86_64 11/31 + Installing : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 12/31 + Installing : python3.12-3.12.13-2.amzn2023.0.5.x86_64 13/31 + Installing : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 14/31 + Installing : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 15/31 + Installing : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Installing : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 17/31 + Installing : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 18/31 + Installing : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 19/31 + Installing : glibc-devel-2.34-231.amzn2023.0.5.x86_64 20/31 + Installing : gc-8.0.4-5.amzn2023.0.2.x86_64 21/31 + Installing : guile22-2.2.7-2.amzn2023.0.3.x86_64 22/31 + Installing : make-1:4.3-5.amzn2023.0.2.x86_64 23/31 + Installing : gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Running scriptlet: gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Installing : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 25/31 + Installing : rust-1.97.0-1.amzn2023.0.1.x86_64 26/31 + Installing : annobin-docs-12.69-1.amzn2023.0.1.noarch 27/31 + Installing : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Running scriptlet: annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Installing : cargo-1.97.0-1.amzn2023.0.1.x86_64 29/31 + Installing : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 30/31 + Installing : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Running scriptlet: python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Verifying : annobin-docs-12.69-1.amzn2023.0.1.noarch 1/31 + Verifying : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 2/31 + Verifying : cargo-1.97.0-1.amzn2023.0.1.x86_64 3/31 + Verifying : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Verifying : gc-8.0.4-5.amzn2023.0.2.x86_64 5/31 + Verifying : gcc-11.5.0-5.amzn2023.0.5.x86_64 6/31 + Verifying : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 7/31 + Verifying : git-2.50.1-1.amzn2023.0.1.x86_64 8/31 + Verifying : git-core-2.50.1-1.amzn2023.0.1.x86_64 9/31 + Verifying : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 10/31 + Verifying : glibc-devel-2.34-231.amzn2023.0.5.x86_64 11/31 + Verifying : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 12/31 + Verifying : guile22-2.2.7-2.amzn2023.0.3.x86_64 13/31 + Verifying : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 14/31 + Verifying : libmpc-1.2.1-2.amzn2023.0.2.x86_64 15/31 + Verifying : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Verifying : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 17/31 + Verifying : make-1:4.3-5.amzn2023.0.2.x86_64 18/31 + Verifying : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 19/31 + Verifying : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 20/31 + Verifying : perl-File-Find-1.37-477.amzn2023.0.9.noarch 21/31 + Verifying : perl-Git-2.50.1-1.amzn2023.0.1.noarch 22/31 + Verifying : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 23/31 + Verifying : perl-lib-0.65-477.amzn2023.0.9.x86_64 24/31 + Verifying : python3.12-3.12.13-2.amzn2023.0.5.x86_64 25/31 + Verifying : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 26/31 + Verifying : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 27/31 + Verifying : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 28/31 + Verifying : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 29/31 + Verifying : rust-1.97.0-1.amzn2023.0.1.x86_64 30/31 + Verifying : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 31/31 + +Installed: + annobin-docs-12.69-1.amzn2023.0.1.noarch + annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 + cargo-1.97.0-1.amzn2023.0.1.x86_64 + cpp-11.5.0-5.amzn2023.0.5.x86_64 + gc-8.0.4-5.amzn2023.0.2.x86_64 + gcc-11.5.0-5.amzn2023.0.5.x86_64 + gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 + git-2.50.1-1.amzn2023.0.1.x86_64 + git-core-2.50.1-1.amzn2023.0.1.x86_64 + git-core-doc-2.50.1-1.amzn2023.0.1.noarch + glibc-devel-2.34-231.amzn2023.0.5.x86_64 + glibc-headers-x86-2.34-231.amzn2023.0.5.noarch + guile22-2.2.7-2.amzn2023.0.3.x86_64 + kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 + libmpc-1.2.1-2.amzn2023.0.2.x86_64 + libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 + libxcrypt-devel-4.4.33-7.amzn2023.x86_64 + make-1:4.3-5.amzn2023.0.2.x86_64 + mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 + perl-Error-1:0.17030-2.amzn2023.0.1.noarch + perl-File-Find-1.37-477.amzn2023.0.9.noarch + perl-Git-2.50.1-1.amzn2023.0.1.noarch + perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 + perl-lib-0.65-477.amzn2023.0.9.x86_64 + python3.12-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-pip-23.2.1-4.amzn2023.0.10.noarch + python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch + python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch + rust-1.97.0-1.amzn2023.0.1.x86_64 + rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 + +Complete! +Cloning into '/opt/qwen35-acceptance-v2/lecore'... +Updating files: 0% (2/903) Updating files: 1% (10/903) Updating files: 2% (19/903) Updating files: 3% (28/903) Updating files: 4% (37/903) Updating files: 5% (46/903) Updating files: 6% (55/903) Updating files: 7% (64/903) Updating files: 8% (73/903) Updating files: 9% (82/903) Updating files: 10% (91/903) Updating files: 11% (100/903) Updating files: 12% (109/903) Updating files: 13% (118/903) Updating files: 14% (127/903) Updating files: 15% (136/903) Updating files: 16% (145/903) Updating files: 17% (154/903) Updating files: 18% (163/903) Updating files: 19% (172/903) Updating files: 20% (181/903) Updating files: 21% (190/903) Updating files: 22% (199/903) Updating files: 23% (208/903) Updating files: 24% (217/903) Updating files: 25% (226/903) Updating files: 26% (235/903) Updating files: 27% (244/903) Updating files: 28% (253/903) Updating files: 29% (262/903) Updating files: 30% (271/903) Updating files: 31% (280/903) Updating files: 32% (289/903) Updating files: 33% (298/903) Updating files: 34% (308/903) Updating files: 35% (317/903) Updating files: 36% (326/903) Updating files: 37% (335/903) Updating files: 38% (344/903) Updating files: 39% (353/903) Updating files: 40% (362/903) Updating files: 41% (371/903) Updating files: 42% (380/903) Updating files: 43% (389/903) Updating files: 44% (398/903) Updating files: 45% (407/903) Updating files: 46% (416/903) Updating files: 47% (425/903) Updating files: 48% (434/903) Updating files: 49% (443/903) Updating files: 50% (452/903) Updating files: 51% (461/903) Updating files: 52% (470/903) Updating files: 53% (479/903) Updating files: 54% (488/903) Updating files: 55% (497/903) Updating files: 56% (506/903) Updating files: 57% (515/903) Updating files: 58% (524/903) Updating files: 59% (533/903) Updating files: 60% (542/903) Updating files: 61% (551/903) Updating files: 62% (560/903) Updating files: 63% (569/903) Updating files: 64% (578/903) Updating files: 65% (587/903) Updating files: 66% (596/903) Updating files: 67% (606/903) Updating files: 68% (615/903) Updating files: 69% (624/903) Updating files: 70% (633/903) Updating files: 71% (642/903) Updating files: 72% (651/903) Updating files: 73% (660/903) Updating files: 74% (669/903) Updating files: 75% (678/903) Updating files: 76% (687/903) Updating files: 77% (696/903) Updating files: 78% (705/903) Updating files: 79% (714/903) Updating files: 80% (723/903) Updating files: 81% (732/903) Updating files: 82% (741/903) Updating files: 83% (750/903) Updating files: 84% (759/903) Updating files: 85% (768/903) Updating files: 86% (777/903) Updating files: 87% (786/903) Updating files: 88% (795/903) Updating files: 89% (804/903) Updating files: 90% (813/903) Updating files: 91% (822/903) Updating files: 92% (831/903) Updating files: 93% (840/903) Updating files: 94% (849/903) Updating files: 95% (858/903) Updating files: 96% (867/903) Updating files: 97% (876/903) Updating files: 98% (885/903) Updating files: 99% (894/903) Updating files: 100% (903/903) Updating files: 100% (903/903), done. +HEAD is now at 2f32151 experiments: version formal qwen attempts +/opt/qwen35-acceptance-v2/lecore/REFERENCE.md: OK +/opt/qwen35-acceptance-v2/inputs/federalist-papers.txt: OK +Cloning into '/opt/qwen35-acceptance-v2/ilxyr'... +HEAD is now at e92382f Publish EXP-006 experiment evidence + Updating crates.io index + Downloading crates ... + Downloaded serde v1.0.229 + Downloaded base64 v0.22.1 + Downloaded rustc_version v0.4.1 + Downloaded memchr v2.8.3 + Downloaded ed25519-dalek v2.2.0 + Downloaded quote v1.0.46 + Downloaded serde_core v1.0.229 + Downloaded curve25519-dalek v4.1.3 + Downloaded cfg-if v1.0.4 + Downloaded digest v0.10.7 + Downloaded block-buffer v0.10.4 + Downloaded getrandom v0.2.17 + Downloaded proc-macro2 v1.0.106 + Downloaded curve25519-dalek-derive v0.1.1 + Downloaded generic-array v0.14.7 + Downloaded itoa v1.0.18 + Downloaded rand_core v0.6.4 + Downloaded version_check v0.9.5 + Downloaded crypto-common v0.1.7 + Downloaded serde_json v1.0.150 + Downloaded spki v0.7.3 + Downloaded zmij v1.0.23 + Downloaded base64ct v1.8.3 + Downloaded sha2 v0.10.9 + Downloaded const-oid v0.9.6 + Downloaded semver v1.0.28 + Downloaded serde_derive v1.0.229 + Downloaded typenum v1.20.1 + Downloaded unicode-ident v1.0.24 + Downloaded syn v2.0.119 + Downloaded der v0.7.10 + Downloaded subtle v2.6.1 + Downloaded pkcs8 v0.10.2 + Downloaded ed25519 v2.2.3 + Downloaded syn v3.0.0 + Downloaded zeroize v1.9.0 + Downloaded signature v2.2.0 + Downloaded cpufeatures v0.2.17 + Downloaded libc v0.2.186 + Compiling proc-macro2 v1.0.106 + Compiling version_check v0.9.5 + Compiling unicode-ident v1.0.24 + Compiling quote v1.0.46 + Compiling typenum v1.20.1 + Compiling semver v1.0.28 + Compiling serde_core v1.0.229 + Compiling cpufeatures v0.2.17 + Compiling zmij v1.0.23 + Compiling cfg-if v1.0.4 + Compiling subtle v2.6.1 + Compiling serde_json v1.0.150 + Compiling serde v1.0.229 + Compiling signature v2.2.0 + Compiling zeroize v1.9.0 + Compiling memchr v2.8.3 + Compiling itoa v1.0.18 + Compiling base64 v0.22.1 + Compiling ed25519 v2.2.3 + Compiling rustc_version v0.4.1 + Compiling generic-array v0.14.7 + Compiling curve25519-dalek v4.1.3 + Compiling syn v2.0.119 + Compiling syn v3.0.0 + Compiling block-buffer v0.10.4 + Compiling crypto-common v0.1.7 + Compiling digest v0.10.7 + Compiling sha2 v0.10.9 + Compiling serde_derive v1.0.229 + Compiling curve25519-dalek-derive v0.1.1 + Compiling ed25519-dalek v2.2.0 + Compiling ilxyr-core v0.1.0 (/opt/qwen35-acceptance-v2/ilxyr/crates/ilxyr-core) + Compiling ilxyr-cli v0.1.0 (/opt/qwen35-acceptance-v2/ilxyr/crates/ilxyr-cli) + Finished `release` profile [optimized] target(s) in 20.15s +Looking in indexes: https://download.pytorch.org/whl/cpu +Collecting torch==2.11.0 + Obtaining dependency information for torch==2.11.0 from https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata + Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata (29 kB) +Collecting filelock (from torch==2.11.0) + Obtaining dependency information for filelock from https://files.pythonhosted.org/packages/81/47/dd9a212ef6e343a6857485ffe25bba537304f1913bdbed446a23f7f592e1/filelock-3.29.0-py3-none-any.whl.metadata + Downloading filelock-3.29.0-py3-none-any.whl.metadata (2.0 kB) +Collecting typing-extensions>=4.10.0 (from torch==2.11.0) + Obtaining dependency information for typing-extensions>=4.10.0 from https://files.pythonhosted.org/packages/18/67/36e9267722cc04a6b9f15c7f3441c2363321a3ea07da7ae0c0707beb2a9c/typing_extensions-4.15.0-py3-none-any.whl.metadata + Downloading typing_extensions-4.15.0-py3-none-any.whl.metadata (3.3 kB) +Collecting setuptools<82 (from torch==2.11.0) + Obtaining dependency information for setuptools<82 from https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata (6.6 kB) +Collecting sympy>=1.13.3 (from torch==2.11.0) + Obtaining dependency information for sympy>=1.13.3 from https://files.pythonhosted.org/packages/a2/09/77d55d46fd61b4a135c444fc97158ef34a095e5681d0a6c10b75bf356191/sympy-1.14.0-py3-none-any.whl.metadata + Downloading sympy-1.14.0-py3-none-any.whl.metadata (12 kB) +Collecting networkx>=2.5.1 (from torch==2.11.0) + Obtaining dependency information for networkx>=2.5.1 from https://files.pythonhosted.org/packages/9e/c9/b2622292ea83fbb4ec318f5b9ab867d0a28ab43c5717bb85b0a5f6b3b0a4/networkx-3.6.1-py3-none-any.whl.metadata + Downloading networkx-3.6.1-py3-none-any.whl.metadata (6.8 kB) +Collecting jinja2 (from torch==2.11.0) + Obtaining dependency information for jinja2 from https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata (2.9 kB) +Collecting fsspec>=0.8.5 (from torch==2.11.0) + Obtaining dependency information for fsspec>=0.8.5 from https://files.pythonhosted.org/packages/d5/0c/043d5e551459da400957a1395e0febbf771446ff34291afcbe3d8be2a279/fsspec-2026.4.0-py3-none-any.whl.metadata + Downloading fsspec-2026.4.0-py3-none-any.whl.metadata (10 kB) +Collecting mpmath<1.4,>=1.1.0 (from sympy>=1.13.3->torch==2.11.0) + Obtaining dependency information for mpmath<1.4,>=1.1.0 from https://files.pythonhosted.org/packages/43/e3/7d92a15f894aa0c9c4b49b8ee9ac9850d6e63b03c9c32c0367a13ae62209/mpmath-1.3.0-py3-none-any.whl.metadata + Downloading mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) +Collecting MarkupSafe>=2.0 (from jinja2->torch==2.11.0) + Obtaining dependency information for MarkupSafe>=2.0 from https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.7 kB) +Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl (190.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 190.3/190.3 MB 14.0 MB/s eta 0:00:00 +Downloading fsspec-2026.4.0-py3-none-any.whl (203 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 203.4/203.4 kB 27.9 MB/s eta 0:00:00 +Downloading networkx-3.6.1-py3-none-any.whl (2.1 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 2.1/2.1 MB 152.6 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl (1.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.3/1.3 MB 196.3 MB/s eta 0:00:00 +Downloading sympy-1.14.0-py3-none-any.whl (6.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.3/6.3 MB 209.5 MB/s eta 0:00:00 +Downloading typing_extensions-4.15.0-py3-none-any.whl (44 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 44.6/44.6 kB 15.4 MB/s eta 0:00:00 +Downloading filelock-3.29.0-py3-none-any.whl (39 kB) +Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl (134 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 134.9/134.9 kB 42.4 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (22 kB) +Downloading mpmath-1.3.0-py3-none-any.whl (536 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 536.2/536.2 kB 144.5 MB/s eta 0:00:00 +Installing collected packages: mpmath, typing-extensions, sympy, setuptools, networkx, MarkupSafe, fsspec, filelock, jinja2, torch +Successfully installed MarkupSafe-3.0.3 filelock-3.29.0 fsspec-2026.4.0 jinja2-3.1.6 mpmath-1.3.0 networkx-3.6.1 setuptools-78.1.0 sympy-1.14.0 torch-2.11.0+cpu typing-extensions-4.15.0 +Collecting numpy==2.4.6 + Obtaining dependency information for numpy==2.4.6 from https://files.pythonhosted.org/packages/99/60/14115e6364fa676c5397c2ad3004e527e9aa487abf5d0706ec81bbd08529/numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (6.6 kB) +Collecting pillow==12.3.0 + Obtaining dependency information for pillow==12.3.0 from https://files.pythonhosted.org/packages/84/21/a35af28dcc61f37ed850a2d64c65c701321dfbf25085e469d5559360cbbf/pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (9.1 kB) +Collecting safetensors==0.8.0 + Obtaining dependency information for safetensors==0.8.0 from https://files.pythonhosted.org/packages/28/50/f203ff3a3ddfe19308efc83c5a3a29ed02bf786732ec35e68bf9162f3365/safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (4.2 kB) +Collecting huggingface-hub==1.27.0 + Obtaining dependency information for huggingface-hub==1.27.0 from https://files.pythonhosted.org/packages/de/d8/95b735e183957c1f26d94c52977f09d466d55119cbbc1558ea4975e4c216/huggingface_hub-1.27.0-py3-none-any.whl.metadata + Downloading huggingface_hub-1.27.0-py3-none-any.whl.metadata (16 kB) +Collecting transformers==5.14.0 + Obtaining dependency information for transformers==5.14.0 from https://files.pythonhosted.org/packages/19/3b/a0a807b957c5f8f2075a8740f597a021420583bb0aa68ff6432d74a530d4/transformers-5.14.0-py3-none-any.whl.metadata + Downloading transformers-5.14.0-py3-none-any.whl.metadata (32 kB) +Collecting click<9.0.0,>=8.4.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for click<9.0.0,>=8.4.2 from https://files.pythonhosted.org/packages/fb/e2/79c688af8b210d232694e31e59da9f6ec747bae31c3f5946e4e9b98860d5/click-8.4.2-py3-none-any.whl.metadata + Downloading click-8.4.2-py3-none-any.whl.metadata (2.6 kB) +Requirement already satisfied: filelock>=3.10.0 in /opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (3.29.0) +Requirement already satisfied: fsspec>=2023.5.0 in /opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (2026.4.0) +Collecting hf-xet<2.0.0,>=1.5.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for hf-xet<2.0.0,>=1.5.2 from https://files.pythonhosted.org/packages/67/4e/a28359bf1c1ecf11eba22123168c138698f7cb576ac678f5a2e16cd5da08/hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata + Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (4.9 kB) +Collecting httpx<1,>=0.23.0 (from huggingface-hub==1.27.0) + Obtaining dependency information for httpx<1,>=0.23.0 from https://files.pythonhosted.org/packages/2a/39/e50c7c3a983047577ee07d2a9e53faf5a69493943ec3f6a384bdc792deb2/httpx-0.28.1-py3-none-any.whl.metadata + Downloading httpx-0.28.1-py3-none-any.whl.metadata (7.1 kB) +Collecting packaging>=20.9 (from huggingface-hub==1.27.0) + Obtaining dependency information for packaging>=20.9 from https://files.pythonhosted.org/packages/63/34/ba1c580383c9eada3711951fef0795c80b829a078d72188184bcab9dd527/packaging-26.3-py3-none-any.whl.metadata + Downloading packaging-26.3-py3-none-any.whl.metadata (3.5 kB) +Collecting pyyaml>=5.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for pyyaml>=5.1 from https://files.pythonhosted.org/packages/8b/9d/b3589d3877982d4f2329302ef98a8026e7f4443c765c46cfecc8858c6b4b/pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.4 kB) +Collecting tqdm>=4.42.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for tqdm>=4.42.1 from https://files.pythonhosted.org/packages/f9/1c/01bfd571a64e7f270e6bab5e33777debe0edc56759233ce84f27dec92d14/tqdm-4.70.0-py3-none-any.whl.metadata + Downloading tqdm-4.70.0-py3-none-any.whl.metadata (57 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 57.3/57.3 kB 12.3 MB/s eta 0:00:00 +Requirement already satisfied: typing-extensions>=4.1.0 in /opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (4.15.0) +Collecting regex>=2025.10.22 (from transformers==5.14.0) + Obtaining dependency information for regex>=2025.10.22 from https://files.pythonhosted.org/packages/2a/8e/096d00c7c480ef2ff4265349b14e2261d4ab787ba1f74e2e80d1c58079c3/regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (40 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 40.5/40.5 kB 11.7 MB/s eta 0:00:00 +Collecting tokenizers<=0.23.0,>=0.22.0 (from transformers==5.14.0) + Obtaining dependency information for tokenizers<=0.23.0,>=0.22.0 from https://files.pythonhosted.org/packages/2e/76/932be4b50ef6ccedf9d3c6639b056a967a86258c6d9200643f01269211ca/tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (7.3 kB) +Collecting typer (from transformers==5.14.0) + Obtaining dependency information for typer from https://files.pythonhosted.org/packages/43/89/9518bc0c3929bee36b3a4a8e3daddd6e03f92f9961c66d4983b837160543/typer-0.27.1-py3-none-any.whl.metadata + Downloading typer-0.27.1-py3-none-any.whl.metadata (16 kB) +Collecting anyio (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for anyio from https://files.pythonhosted.org/packages/da/35/f2287558c17e29fafc8ef3daf819bb9834061cfa43bff8014f7df7f63bdc/anyio-4.14.2-py3-none-any.whl.metadata + Downloading anyio-4.14.2-py3-none-any.whl.metadata (4.6 kB) +Collecting certifi (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for certifi from https://files.pythonhosted.org/packages/0b/a7/71ac2cff56fec219ed242bb11b8efb69fcc4bec75db06fb7bfe35de520e6/certifi-2026.7.22-py3-none-any.whl.metadata + Downloading certifi-2026.7.22-py3-none-any.whl.metadata (2.5 kB) +Collecting httpcore==1.* (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for httpcore==1.* from https://files.pythonhosted.org/packages/7e/f5/f66802a942d491edb555dd61e3a9961140fd64c90bce1eafd741609d334d/httpcore-1.0.9-py3-none-any.whl.metadata + Downloading httpcore-1.0.9-py3-none-any.whl.metadata (21 kB) +Collecting idna (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for idna from https://files.pythonhosted.org/packages/1e/5e/d4e9f1a599fb8e573b7b87160658329fbf28d19eac2718f51fc3def3aa5a/idna-3.18-py3-none-any.whl.metadata + Downloading idna-3.18-py3-none-any.whl.metadata (6.1 kB) +Collecting h11>=0.16 (from httpcore==1.*->httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for h11>=0.16 from https://files.pythonhosted.org/packages/04/4b/29cac41a4d98d144bf5f6d33995617b185d14b22401f75ca86f384e87ff1/h11-0.16.0-py3-none-any.whl.metadata + Downloading h11-0.16.0-py3-none-any.whl.metadata (8.3 kB) +Collecting shellingham>=1.3.0 (from typer->transformers==5.14.0) + Obtaining dependency information for shellingham>=1.3.0 from https://files.pythonhosted.org/packages/e0/f9/0595336914c5619e5f28a1fb793285925a8cd4b432c9da0a987836c7f822/shellingham-1.5.4-py2.py3-none-any.whl.metadata + Downloading shellingham-1.5.4-py2.py3-none-any.whl.metadata (3.5 kB) +Collecting rich>=13.8.0 (from typer->transformers==5.14.0) + Obtaining dependency information for rich>=13.8.0 from https://files.pythonhosted.org/packages/82/3b/64d4899d73f91ba49a8c18a8ff3f0ea8f1c1d75481760df8c68ef5235bf5/rich-15.0.0-py3-none-any.whl.metadata + Downloading rich-15.0.0-py3-none-any.whl.metadata (18 kB) +Collecting annotated-doc>=0.0.2 (from typer->transformers==5.14.0) + Obtaining dependency information for annotated-doc>=0.0.2 from https://files.pythonhosted.org/packages/3e/30/e900b21425a860e195f32e37657aa1f7c7f2b1bfb26f03ca209b90933c06/annotated_doc-0.0.5-py3-none-any.whl.metadata + Downloading annotated_doc-0.0.5-py3-none-any.whl.metadata (6.5 kB) +Collecting markdown-it-py>=2.2.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for markdown-it-py>=2.2.0 from https://files.pythonhosted.org/packages/b3/81/4da04ced5a082363ecfa159c010d200ecbd959ae410c10c0264a38cac0f5/markdown_it_py-4.2.0-py3-none-any.whl.metadata + Downloading markdown_it_py-4.2.0-py3-none-any.whl.metadata (7.4 kB) +Collecting pygments<3.0.0,>=2.13.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for pygments<3.0.0,>=2.13.0 from https://files.pythonhosted.org/packages/f4/7e/a72dd26f3b0f4f2bf1dd8923c85f7ceb43172af56d63c7383eb62b332364/pygments-2.20.0-py3-none-any.whl.metadata + Downloading pygments-2.20.0-py3-none-any.whl.metadata (2.5 kB) +Collecting mdurl~=0.1 (from markdown-it-py>=2.2.0->rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for mdurl~=0.1 from https://files.pythonhosted.org/packages/b3/38/89ba8ad64ae25be8de66a6d463314cf1eb366222074cfda9ee839c56a4b4/mdurl-0.1.2-py3-none-any.whl.metadata + Downloading mdurl-0.1.2-py3-none-any.whl.metadata (1.6 kB) +Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (16.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 16.6/16.6 MB 138.4 MB/s eta 0:00:00 +Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.9 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.9/6.9 MB 234.3 MB/s eta 0:00:00 +Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (516 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 516.0/516.0 kB 119.5 MB/s eta 0:00:00 +Downloading huggingface_hub-1.27.0-py3-none-any.whl (784 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 784.9/784.9 kB 130.6 MB/s eta 0:00:00 +Downloading transformers-5.14.0-py3-none-any.whl (11.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 11.6/11.6 MB 115.1 MB/s eta 0:00:00 +Downloading click-8.4.2-py3-none-any.whl (119 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 119.2/119.2 kB 40.9 MB/s eta 0:00:00 +Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (4.5 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 4.5/4.5 MB 231.2 MB/s eta 0:00:00 +Downloading httpx-0.28.1-py3-none-any.whl (73 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 73.5/73.5 kB 27.6 MB/s eta 0:00:00 +Downloading httpcore-1.0.9-py3-none-any.whl (78 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 78.8/78.8 kB 24.6 MB/s eta 0:00:00 +Downloading packaging-26.3-py3-none-any.whl (129 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 130.0/130.0 kB 40.2 MB/s eta 0:00:00 +Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (807 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 807.9/807.9 kB 163.7 MB/s eta 0:00:00 +Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (801 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 801.8/801.8 kB 177.2 MB/s eta 0:00:00 +Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (3.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 3.3/3.3 MB 261.5 MB/s eta 0:00:00 +Downloading tqdm-4.70.0-py3-none-any.whl (80 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 80.2/80.2 kB 30.4 MB/s eta 0:00:00 +Downloading typer-0.27.1-py3-none-any.whl (122 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 122.9/122.9 kB 41.7 MB/s eta 0:00:00 +Downloading annotated_doc-0.0.5-py3-none-any.whl (5.3 kB) +Downloading rich-15.0.0-py3-none-any.whl (310 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 310.7/310.7 kB 72.0 MB/s eta 0:00:00 +Downloading shellingham-1.5.4-py2.py3-none-any.whl (9.8 kB) +Downloading anyio-4.14.2-py3-none-any.whl (125 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 125.8/125.8 kB 40.3 MB/s eta 0:00:00 +Downloading idna-3.18-py3-none-any.whl (65 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 65.5/65.5 kB 22.8 MB/s eta 0:00:00 +Downloading certifi-2026.7.22-py3-none-any.whl (136 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 137.0/137.0 kB 47.3 MB/s eta 0:00:00 +Downloading h11-0.16.0-py3-none-any.whl (37 kB) +Downloading markdown_it_py-4.2.0-py3-none-any.whl (91 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 91.7/91.7 kB 32.8 MB/s eta 0:00:00 +Downloading pygments-2.20.0-py3-none-any.whl (1.2 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.2/1.2 MB 200.2 MB/s eta 0:00:00 +Downloading mdurl-0.1.2-py3-none-any.whl (10.0 kB) +Installing collected packages: tqdm, shellingham, safetensors, regex, pyyaml, pygments, pillow, packaging, numpy, mdurl, idna, hf-xet, h11, click, certifi, annotated-doc, markdown-it-py, httpcore, anyio, rich, httpx, typer, huggingface-hub, tokenizers, transformers +Successfully installed annotated-doc-0.0.5 anyio-4.14.2 certifi-2026.7.22 click-8.4.2 h11-0.16.0 hf-xet-1.6.0 httpcore-1.0.9 httpx-0.28.1 huggingface-hub-1.27.0 idna-3.18 markdown-it-py-4.2.0 mdurl-0.1.2 numpy-2.4.6 packaging-26.3 pillow-12.3.0 pygments-2.20.0 pyyaml-6.0.3 regex-2026.7.19 rich-15.0.0 safetensors-0.8.0 shellingham-1.5.4 tokenizers-0.22.2 tqdm-4.70.0 transformers-5.14.0 typer-0.27.1 + +[notice] A new release of pip is available: 23.2.1 -> 26.2.1 +[notice] To update, run: python3.12 -m pip install --upgrade pip +No broken requirements found. +torch 2.11.0+cpu +transformers 5.14.0 +AutoProcessor Qwen3_5ForConditionalGeneration + Fetching 13 files: 0%| | 0/13 [00:00 +ilxyr contribute +ilxyr shared-task-register +ilxyr compile +ilxyr retro --execute +ilxyr forecast +ilxyr fund +ilxyr trust-key +ilxyr budget-payload +ilxyr budget-register +ilxyr allocate ... +ilxyr admit +ilxyr run --execute +ilxyr authorize +ilxyr run-auto +ilxyr loop-cycle +ilxyr sandbox +ilxyr certify +ilxyr calibration +ilxyr status +ilxyr verify diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/execution-summary.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/execution-summary.json new file mode 100644 index 0000000..36fe671 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/execution-summary.json @@ -0,0 +1,8 @@ +{ + "completed_at": "2026-08-13T04:51:28.824571+00:00", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "run_exit_code": 0, + "schema": "lecore.qwen35-execution-summary.v2", + "status_exit_code": 0, + "verify_exit_code": 0 +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/generation.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/generation.json new file mode 100644 index 0000000..3bb2bc7 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/generation.json @@ -0,0 +1 @@ +{"experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", "project": "/opt/qwen35-acceptance-v2/project/project.json", "project_dir": "/opt/qwen35-acceptance-v2/project"} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-admission.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-admission.json new file mode 100644 index 0000000..e8a745b --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-admission.json @@ -0,0 +1,68 @@ +{ + "schema": "ilxyr.admission.v1", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "accepted": true, + "checks": [ + { + "gate": "methodology", + "passed": true, + "detail": "4/4 required research stages resolved" + }, + { + "gate": "outcome_contract", + "passed": true, + "detail": "3 frozen mutually evaluated outcomes declared" + }, + { + "gate": "forecast_participation", + "passed": true, + "detail": "2/2 distinct forecasters" + }, + { + "gate": "forecast_stake", + "passed": true, + "detail": "10/10 forecast credits staked" + }, + { + "gate": "compute_funding", + "passed": true, + "detail": "100/100 compute credits committed" + }, + { + "gate": "executor_available", + "passed": true, + "detail": "local-command adapter is installed" + }, + { + "gate": "weight_protection", + "passed": true, + "detail": "local execution is limited to public-weight handles" + }, + { + "gate": "local_execution_policy", + "passed": true, + "detail": "local execution requires network=open and an absolute executable path" + }, + { + "gate": "code_policy", + "passed": true, + "detail": "local-command directly executes the declared program" + }, + { + "gate": "export_policy", + "passed": true, + "detail": "local-command records stdout, stderr, and metric artifacts" + }, + { + "gate": "reviewer_separation", + "passed": true, + "detail": "proposer and engineering reviewer use distinct handles" + }, + { + "gate": "forecaster_separation", + "passed": true, + "detail": "proposer and forecasters use distinct handles" + } + ], + "decided_at_ms": 1786580796248 +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt new file mode 100644 index 0000000..c29264e --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt @@ -0,0 +1,7 @@ +commit e92382ff2a5e8714466533a160f6609b4ef9cee8 +Author: atimics +AuthorDate: Sat Aug 8 12:46:56 2026 -0700 +Commit: atimics +CommitDate: Sat Aug 8 12:46:56 2026 -0700 + + Publish EXP-006 experiment evidence diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-compile.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-compile.json new file mode 100644 index 0000000..f81351a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-compile.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json new file mode 100644 index 0000000..79d1d1f --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json new file mode 100644 index 0000000..f64cb2a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json new file mode 100644 index 0000000..93f3b24 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json new file mode 100644 index 0000000..669e00b --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json new file mode 100644 index 0000000..f0f2daa --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/32f36305742072ad04d02059b866f824532c2afaefe72151197de23d8967bec6" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json new file mode 100644 index 0000000..8e34947 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/b285b2b9ee58772ca6059775c41bb01f2cc75e3e097ca538dea75040c997e451" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-funding.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-funding.json new file mode 100644 index 0000000..253ab01 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-funding.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/a69efa569914df801764e85672dcdad748c424792de726bcbdd5afd9fa13fdb7" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-init.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-init.json new file mode 100644 index 0000000..0b7bc81 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-init.json @@ -0,0 +1,4 @@ +{ + "initialized": true, + "workspace": "/opt/qwen35-acceptance-v2/workspace" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-run.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-run.json new file mode 100644 index 0000000..a5bdf16 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-run.json @@ -0,0 +1,94 @@ +{ + "run": { + "schema": "ilxyr.run.v1", + "id": "run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "started_at_ms": 1786580796737, + "completed_at_ms": 1786596688293, + "exit_code": 1, + "timed_out": false, + "stdout": " qkv layout: flat (ppl 20.34 vs 201031.79 for grouped -- 9881.2x better)\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n", + "stderr": "[transformers] `torch_dtype` is deprecated! Use `dtype` instead!\n[transformers] The fast path is not available because one of the required library is not installed. Falling back to torch implementation. To install follow https://github.com/fla-org/flash-linear-attention#installation and https://github.com/Dao-AILab/causal-conv1d\n\rLoading weights: 0%| | 0/320 [00:00\n raise SystemExit(main())\n ^^^^^^\n File \"/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py\", line 296, in main\n text_pass, vision_pass, peak_gpu = official_output_smokes(installed_dir)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py\", line 158, in official_output_smokes\n model = Qwen3_5ForConditionalGeneration.from_pretrained(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/modeling_utils.py\", line 4457, in from_pretrained\n loading_info = cls._finalize_model_loading(model, load_config, loading_info)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/modeling_utils.py\", line 4633, in _finalize_model_loading\n log_state_dict_report(\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/utils/loading_report.py\", line 278, in log_state_dict_report\n raise RuntimeError(\nRuntimeError: You set `ignore_mismatched_sizes` to `False`, thus raising an error. For details look at the above report!\n", + "output_truncated": false, + "metrics": {} + }, + "evidence": { + "schema": "ilxyr.evidence.v1", + "id": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "run_ref": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "resolved_outcome": "execution_failure", + "metrics": {}, + "recorded_at_ms": 1786596688298, + "authority": { + "level": "corpus_proxy", + "scope": { + "seeds": [ + 0 + ], + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "coverage": 1.0 + }, + "provenance": { + "artifact_hashes": [ + "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6", + "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91", + "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c", + "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" + ], + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "checker": "checker://lecore/qwen35-acceptance/v2" + } + }, + "lane": "promoted" + }, + "settlements": [ + { + "schema": "ilxyr.forecast_settlement.v1", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "resolved_outcome": "execution_failure", + "brier_score": 1.185, + "stake": 5 + }, + { + "schema": "ilxyr.forecast_settlement.v1", + "forecast_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "resolved_outcome": "execution_failure", + "brier_score": 0.815, + "stake": 5 + } + ], + "calibrations": [ + { + "schema": "ilxyr.calibration_record.v1", + "handle": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical", + "forecasts_settled": 1, + "reliability": 1.185, + "resolution": 0.0, + "brier_score": 1.185, + "uncertainty": 0.0, + "probationary": true, + "last_settlement_at_ms": 1786596688305, + "forecast_ids": [ + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast.empirical" + ] + }, + { + "schema": "ilxyr.calibration_record.v1", + "handle": "model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic", + "forecasts_settled": 1, + "reliability": 0.815, + "resolution": 0.0, + "brier_score": 0.815, + "uncertainty": 0.0, + "probationary": true, + "last_settlement_at_ms": 1786596688309, + "forecast_ids": [ + "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.forecast-mechanistic" + ] + } + ] +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-status.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-status.json new file mode 100644 index 0000000..3e6e18c --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-status.json @@ -0,0 +1,121 @@ +{ + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "compiled_ref": "artifact://sha256/ae9116269123dd22f0530f72dc3a328921a28f367a0e7a41f16004f367b0515d", + "forecasts": 2, + "total_stake": 10, + "funding_commitments": 1, + "funded_compute_credits": 100, + "latest_admission": { + "schema": "ilxyr.admission.v1", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "accepted": true, + "checks": [ + { + "gate": "methodology", + "passed": true, + "detail": "4/4 required research stages resolved" + }, + { + "gate": "outcome_contract", + "passed": true, + "detail": "3 frozen mutually evaluated outcomes declared" + }, + { + "gate": "forecast_participation", + "passed": true, + "detail": "2/2 distinct forecasters" + }, + { + "gate": "forecast_stake", + "passed": true, + "detail": "10/10 forecast credits staked" + }, + { + "gate": "compute_funding", + "passed": true, + "detail": "100/100 compute credits committed" + }, + { + "gate": "executor_available", + "passed": true, + "detail": "local-command adapter is installed" + }, + { + "gate": "weight_protection", + "passed": true, + "detail": "local execution is limited to public-weight handles" + }, + { + "gate": "local_execution_policy", + "passed": true, + "detail": "local execution requires network=open and an absolute executable path" + }, + { + "gate": "code_policy", + "passed": true, + "detail": "local-command directly executes the declared program" + }, + { + "gate": "export_policy", + "passed": true, + "detail": "local-command records stdout, stderr, and metric artifacts" + }, + { + "gate": "reviewer_separation", + "passed": true, + "detail": "proposer and engineering reviewer use distinct handles" + }, + { + "gate": "forecaster_separation", + "passed": true, + "detail": "proposer and forecasters use distinct handles" + } + ], + "decided_at_ms": 1786580796248 + }, + "execution_started": true, + "latest_run": { + "schema": "ilxyr.run.v1", + "id": "run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "started_at_ms": 1786580796737, + "completed_at_ms": 1786596688293, + "exit_code": 1, + "timed_out": false, + "stdout": " qkv layout: flat (ppl 20.34 vs 201031.79 for grouped -- 9881.2x better)\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n", + "stderr": "[transformers] `torch_dtype` is deprecated! Use `dtype` instead!\n[transformers] The fast path is not available because one of the required library is not installed. Falling back to torch implementation. To install follow https://github.com/fla-org/flash-linear-attention#installation and https://github.com/Dao-AILab/causal-conv1d\n\rLoading weights: 0%| | 0/320 [00:00\n raise SystemExit(main())\n ^^^^^^\n File \"/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py\", line 296, in main\n text_pass, vision_pass, peak_gpu = official_output_smokes(installed_dir)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/lecore/experiments/qwen35_acceptance/run.py\", line 158, in official_output_smokes\n model = Qwen3_5ForConditionalGeneration.from_pretrained(\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/modeling_utils.py\", line 4457, in from_pretrained\n loading_info = cls._finalize_model_loading(model, load_config, loading_info)\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/modeling_utils.py\", line 4633, in _finalize_model_loading\n log_state_dict_report(\n File \"/opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages/transformers/utils/loading_report.py\", line 278, in log_state_dict_report\n raise RuntimeError(\nRuntimeError: You set `ignore_mismatched_sizes` to `False`, thus raising an error. For details look at the above report!\n", + "output_truncated": false, + "metrics": {} + }, + "latest_evidence": { + "schema": "ilxyr.evidence.v1", + "id": "evidence:run:lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance:1786580796737", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", + "run_ref": "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "resolved_outcome": "execution_failure", + "metrics": {}, + "recorded_at_ms": 1786596688298, + "authority": { + "level": "corpus_proxy", + "scope": { + "seeds": [ + 0 + ], + "eval_set": "dataset://lecore/qwen-evaluation/a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94", + "coverage": 1.0 + }, + "provenance": { + "artifact_hashes": [ + "artifact://sha256/949c66e8b6cceb11c13828d9139b43a2f473d58ec8aa0b41f800d0eab681a3b6", + "artifact://sha256/9a040b3d177a98dfca6cfcf172f7c5b42017dbf5be4679ae49a7bc241d1c3d91", + "artifact://sha256/9c9915011dbd3857bdc40cf86b99b0abf48ea0a9cf2eead10d50d47179bd335c", + "artifact://sha256/dee0b19888937eb4b4109ef6558ce3e5c6a530469019b94f4905c01d144a6d8f", + "artifact://sha256/ecb26ed6cbf11c3b85d05e1c6ade71ea3605755a389d067ccb712944cf9ef2b0" + ], + "model_lineage": "model://Qwen/Qwen3.5-0.8B/20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "checker": "checker://lecore/qwen35-acceptance/v2" + } + }, + "lane": "promoted" + } +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-verify.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-verify.json new file mode 100644 index 0000000..d3d17dd --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/ilxyr-verify.json @@ -0,0 +1,5 @@ +{ + "objects_checked": 15, + "events_checked": 16, + "valid": true +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/launch-manifest.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/launch-manifest.json new file mode 100644 index 0000000..d5e6e2e --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/launch-manifest.json @@ -0,0 +1,106 @@ +{ + "schema": "lecore.qwen35-launch.v1", + "state": "authorized_prelaunch", + "attempt": { + "experiment_version": 2, + "maximum_formal_runs": 1, + "retry_after_admission": false, + "seed": 0, + "authorized_at": "2026-08-13T00:09:30Z", + "predecessor": { + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "outcome": "execution_failure", + "scientific_result": false, + "corrective_change": "preserve the selected virtual-environment interpreter path" + } + }, + "source": { + "repository": "https://github.com/atimics/holostuff.git", + "commit": "2f32151c6d44c80a7e2ff98c0696650cd1d9c6f1" + }, + "ilxyr": { + "repository": "https://github.com/cenetex/ilXyr.git", + "commit": "e92382ff2a5e8714466533a160f6609b4ef9cee8" + }, + "model": { + "repository": "Qwen/Qwen3.5-0.8B", + "revision": "2fc06364715b967f1860aea9cf38778875588b17", + "license": "Apache-2.0", + "critical_files": [ + { + "path": "model.safetensors-00001-of-00001.safetensors", + "bytes": 1746942600, + "sha256": "04b1c301231dd422b8860db31311ab2721511346a32cb1e079c4c4e5f1fe4696" + }, + { + "path": "tokenizer.json", + "bytes": 12807982, + "sha256": "5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42" + } + ], + "complete_manifest": "generated and frozen before ilxyr admission" + }, + "corpora": { + "installation": { + "handle": "git:a04ab5692be38f06120aba4b0bc5e2a284eb2c79:REFERENCE.md", + "license": "MIT", + "bytes": 2300089, + "sha256": "d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f" + }, + "evaluation": { + "handle": "https://www.gutenberg.org/ebooks/18", + "retrieved_at": "2026-08-12T21:32:31Z", + "last_modified": "2026-08-01T07:33:54Z", + "copyright": "Public domain in the USA", + "bytes": 1213410, + "sha256": "a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + } + }, + "method": { + "minimum_paired_positions": 4096, + "reference_logit_relative_tolerance": 0.001, + "maximum_regression": 0.01, + "confidence": 0.95, + "bootstrap": "paired moving block", + "spectral_filtering": false, + "installer_experimental": true + }, + "dependencies": { + "python": "3.12", + "numpy": "2.4.6", + "pillow": "12.3.0", + "torch": "2.11.0+cpu", + "transformers": "5.14.0", + "resolved_environment": "recorded before ilxyr admission" + }, + "aws": { + "region": "us-east-1", + "availability_zone": "us-east-1a", + "ami": "ami-07a5b367e8dc8bd92", + "ami_name": "al2023-ami-2023.12.20260803.3-kernel-6.1-x86_64", + "instance_type": "r7i.4xlarge", + "vcpu": 16, + "memory_gib": 128, + "market": "on-demand", + "hourly_compute_usd": 1.0584, + "root_volume": { + "type": "gp3", + "size_gib": 100, + "encrypted": true, + "delete_on_termination": true + }, + "inbound_ports": [], + "experiment_timeout_seconds": 21600, + "instance_lifetime_seconds": 28800, + "instance_initiated_shutdown_behavior": "terminate", + "maximum_compute_usd": 8.4672, + "total_ceiling_usd": 10.0 + }, + "publication": { + "temporary_staging": "s3://zero-training-022118847419/qwen35-acceptance/2f32151-2fc06364-a6c9d113/v2-formal/", + "review_index": "https://github.com/atimics/holostuff/pull/3", + "permanent_transport": "Arweave", + "publish_all_outcomes": true, + "exclude_model_weights": true + } +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/lecore-commit.txt b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/lecore-commit.txt new file mode 100644 index 0000000..67ff02b --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/lecore-commit.txt @@ -0,0 +1,7 @@ +commit 2f32151c6d44c80a7e2ff98c0696650cd1d9c6f1 +Author: atimics +AuthorDate: Wed Aug 12 17:08:37 2026 -0700 +Commit: atimics +CommitDate: Wed Aug 12 17:08:37 2026 -0700 + + experiments: version formal qwen attempts diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/lecore-status.txt b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/lecore-status.txt new file mode 100644 index 0000000..e69de29 diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/model-manifest.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/model-manifest.json new file mode 100644 index 0000000..65559ac --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/model-manifest.json @@ -0,0 +1,66 @@ +{ + "digest": "20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "files": [ + { + "bytes": 11544, + "path": "LICENSE", + "sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a" + }, + { + "bytes": 61705, + "path": "README.md", + "sha256": "87a163af54f32fa608a0f8d3ac67945c53dd2b4a7c96740b3d7fdc28e8458864" + }, + { + "bytes": 7755, + "path": "chat_template.jinja", + "sha256": "273d8e0e683b885071fb17e08d71e5f2a5ddfb5309756181681de4f5a1822d80" + }, + { + "bytes": 2907, + "path": "config.json", + "sha256": "b90b86f35c8e6925ef74ee04d0e758f0a845c83a42089ad82bbaa948de9b4204" + }, + { + "bytes": 3353259, + "path": "merges.txt", + "sha256": "a9d356d7bdf1ef4949e3e748e95b8e10ad9d4e2e838eddc38a0a7b6b94d1db8d" + }, + { + "bytes": 1746942600, + "path": "model.safetensors-00001-of-00001.safetensors", + "sha256": "04b1c301231dd422b8860db31311ab2721511346a32cb1e079c4c4e5f1fe4696" + }, + { + "bytes": 50900, + "path": "model.safetensors.index.json", + "sha256": "d8a08838a613b025eb7952ed9db11696213e57e76a375661ef5c12f9dd5dcf4e" + }, + { + "bytes": 390, + "path": "preprocessor_config.json", + "sha256": "27225450ac9c6529872ee1924fcb0962ff5634834f817040f444118116f4e516" + }, + { + "bytes": 12807982, + "path": "tokenizer.json", + "sha256": "5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42" + }, + { + "bytes": 16709, + "path": "tokenizer_config.json", + "sha256": "49e2b6e395f959f077f1e992b338919c0d4a9732fc6e613995e06557f843500c" + }, + { + "bytes": 385, + "path": "video_preprocessor_config.json", + "sha256": "7768af27c1fafa9cc9011c1dc20067e03f8915e03b63504550e11d5066986d13" + }, + { + "bytes": 6722759, + "path": "vocab.json", + "sha256": "ce99b4cb2983d118806ce0a8b777a35b093e2000a503ebde25853284c9dfa003" + } + ], + "revision": "2fc06364715b967f1860aea9cf38778875588b17" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/requirements-resolved.txt b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/requirements-resolved.txt new file mode 100644 index 0000000..7bfa4f7 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/requirements-resolved.txt @@ -0,0 +1,36 @@ +Jinja2==3.1.6 +MarkupSafe==3.0.3 +PyYAML==6.0.3 +Pygments==2.20.0 +annotated-doc==0.0.5 +anyio==4.14.2 +certifi==2026.7.22 +click==8.4.2 +filelock==3.29.0 +fsspec==2026.4.0 +h11==0.16.0 +hf-xet==1.6.0 +httpcore==1.0.9 +httpx==0.28.1 +huggingface_hub==1.27.0 +idna==3.18 +markdown-it-py==4.2.0 +mdurl==0.1.2 +mpmath==1.3.0 +networkx==3.6.1 +numpy==2.4.6 +packaging==26.3 +pillow==12.3.0 +pip==23.2.1 +regex==2026.7.19 +rich==15.0.0 +safetensors==0.8.0 +setuptools==78.1.0 +shellingham==1.5.4 +sympy==1.14.0 +tokenizers==0.22.2 +torch==2.11.0+cpu +tqdm==4.70.0 +transformers==5.14.0 +typer==0.27.1 +typing_extensions==4.15.0 diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/runner-status.json b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/runner-status.json new file mode 100644 index 0000000..d5b3d7a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/runner-status.json @@ -0,0 +1,7 @@ +{ + "exit_code": 0, + "formal_run_started": true, + "schema": "lecore.qwen35-runner-status.v2", + "stage": "complete", + "updated_at": "2026-08-13T04:51:28.842133+00:00" +} diff --git a/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/user-data.log b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/user-data.log new file mode 100644 index 0000000..35cff55 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v2-20c3330d0b3e-execution-failure/result/user-data.log @@ -0,0 +1,463 @@ +Shutdown scheduled for Thu 2026-08-13 08:24:56 UTC, use 'shutdown -c' to cancel. +Amazon Linux 2023 repository 80 MB/s | 75 MB 00:00 +Amazon Linux 2023 Kernel Livepatch repository 651 kB/s | 69 kB 00:00 +Package tar-2:1.34-1.amzn2023.0.4.x86_64 is already installed. +Package gzip-1.12-1.amzn2023.0.1.x86_64 is already installed. +Dependencies resolved. +================================================================================ + Package Arch Version Repository Size +================================================================================ +Installing: + cargo x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 8.2 M + git x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 53 k + python3.12 x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 27 k + python3.12-pip noarch 23.2.1-4.amzn2023.0.10 amazonlinux 2.8 M + rust x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 62 M +Installing dependencies: + annobin-docs noarch 12.69-1.amzn2023.0.1 amazonlinux 93 k + annobin-plugin-gcc x86_64 12.69-1.amzn2023.0.1 amazonlinux 974 k + cpp x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 11 M + gc x86_64 8.0.4-5.amzn2023.0.2 amazonlinux 105 k + gcc x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 36 M + gcc-plugin-annobin x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 38 k + git-core x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 4.9 M + git-core-doc noarch 2.50.1-1.amzn2023.0.1 amazonlinux 2.8 M + glibc-devel x86_64 2.34-231.amzn2023.0.5 amazonlinux 33 k + glibc-headers-x86 noarch 2.34-231.amzn2023.0.5 amazonlinux 439 k + guile22 x86_64 2.2.7-2.amzn2023.0.3 amazonlinux 6.4 M + kernel-headers x86_64 1:6.1.177-224.371.amzn2023 amazonlinux 1.4 M + libmpc x86_64 1.2.1-2.amzn2023.0.2 amazonlinux 62 k + libtool-ltdl x86_64 2.4.7-1.amzn2023.0.3 amazonlinux 38 k + libxcrypt-devel x86_64 4.4.33-7.amzn2023 amazonlinux 32 k + make x86_64 1:4.3-5.amzn2023.0.2 amazonlinux 534 k + mpdecimal x86_64 2.5.1-3.amzn2023.0.3 amazonlinux 101 k + perl-Error noarch 1:0.17030-2.amzn2023.0.1 amazonlinux 42 k + perl-File-Find noarch 1.37-477.amzn2023.0.9 amazonlinux 26 k + perl-Git noarch 2.50.1-1.amzn2023.0.1 amazonlinux 41 k + perl-TermReadKey x86_64 2.38-9.amzn2023.0.3 amazonlinux 36 k + perl-lib x86_64 0.65-477.amzn2023.0.9 amazonlinux 15 k + python3.12-libs x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 9.2 M + python3.12-pip-wheel noarch 23.2.1-4.amzn2023.0.10 amazonlinux 1.5 M + rust-std-static x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 39 M +Installing weak dependencies: + python3.12-setuptools noarch 68.2.2-4.amzn2023.0.3 amazonlinux 1.4 M + +Transaction Summary +================================================================================ +Install 31 Packages + +Total download size: 189 M +Installed size: 700 M +Downloading Packages: +(1/31): annobin-docs-12.69-1.amzn2023.0.1.noarc 2.5 MB/s | 93 kB 00:00 +(2/31): annobin-plugin-gcc-12.69-1.amzn2023.0.1 24 MB/s | 974 kB 00:00 +(3/31): gc-8.0.4-5.amzn2023.0.2.x86_64.rpm 4.0 MB/s | 105 kB 00:00 +(4/31): cpp-11.5.0-5.amzn2023.0.5.x86_64.rpm 79 MB/s | 11 MB 00:00 +(5/31): gcc-plugin-annobin-11.5.0-5.amzn2023.0. 1.2 MB/s | 38 kB 00:00 +(6/31): git-2.50.1-1.amzn2023.0.1.x86_64.rpm 1.7 MB/s | 53 kB 00:00 +(7/31): git-core-2.50.1-1.amzn2023.0.1.x86_64.r 74 MB/s | 4.9 MB 00:00 +(8/31): git-core-doc-2.50.1-1.amzn2023.0.1.noar 56 MB/s | 2.8 MB 00:00 +(9/31): glibc-devel-2.34-231.amzn2023.0.5.x86_6 941 kB/s | 33 kB 00:00 +(10/31): cargo-1.97.0-1.amzn2023.0.1.x86_64.rpm 19 MB/s | 8.2 MB 00:00 +(11/31): glibc-headers-x86-2.34-231.amzn2023.0. 12 MB/s | 439 kB 00:00 +(12/31): gcc-11.5.0-5.amzn2023.0.5.x86_64.rpm 78 MB/s | 36 MB 00:00 +(13/31): kernel-headers-6.1.177-224.371.amzn202 13 MB/s | 1.4 MB 00:00 +(14/31): guile22-2.2.7-2.amzn2023.0.3.x86_64.rp 50 MB/s | 6.4 MB 00:00 +(15/31): libmpc-1.2.1-2.amzn2023.0.2.x86_64.rpm 2.2 MB/s | 62 kB 00:00 +(16/31): libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_ 1.6 MB/s | 38 kB 00:00 +(17/31): libxcrypt-devel-4.4.33-7.amzn2023.x86_ 1.2 MB/s | 32 kB 00:00 +(18/31): mpdecimal-2.5.1-3.amzn2023.0.3.x86_64. 4.3 MB/s | 101 kB 00:00 +(19/31): make-4.3-5.amzn2023.0.2.x86_64.rpm 19 MB/s | 534 kB 00:00 +(20/31): perl-Error-0.17030-2.amzn2023.0.1.noar 1.6 MB/s | 42 kB 00:00 +(21/31): perl-File-Find-1.37-477.amzn2023.0.9.n 1.0 MB/s | 26 kB 00:00 +(22/31): perl-Git-2.50.1-1.amzn2023.0.1.noarch. 1.5 MB/s | 41 kB 00:00 +(23/31): perl-lib-0.65-477.amzn2023.0.9.x86_64. 536 kB/s | 15 kB 00:00 +(24/31): python3.12-3.12.13-2.amzn2023.0.5.x86_ 1.0 MB/s | 27 kB 00:00 +(25/31): perl-TermReadKey-2.38-9.amzn2023.0.3.x 1.0 MB/s | 36 kB 00:00 +(26/31): python3.12-pip-23.2.1-4.amzn2023.0.10. 70 MB/s | 2.8 MB 00:00 +(27/31): python3.12-pip-wheel-23.2.1-4.amzn2023 35 MB/s | 1.5 MB 00:00 +(28/31): python3.12-setuptools-68.2.2-4.amzn202 27 MB/s | 1.4 MB 00:00 +(29/31): python3.12-libs-3.12.13-2.amzn2023.0.5 77 MB/s | 9.2 MB 00:00 +(30/31): rust-std-static-1.97.0-1.amzn2023.0.1. 72 MB/s | 39 MB 00:00 +(31/31): rust-1.97.0-1.amzn2023.0.1.x86_64.rpm 73 MB/s | 62 MB 00:00 +-------------------------------------------------------------------------------- +Total 121 MB/s | 189 MB 00:01 +Running transaction check +Transaction check succeeded. +Running transaction test +Transaction test succeeded. +Running transaction + Preparing : 1/1 + Installing : libmpc-1.2.1-2.amzn2023.0.2.x86_64 1/31 + Installing : git-core-2.50.1-1.amzn2023.0.1.x86_64 2/31 + Installing : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 3/31 + Installing : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Installing : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 5/31 + Installing : perl-lib-0.65-477.amzn2023.0.9.x86_64 6/31 + Installing : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 7/31 + Installing : perl-File-Find-1.37-477.amzn2023.0.9.noarch 8/31 + Installing : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 9/31 + Installing : perl-Git-2.50.1-1.amzn2023.0.1.noarch 10/31 + Installing : git-2.50.1-1.amzn2023.0.1.x86_64 11/31 + Installing : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 12/31 + Installing : python3.12-3.12.13-2.amzn2023.0.5.x86_64 13/31 + Installing : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 14/31 + Installing : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 15/31 + Installing : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Installing : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 17/31 + Installing : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 18/31 + Installing : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 19/31 + Installing : glibc-devel-2.34-231.amzn2023.0.5.x86_64 20/31 + Installing : gc-8.0.4-5.amzn2023.0.2.x86_64 21/31 + Installing : guile22-2.2.7-2.amzn2023.0.3.x86_64 22/31 + Installing : make-1:4.3-5.amzn2023.0.2.x86_64 23/31 + Installing : gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Running scriptlet: gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Installing : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 25/31 + Installing : rust-1.97.0-1.amzn2023.0.1.x86_64 26/31 + Installing : annobin-docs-12.69-1.amzn2023.0.1.noarch 27/31 + Installing : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Running scriptlet: annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Installing : cargo-1.97.0-1.amzn2023.0.1.x86_64 29/31 + Installing : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 30/31 + Installing : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Running scriptlet: python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Verifying : annobin-docs-12.69-1.amzn2023.0.1.noarch 1/31 + Verifying : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 2/31 + Verifying : cargo-1.97.0-1.amzn2023.0.1.x86_64 3/31 + Verifying : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Verifying : gc-8.0.4-5.amzn2023.0.2.x86_64 5/31 + Verifying : gcc-11.5.0-5.amzn2023.0.5.x86_64 6/31 + Verifying : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 7/31 + Verifying : git-2.50.1-1.amzn2023.0.1.x86_64 8/31 + Verifying : git-core-2.50.1-1.amzn2023.0.1.x86_64 9/31 + Verifying : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 10/31 + Verifying : glibc-devel-2.34-231.amzn2023.0.5.x86_64 11/31 + Verifying : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 12/31 + Verifying : guile22-2.2.7-2.amzn2023.0.3.x86_64 13/31 + Verifying : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 14/31 + Verifying : libmpc-1.2.1-2.amzn2023.0.2.x86_64 15/31 + Verifying : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Verifying : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 17/31 + Verifying : make-1:4.3-5.amzn2023.0.2.x86_64 18/31 + Verifying : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 19/31 + Verifying : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 20/31 + Verifying : perl-File-Find-1.37-477.amzn2023.0.9.noarch 21/31 + Verifying : perl-Git-2.50.1-1.amzn2023.0.1.noarch 22/31 + Verifying : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 23/31 + Verifying : perl-lib-0.65-477.amzn2023.0.9.x86_64 24/31 + Verifying : python3.12-3.12.13-2.amzn2023.0.5.x86_64 25/31 + Verifying : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 26/31 + Verifying : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 27/31 + Verifying : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 28/31 + Verifying : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 29/31 + Verifying : rust-1.97.0-1.amzn2023.0.1.x86_64 30/31 + Verifying : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 31/31 + +Installed: + annobin-docs-12.69-1.amzn2023.0.1.noarch + annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 + cargo-1.97.0-1.amzn2023.0.1.x86_64 + cpp-11.5.0-5.amzn2023.0.5.x86_64 + gc-8.0.4-5.amzn2023.0.2.x86_64 + gcc-11.5.0-5.amzn2023.0.5.x86_64 + gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 + git-2.50.1-1.amzn2023.0.1.x86_64 + git-core-2.50.1-1.amzn2023.0.1.x86_64 + git-core-doc-2.50.1-1.amzn2023.0.1.noarch + glibc-devel-2.34-231.amzn2023.0.5.x86_64 + glibc-headers-x86-2.34-231.amzn2023.0.5.noarch + guile22-2.2.7-2.amzn2023.0.3.x86_64 + kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 + libmpc-1.2.1-2.amzn2023.0.2.x86_64 + libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 + libxcrypt-devel-4.4.33-7.amzn2023.x86_64 + make-1:4.3-5.amzn2023.0.2.x86_64 + mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 + perl-Error-1:0.17030-2.amzn2023.0.1.noarch + perl-File-Find-1.37-477.amzn2023.0.9.noarch + perl-Git-2.50.1-1.amzn2023.0.1.noarch + perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 + perl-lib-0.65-477.amzn2023.0.9.x86_64 + python3.12-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-pip-23.2.1-4.amzn2023.0.10.noarch + python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch + python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch + rust-1.97.0-1.amzn2023.0.1.x86_64 + rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 + +Complete! +Cloning into '/opt/qwen35-acceptance-v2/lecore'... +Updating files: 0% (2/903) Updating files: 1% (10/903) Updating files: 2% (19/903) Updating files: 3% (28/903) Updating files: 4% (37/903) Updating files: 5% (46/903) Updating files: 6% (55/903) Updating files: 7% (64/903) Updating files: 8% (73/903) Updating files: 9% (82/903) Updating files: 10% (91/903) Updating files: 11% (100/903) Updating files: 12% (109/903) Updating files: 13% (118/903) Updating files: 14% (127/903) Updating files: 15% (136/903) Updating files: 16% (145/903) Updating files: 17% (154/903) Updating files: 18% (163/903) Updating files: 19% (172/903) Updating files: 20% (181/903) Updating files: 21% (190/903) Updating files: 22% (199/903) Updating files: 23% (208/903) Updating files: 24% (217/903) Updating files: 25% (226/903) Updating files: 26% (235/903) Updating files: 27% (244/903) Updating files: 28% (253/903) Updating files: 29% (262/903) Updating files: 30% (271/903) Updating files: 31% (280/903) Updating files: 32% (289/903) Updating files: 33% (298/903) Updating files: 34% (308/903) Updating files: 35% (317/903) Updating files: 36% (326/903) Updating files: 37% (335/903) Updating files: 38% (344/903) Updating files: 39% (353/903) Updating files: 40% (362/903) Updating files: 41% (371/903) Updating files: 42% (380/903) Updating files: 43% (389/903) Updating files: 44% (398/903) Updating files: 45% (407/903) Updating files: 46% (416/903) Updating files: 47% (425/903) Updating files: 48% (434/903) Updating files: 49% (443/903) Updating files: 50% (452/903) Updating files: 51% (461/903) Updating files: 52% (470/903) Updating files: 53% (479/903) Updating files: 54% (488/903) Updating files: 55% (497/903) Updating files: 56% (506/903) Updating files: 57% (515/903) Updating files: 58% (524/903) Updating files: 59% (533/903) Updating files: 60% (542/903) Updating files: 61% (551/903) Updating files: 62% (560/903) Updating files: 63% (569/903) Updating files: 64% (578/903) Updating files: 65% (587/903) Updating files: 66% (596/903) Updating files: 67% (606/903) Updating files: 68% (615/903) Updating files: 69% (624/903) Updating files: 70% (633/903) Updating files: 71% (642/903) Updating files: 72% (651/903) Updating files: 73% (660/903) Updating files: 74% (669/903) Updating files: 75% (678/903) Updating files: 76% (687/903) Updating files: 77% (696/903) Updating files: 78% (705/903) Updating files: 79% (714/903) Updating files: 80% (723/903) Updating files: 81% (732/903) Updating files: 82% (741/903) Updating files: 83% (750/903) Updating files: 84% (759/903) Updating files: 85% (768/903) Updating files: 86% (777/903) Updating files: 87% (786/903) Updating files: 88% (795/903) Updating files: 89% (804/903) Updating files: 90% (813/903) Updating files: 91% (822/903) Updating files: 92% (831/903) Updating files: 93% (840/903) Updating files: 94% (849/903) Updating files: 95% (858/903) Updating files: 96% (867/903) Updating files: 97% (876/903) Updating files: 98% (885/903) Updating files: 99% (894/903) Updating files: 100% (903/903) Updating files: 100% (903/903), done. +HEAD is now at 2f32151 experiments: version formal qwen attempts +/opt/qwen35-acceptance-v2/lecore/REFERENCE.md: OK +/opt/qwen35-acceptance-v2/inputs/federalist-papers.txt: OK +Cloning into '/opt/qwen35-acceptance-v2/ilxyr'... +HEAD is now at e92382f Publish EXP-006 experiment evidence + Updating crates.io index + Downloading crates ... + Downloaded serde v1.0.229 + Downloaded base64 v0.22.1 + Downloaded rustc_version v0.4.1 + Downloaded memchr v2.8.3 + Downloaded ed25519-dalek v2.2.0 + Downloaded quote v1.0.46 + Downloaded serde_core v1.0.229 + Downloaded curve25519-dalek v4.1.3 + Downloaded cfg-if v1.0.4 + Downloaded digest v0.10.7 + Downloaded block-buffer v0.10.4 + Downloaded getrandom v0.2.17 + Downloaded proc-macro2 v1.0.106 + Downloaded curve25519-dalek-derive v0.1.1 + Downloaded generic-array v0.14.7 + Downloaded itoa v1.0.18 + Downloaded rand_core v0.6.4 + Downloaded version_check v0.9.5 + Downloaded crypto-common v0.1.7 + Downloaded serde_json v1.0.150 + Downloaded spki v0.7.3 + Downloaded zmij v1.0.23 + Downloaded base64ct v1.8.3 + Downloaded sha2 v0.10.9 + Downloaded const-oid v0.9.6 + Downloaded semver v1.0.28 + Downloaded serde_derive v1.0.229 + Downloaded typenum v1.20.1 + Downloaded unicode-ident v1.0.24 + Downloaded syn v2.0.119 + Downloaded der v0.7.10 + Downloaded subtle v2.6.1 + Downloaded pkcs8 v0.10.2 + Downloaded ed25519 v2.2.3 + Downloaded syn v3.0.0 + Downloaded zeroize v1.9.0 + Downloaded signature v2.2.0 + Downloaded cpufeatures v0.2.17 + Downloaded libc v0.2.186 + Compiling proc-macro2 v1.0.106 + Compiling version_check v0.9.5 + Compiling unicode-ident v1.0.24 + Compiling quote v1.0.46 + Compiling typenum v1.20.1 + Compiling semver v1.0.28 + Compiling serde_core v1.0.229 + Compiling cpufeatures v0.2.17 + Compiling zmij v1.0.23 + Compiling cfg-if v1.0.4 + Compiling subtle v2.6.1 + Compiling serde_json v1.0.150 + Compiling serde v1.0.229 + Compiling signature v2.2.0 + Compiling zeroize v1.9.0 + Compiling memchr v2.8.3 + Compiling itoa v1.0.18 + Compiling base64 v0.22.1 + Compiling ed25519 v2.2.3 + Compiling rustc_version v0.4.1 + Compiling generic-array v0.14.7 + Compiling curve25519-dalek v4.1.3 + Compiling syn v2.0.119 + Compiling syn v3.0.0 + Compiling block-buffer v0.10.4 + Compiling crypto-common v0.1.7 + Compiling digest v0.10.7 + Compiling sha2 v0.10.9 + Compiling serde_derive v1.0.229 + Compiling curve25519-dalek-derive v0.1.1 + Compiling ed25519-dalek v2.2.0 + Compiling ilxyr-core v0.1.0 (/opt/qwen35-acceptance-v2/ilxyr/crates/ilxyr-core) + Compiling ilxyr-cli v0.1.0 (/opt/qwen35-acceptance-v2/ilxyr/crates/ilxyr-cli) + Finished `release` profile [optimized] target(s) in 20.15s +Looking in indexes: https://download.pytorch.org/whl/cpu +Collecting torch==2.11.0 + Obtaining dependency information for torch==2.11.0 from https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata + Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata (29 kB) +Collecting filelock (from torch==2.11.0) + Obtaining dependency information for filelock from https://files.pythonhosted.org/packages/81/47/dd9a212ef6e343a6857485ffe25bba537304f1913bdbed446a23f7f592e1/filelock-3.29.0-py3-none-any.whl.metadata + Downloading filelock-3.29.0-py3-none-any.whl.metadata (2.0 kB) +Collecting typing-extensions>=4.10.0 (from torch==2.11.0) + Obtaining dependency information for typing-extensions>=4.10.0 from https://files.pythonhosted.org/packages/18/67/36e9267722cc04a6b9f15c7f3441c2363321a3ea07da7ae0c0707beb2a9c/typing_extensions-4.15.0-py3-none-any.whl.metadata + Downloading typing_extensions-4.15.0-py3-none-any.whl.metadata (3.3 kB) +Collecting setuptools<82 (from torch==2.11.0) + Obtaining dependency information for setuptools<82 from https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata (6.6 kB) +Collecting sympy>=1.13.3 (from torch==2.11.0) + Obtaining dependency information for sympy>=1.13.3 from https://files.pythonhosted.org/packages/a2/09/77d55d46fd61b4a135c444fc97158ef34a095e5681d0a6c10b75bf356191/sympy-1.14.0-py3-none-any.whl.metadata + Downloading sympy-1.14.0-py3-none-any.whl.metadata (12 kB) +Collecting networkx>=2.5.1 (from torch==2.11.0) + Obtaining dependency information for networkx>=2.5.1 from https://files.pythonhosted.org/packages/9e/c9/b2622292ea83fbb4ec318f5b9ab867d0a28ab43c5717bb85b0a5f6b3b0a4/networkx-3.6.1-py3-none-any.whl.metadata + Downloading networkx-3.6.1-py3-none-any.whl.metadata (6.8 kB) +Collecting jinja2 (from torch==2.11.0) + Obtaining dependency information for jinja2 from https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata (2.9 kB) +Collecting fsspec>=0.8.5 (from torch==2.11.0) + Obtaining dependency information for fsspec>=0.8.5 from https://files.pythonhosted.org/packages/d5/0c/043d5e551459da400957a1395e0febbf771446ff34291afcbe3d8be2a279/fsspec-2026.4.0-py3-none-any.whl.metadata + Downloading fsspec-2026.4.0-py3-none-any.whl.metadata (10 kB) +Collecting mpmath<1.4,>=1.1.0 (from sympy>=1.13.3->torch==2.11.0) + Obtaining dependency information for mpmath<1.4,>=1.1.0 from https://files.pythonhosted.org/packages/43/e3/7d92a15f894aa0c9c4b49b8ee9ac9850d6e63b03c9c32c0367a13ae62209/mpmath-1.3.0-py3-none-any.whl.metadata + Downloading mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) +Collecting MarkupSafe>=2.0 (from jinja2->torch==2.11.0) + Obtaining dependency information for MarkupSafe>=2.0 from https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.7 kB) +Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl (190.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 190.3/190.3 MB 14.0 MB/s eta 0:00:00 +Downloading fsspec-2026.4.0-py3-none-any.whl (203 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 203.4/203.4 kB 27.9 MB/s eta 0:00:00 +Downloading networkx-3.6.1-py3-none-any.whl (2.1 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 2.1/2.1 MB 152.6 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl (1.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.3/1.3 MB 196.3 MB/s eta 0:00:00 +Downloading sympy-1.14.0-py3-none-any.whl (6.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.3/6.3 MB 209.5 MB/s eta 0:00:00 +Downloading typing_extensions-4.15.0-py3-none-any.whl (44 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 44.6/44.6 kB 15.4 MB/s eta 0:00:00 +Downloading filelock-3.29.0-py3-none-any.whl (39 kB) +Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl (134 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 134.9/134.9 kB 42.4 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (22 kB) +Downloading mpmath-1.3.0-py3-none-any.whl (536 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 536.2/536.2 kB 144.5 MB/s eta 0:00:00 +Installing collected packages: mpmath, typing-extensions, sympy, setuptools, networkx, MarkupSafe, fsspec, filelock, jinja2, torch +Successfully installed MarkupSafe-3.0.3 filelock-3.29.0 fsspec-2026.4.0 jinja2-3.1.6 mpmath-1.3.0 networkx-3.6.1 setuptools-78.1.0 sympy-1.14.0 torch-2.11.0+cpu typing-extensions-4.15.0 +Collecting numpy==2.4.6 + Obtaining dependency information for numpy==2.4.6 from https://files.pythonhosted.org/packages/99/60/14115e6364fa676c5397c2ad3004e527e9aa487abf5d0706ec81bbd08529/numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (6.6 kB) +Collecting pillow==12.3.0 + Obtaining dependency information for pillow==12.3.0 from https://files.pythonhosted.org/packages/84/21/a35af28dcc61f37ed850a2d64c65c701321dfbf25085e469d5559360cbbf/pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (9.1 kB) +Collecting safetensors==0.8.0 + Obtaining dependency information for safetensors==0.8.0 from https://files.pythonhosted.org/packages/28/50/f203ff3a3ddfe19308efc83c5a3a29ed02bf786732ec35e68bf9162f3365/safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (4.2 kB) +Collecting huggingface-hub==1.27.0 + Obtaining dependency information for huggingface-hub==1.27.0 from https://files.pythonhosted.org/packages/de/d8/95b735e183957c1f26d94c52977f09d466d55119cbbc1558ea4975e4c216/huggingface_hub-1.27.0-py3-none-any.whl.metadata + Downloading huggingface_hub-1.27.0-py3-none-any.whl.metadata (16 kB) +Collecting transformers==5.14.0 + Obtaining dependency information for transformers==5.14.0 from https://files.pythonhosted.org/packages/19/3b/a0a807b957c5f8f2075a8740f597a021420583bb0aa68ff6432d74a530d4/transformers-5.14.0-py3-none-any.whl.metadata + Downloading transformers-5.14.0-py3-none-any.whl.metadata (32 kB) +Collecting click<9.0.0,>=8.4.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for click<9.0.0,>=8.4.2 from https://files.pythonhosted.org/packages/fb/e2/79c688af8b210d232694e31e59da9f6ec747bae31c3f5946e4e9b98860d5/click-8.4.2-py3-none-any.whl.metadata + Downloading click-8.4.2-py3-none-any.whl.metadata (2.6 kB) +Requirement already satisfied: filelock>=3.10.0 in /opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (3.29.0) +Requirement already satisfied: fsspec>=2023.5.0 in /opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (2026.4.0) +Collecting hf-xet<2.0.0,>=1.5.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for hf-xet<2.0.0,>=1.5.2 from https://files.pythonhosted.org/packages/67/4e/a28359bf1c1ecf11eba22123168c138698f7cb576ac678f5a2e16cd5da08/hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata + Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (4.9 kB) +Collecting httpx<1,>=0.23.0 (from huggingface-hub==1.27.0) + Obtaining dependency information for httpx<1,>=0.23.0 from https://files.pythonhosted.org/packages/2a/39/e50c7c3a983047577ee07d2a9e53faf5a69493943ec3f6a384bdc792deb2/httpx-0.28.1-py3-none-any.whl.metadata + Downloading httpx-0.28.1-py3-none-any.whl.metadata (7.1 kB) +Collecting packaging>=20.9 (from huggingface-hub==1.27.0) + Obtaining dependency information for packaging>=20.9 from https://files.pythonhosted.org/packages/63/34/ba1c580383c9eada3711951fef0795c80b829a078d72188184bcab9dd527/packaging-26.3-py3-none-any.whl.metadata + Downloading packaging-26.3-py3-none-any.whl.metadata (3.5 kB) +Collecting pyyaml>=5.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for pyyaml>=5.1 from https://files.pythonhosted.org/packages/8b/9d/b3589d3877982d4f2329302ef98a8026e7f4443c765c46cfecc8858c6b4b/pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.4 kB) +Collecting tqdm>=4.42.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for tqdm>=4.42.1 from https://files.pythonhosted.org/packages/f9/1c/01bfd571a64e7f270e6bab5e33777debe0edc56759233ce84f27dec92d14/tqdm-4.70.0-py3-none-any.whl.metadata + Downloading tqdm-4.70.0-py3-none-any.whl.metadata (57 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 57.3/57.3 kB 12.3 MB/s eta 0:00:00 +Requirement already satisfied: typing-extensions>=4.1.0 in /opt/qwen35-acceptance-v2/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (4.15.0) +Collecting regex>=2025.10.22 (from transformers==5.14.0) + Obtaining dependency information for regex>=2025.10.22 from https://files.pythonhosted.org/packages/2a/8e/096d00c7c480ef2ff4265349b14e2261d4ab787ba1f74e2e80d1c58079c3/regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (40 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 40.5/40.5 kB 11.7 MB/s eta 0:00:00 +Collecting tokenizers<=0.23.0,>=0.22.0 (from transformers==5.14.0) + Obtaining dependency information for tokenizers<=0.23.0,>=0.22.0 from https://files.pythonhosted.org/packages/2e/76/932be4b50ef6ccedf9d3c6639b056a967a86258c6d9200643f01269211ca/tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (7.3 kB) +Collecting typer (from transformers==5.14.0) + Obtaining dependency information for typer from https://files.pythonhosted.org/packages/43/89/9518bc0c3929bee36b3a4a8e3daddd6e03f92f9961c66d4983b837160543/typer-0.27.1-py3-none-any.whl.metadata + Downloading typer-0.27.1-py3-none-any.whl.metadata (16 kB) +Collecting anyio (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for anyio from https://files.pythonhosted.org/packages/da/35/f2287558c17e29fafc8ef3daf819bb9834061cfa43bff8014f7df7f63bdc/anyio-4.14.2-py3-none-any.whl.metadata + Downloading anyio-4.14.2-py3-none-any.whl.metadata (4.6 kB) +Collecting certifi (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for certifi from https://files.pythonhosted.org/packages/0b/a7/71ac2cff56fec219ed242bb11b8efb69fcc4bec75db06fb7bfe35de520e6/certifi-2026.7.22-py3-none-any.whl.metadata + Downloading certifi-2026.7.22-py3-none-any.whl.metadata (2.5 kB) +Collecting httpcore==1.* (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for httpcore==1.* from https://files.pythonhosted.org/packages/7e/f5/f66802a942d491edb555dd61e3a9961140fd64c90bce1eafd741609d334d/httpcore-1.0.9-py3-none-any.whl.metadata + Downloading httpcore-1.0.9-py3-none-any.whl.metadata (21 kB) +Collecting idna (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for idna from https://files.pythonhosted.org/packages/1e/5e/d4e9f1a599fb8e573b7b87160658329fbf28d19eac2718f51fc3def3aa5a/idna-3.18-py3-none-any.whl.metadata + Downloading idna-3.18-py3-none-any.whl.metadata (6.1 kB) +Collecting h11>=0.16 (from httpcore==1.*->httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for h11>=0.16 from https://files.pythonhosted.org/packages/04/4b/29cac41a4d98d144bf5f6d33995617b185d14b22401f75ca86f384e87ff1/h11-0.16.0-py3-none-any.whl.metadata + Downloading h11-0.16.0-py3-none-any.whl.metadata (8.3 kB) +Collecting shellingham>=1.3.0 (from typer->transformers==5.14.0) + Obtaining dependency information for shellingham>=1.3.0 from https://files.pythonhosted.org/packages/e0/f9/0595336914c5619e5f28a1fb793285925a8cd4b432c9da0a987836c7f822/shellingham-1.5.4-py2.py3-none-any.whl.metadata + Downloading shellingham-1.5.4-py2.py3-none-any.whl.metadata (3.5 kB) +Collecting rich>=13.8.0 (from typer->transformers==5.14.0) + Obtaining dependency information for rich>=13.8.0 from https://files.pythonhosted.org/packages/82/3b/64d4899d73f91ba49a8c18a8ff3f0ea8f1c1d75481760df8c68ef5235bf5/rich-15.0.0-py3-none-any.whl.metadata + Downloading rich-15.0.0-py3-none-any.whl.metadata (18 kB) +Collecting annotated-doc>=0.0.2 (from typer->transformers==5.14.0) + Obtaining dependency information for annotated-doc>=0.0.2 from https://files.pythonhosted.org/packages/3e/30/e900b21425a860e195f32e37657aa1f7c7f2b1bfb26f03ca209b90933c06/annotated_doc-0.0.5-py3-none-any.whl.metadata + Downloading annotated_doc-0.0.5-py3-none-any.whl.metadata (6.5 kB) +Collecting markdown-it-py>=2.2.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for markdown-it-py>=2.2.0 from https://files.pythonhosted.org/packages/b3/81/4da04ced5a082363ecfa159c010d200ecbd959ae410c10c0264a38cac0f5/markdown_it_py-4.2.0-py3-none-any.whl.metadata + Downloading markdown_it_py-4.2.0-py3-none-any.whl.metadata (7.4 kB) +Collecting pygments<3.0.0,>=2.13.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for pygments<3.0.0,>=2.13.0 from https://files.pythonhosted.org/packages/f4/7e/a72dd26f3b0f4f2bf1dd8923c85f7ceb43172af56d63c7383eb62b332364/pygments-2.20.0-py3-none-any.whl.metadata + Downloading pygments-2.20.0-py3-none-any.whl.metadata (2.5 kB) +Collecting mdurl~=0.1 (from markdown-it-py>=2.2.0->rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for mdurl~=0.1 from https://files.pythonhosted.org/packages/b3/38/89ba8ad64ae25be8de66a6d463314cf1eb366222074cfda9ee839c56a4b4/mdurl-0.1.2-py3-none-any.whl.metadata + Downloading mdurl-0.1.2-py3-none-any.whl.metadata (1.6 kB) +Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (16.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 16.6/16.6 MB 138.4 MB/s eta 0:00:00 +Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.9 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.9/6.9 MB 234.3 MB/s eta 0:00:00 +Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (516 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 516.0/516.0 kB 119.5 MB/s eta 0:00:00 +Downloading huggingface_hub-1.27.0-py3-none-any.whl (784 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 784.9/784.9 kB 130.6 MB/s eta 0:00:00 +Downloading transformers-5.14.0-py3-none-any.whl (11.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 11.6/11.6 MB 115.1 MB/s eta 0:00:00 +Downloading click-8.4.2-py3-none-any.whl (119 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 119.2/119.2 kB 40.9 MB/s eta 0:00:00 +Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (4.5 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 4.5/4.5 MB 231.2 MB/s eta 0:00:00 +Downloading httpx-0.28.1-py3-none-any.whl (73 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 73.5/73.5 kB 27.6 MB/s eta 0:00:00 +Downloading httpcore-1.0.9-py3-none-any.whl (78 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 78.8/78.8 kB 24.6 MB/s eta 0:00:00 +Downloading packaging-26.3-py3-none-any.whl (129 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 130.0/130.0 kB 40.2 MB/s eta 0:00:00 +Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (807 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 807.9/807.9 kB 163.7 MB/s eta 0:00:00 +Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (801 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 801.8/801.8 kB 177.2 MB/s eta 0:00:00 +Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (3.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 3.3/3.3 MB 261.5 MB/s eta 0:00:00 +Downloading tqdm-4.70.0-py3-none-any.whl (80 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 80.2/80.2 kB 30.4 MB/s eta 0:00:00 +Downloading typer-0.27.1-py3-none-any.whl (122 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 122.9/122.9 kB 41.7 MB/s eta 0:00:00 +Downloading annotated_doc-0.0.5-py3-none-any.whl (5.3 kB) +Downloading rich-15.0.0-py3-none-any.whl (310 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 310.7/310.7 kB 72.0 MB/s eta 0:00:00 +Downloading shellingham-1.5.4-py2.py3-none-any.whl (9.8 kB) +Downloading anyio-4.14.2-py3-none-any.whl (125 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 125.8/125.8 kB 40.3 MB/s eta 0:00:00 +Downloading idna-3.18-py3-none-any.whl (65 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 65.5/65.5 kB 22.8 MB/s eta 0:00:00 +Downloading certifi-2026.7.22-py3-none-any.whl (136 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 137.0/137.0 kB 47.3 MB/s eta 0:00:00 +Downloading h11-0.16.0-py3-none-any.whl (37 kB) +Downloading markdown_it_py-4.2.0-py3-none-any.whl (91 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 91.7/91.7 kB 32.8 MB/s eta 0:00:00 +Downloading pygments-2.20.0-py3-none-any.whl (1.2 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.2/1.2 MB 200.2 MB/s eta 0:00:00 +Downloading mdurl-0.1.2-py3-none-any.whl (10.0 kB) +Installing collected packages: tqdm, shellingham, safetensors, regex, pyyaml, pygments, pillow, packaging, numpy, mdurl, idna, hf-xet, h11, click, certifi, annotated-doc, markdown-it-py, httpcore, anyio, rich, httpx, typer, huggingface-hub, tokenizers, transformers +Successfully installed annotated-doc-0.0.5 anyio-4.14.2 certifi-2026.7.22 click-8.4.2 h11-0.16.0 hf-xet-1.6.0 httpcore-1.0.9 httpx-0.28.1 huggingface-hub-1.27.0 idna-3.18 markdown-it-py-4.2.0 mdurl-0.1.2 numpy-2.4.6 packaging-26.3 pillow-12.3.0 pygments-2.20.0 pyyaml-6.0.3 regex-2026.7.19 rich-15.0.0 safetensors-0.8.0 shellingham-1.5.4 tokenizers-0.22.2 tqdm-4.70.0 transformers-5.14.0 typer-0.27.1 + +[notice] A new release of pip is available: 23.2.1 -> 26.2.1 +[notice] To update, run: python3.12 -m pip install --upgrade pip +No broken requirements found. +torch 2.11.0+cpu +transformers 5.14.0 +AutoProcessor Qwen3_5ForConditionalGeneration + Fetching 13 files: 0%| | 0/13 [00:00 Date: Thu, 13 Aug 2026 05:06:36 +0000 Subject: [PATCH 10/17] docs: refresh generated docs (REFERENCE, CAPABILITIES, API_QUICKREF, PIPELINE_MAP, FACULTY_MAP, DOC_MAP) [skip ci] --- REFERENCE.md | 278 +++++++++++++++++++++++++-------------------------- 1 file changed, 139 insertions(+), 139 deletions(-) diff --git a/REFERENCE.md b/REFERENCE.md index 164a4cd..e06473d 100644 --- a/REFERENCE.md +++ b/REFERENCE.md @@ -76,8 +76,8 @@ | module | what it is | lines | |---|---|---| | [`holographic_session.py`](#holographic-session) | holographic_session.py -- ONE render session that ties the disconnected rendering threads together. | 227 | -| [`holographic_session.py`](#holographic-session) | SESSION STORE -- contexts that outlive the process. | 261 | | [`holographic_session.py`](#holographic-session) | SESSION -- never compute the same conversation prefix twice. | 253 | +| [`holographic_session.py`](#holographic-session) | SESSION STORE -- contexts that outlive the process. | 261 | ### `splat*` family (6) @@ -423,8 +423,8 @@ | [`holographic_materialio.py`](#holographic-materialio) | PBR material interchange (G2-IO): the standard-format bridge for materials, and a VSA-native carrier. | 307 | | [`holographic_matlib.py`](#holographic-matlib) | holographic_matlib.py -- a comprehensive RENDER material library: plain diffuse -> a fractal planet. | 801 | | [`holographic_meaning_predict.py`](#holographic-meaning-predict) | Generation with structure: predict a next-MEANING vector and settle it, rather | 198 | -| [`holographic_measure.py`](#holographic-measure) | MEASURE -- perplexity with error bars, and decisions that respect them. | 239 | | [`holographic_measure.py`](#holographic-measure) | The variance harness: every headline number gets a mean, a spread, and a confidence | 187 | +| [`holographic_measure.py`](#holographic-measure) | MEASURE -- perplexity with error bars, and decisions that respect them. | 239 | | [`holographic_memory.py`](#holographic-memory) | MEMORY -- the Galvatron's own store, built on leCore's holographic database. | 241 | | [`holographic_memoryhome.py`](#holographic-memoryhome) | holographic_memoryhome.py -- the MEMORY home (consolidation backlog H6): keep the hot working set where the CP | 165 | | [`holographic_memsearch.py`](#holographic-memsearch) | MEMSEARCH -- searchable memory that lives in the weights and runs in the pass. | 212 | @@ -577,8 +577,8 @@ | [`holographic_rolling.py`](#holographic-rolling) | holographic_rolling.py -- the CAUSAL rolling / streaming statistics kit: trailing mean, std, min, max, | 378 | | [`holographic_roomacoustic.py`](#holographic-roomacoustic) | holographic_roomacoustic.py -- A6: GEOMETRIC ROOM ACOUSTICS. How a room echoes -- reflections and reverb. | 161 | | [`holographic_route.py`](#holographic-route) | Representation routing (ARCH-7): route each operation to the representation that supports it -- the policy lay | 183 | -| [`holographic_router.py`](#holographic-router) | Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. | 173 | | [`holographic_router.py`](#holographic-router) | ROUTER -- the model DECIDING, inside one forward pass. | 229 | +| [`holographic_router.py`](#holographic-router) | Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. | 173 | | [`holographic_sampler.py`](#holographic-sampler) | holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). | 223 | | [`holographic_sampling.py`](#holographic-sampling) | Blue-noise / Poisson-disk point sampling -- the EXCLUSION principle, done right. | 134 | | [`holographic_samplinghome.py`](#holographic-samplinghome) | holographic_samplinghome.py -- the SAMPLING home (consolidation backlog R4): one place for the Monte-Carlo | 127 | @@ -683,8 +683,8 @@ | [`holographic_toolclient.py`](#holographic-toolclient) | holographic_toolclient.py -- call another node the same way leCore is called. | 98 | | [`holographic_topology.py`](#holographic-topology) | Principled topology by persistent homology (EXP-7). | 288 | | [`holographic_transfer.py`](#holographic-transfer) | holographic_transfer.py -- KERNEL SCATTER / GATHER: the ONE bundle/readout under every particle<->grid transfe | 224 | -| [`holographic_transform.py`](#holographic-transform) | TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. | 189 | | [`holographic_transform.py`](#holographic-transform) | holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). | 297 | +| [`holographic_transform.py`](#holographic-transform) | TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. | 189 | | [`holographic_transform_space.py`](#holographic-transform-space) | holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it | 176 | | [`holographic_transformbank.py`](#holographic-transformbank) | holographic_transformbank.py -- a prebuilt map of hypervector transforms, and what it can and cannot hold. | 358 | | [`holographic_transformhome.py`](#holographic-transformhome) | holographic_transformhome.py -- the TRANSFORM home (consolidation backlog H5): one facade over "move / rotate | 223 | @@ -14848,6 +14848,39 @@ ### holographic_measure.py +> The variance harness: every headline number gets a mean, a spread, and a confidence +> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. +> +> WHY this exists +> --------------- +> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's +> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A +> single-seed score is therefore a sample from a distribution, and reporting it alone hides +> how wide that distribution is. For an engine whose entire pitch is "measured, not +> promised," reporting a number without its noise is the sharpest blind spot -- so this +> points that same discipline at the numbers themselves. +> +> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the +> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no +> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI +> bound clears the floor -- which is what stops a single fortunate seed from passing a +> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". +> +> USE REAL DATA. The point of the harness is to characterise the real distribution of a +> real claim; running it on a toy makes the spread meaningless. The measurements wired +> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). + +**Public API:** + +- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. +- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. +- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. +- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is +- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above +- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a + +### holographic_measure.py + > MEASURE -- perplexity with error bars, and decisions that respect them. > > Moose asked what assimilation is actually doing. The answer, from his own run: @@ -14880,39 +14913,6 @@ - `def better_than(a, b, alpha, seed, resamples)` -- Is model A better than model B, or is the difference undecidable? - `def tokens_needed(reference, effect_pct, alpha)` -- How many tokens would be needed to RESOLVE an effect of this size. -### holographic_measure.py - -> The variance harness: every headline number gets a mean, a spread, and a confidence -> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. -> -> WHY this exists -> --------------- -> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's -> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A -> single-seed score is therefore a sample from a distribution, and reporting it alone hides -> how wide that distribution is. For an engine whose entire pitch is "measured, not -> promised," reporting a number without its noise is the sharpest blind spot -- so this -> points that same discipline at the numbers themselves. -> -> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the -> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no -> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI -> bound clears the floor -- which is what stops a single fortunate seed from passing a -> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". -> -> USE REAL DATA. The point of the harness is to characterise the real distribution of a -> real claim; running it on a toy makes the spread meaningless. The measurements wired -> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). - -**Public API:** - -- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. -- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. -- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. -- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is -- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above -- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a - ### holographic_memory.py > MEMORY -- the Galvatron's own store, built on leCore's holographic database. @@ -22656,33 +22656,6 @@ ### holographic_router.py -> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. -> -> WHY THIS EXISTS (measured, backlog N9/N28): -> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, -> but it has no notion of MEANING: "squish a big array down for storage" shares no token with -> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics -> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median -> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. -> -> WHAT SHIPS (and what does NOT): -> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted -> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a -> query it already has a VECTOR for: -> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) -> * a caller that supplies its own query vector (an app that ran the encoder itself) -> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this -> router says so and returns None, so the caller falls back to the token router rather than guessing. That -> honesty is the whole point: silence beats a confident wrong route. -> -> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. - -**Public API:** - -- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. - -### holographic_router.py - > ROUTER -- the model DECIDING, inside one forward pass. > > Moose raised the architecture that dissolves the wall this project kept hitting: @@ -22722,6 +22695,33 @@ - `def route(runtime, router, text, tokenize)` -- Would this model choose to use the capability on this prompt? - `def install_routed(weights, cfg, operator, router, layer, gain, temperature)` -- Install a circuit whose GATE is the router, not a constant. +### holographic_router.py + +> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. +> +> WHY THIS EXISTS (measured, backlog N9/N28): +> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, +> but it has no notion of MEANING: "squish a big array down for storage" shares no token with +> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics +> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median +> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. +> +> WHAT SHIPS (and what does NOT): +> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted +> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a +> query it already has a VECTOR for: +> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) +> * a caller that supplies its own query vector (an app that ran the encoder itself) +> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this +> router says so and returns None, so the caller falls back to the token router rather than guessing. That +> honesty is the whole point: silence beats a confident wrong route. +> +> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. + +**Public API:** + +- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. + ### holographic_sampler.py > holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). @@ -24352,6 +24352,39 @@ ### holographic_session.py +> SESSION -- never compute the same conversation prefix twice. +> +> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a +> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS +> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens +> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS +> REPEATED, and the fraction grows with every turn. +> +> leCore already had the pieces and never joined them: the runtime exposes +> prefill/step over an InferenceState with copy(), galvacache memoises pure +> functions at a measured 75% hit rate with bit-identical output, and sessions +> exist. What was missing is the RADIX TREE -- the structure that answers "what is +> the longest prefix of this prompt that I have already computed?" +> +> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three +> turns with turn 3, and a dictionary keyed on the whole prompt misses that +> completely. vLLM and SGLang call this RadixAttention; the idea is the same +> whatever the model: index by prefix, resume from the deepest match, compute only +> the tail. +> +> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no +> cache: resuming from a cached state reproduces a full recompute TO FLOAT +> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the +> difference is real rather than pedantic: resuming STEPS the tail while a fresh +> call PREFILLS it, and the two associate their sums differently. The selftest +> asserts the measured bound against the live runtime rather than assuming it. + +**Public API:** + +- `class PrefixCache` -- A radix tree over token sequences, holding inference states. + +### holographic_session.py + > SESSION STORE -- contexts that outlive the process. > > A Galvatron's context is not a transcript, it is STATE: the GDN recurrent @@ -24384,39 +24417,6 @@ - `class SessionStore` -- Named, persistent, independent contexts on disk. - `def runtime_fingerprint(runtime)` -- A cheap, deterministic id for the checkpoint behind a runtime, so a -### holographic_session.py - -> SESSION -- never compute the same conversation prefix twice. -> -> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a -> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS -> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens -> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS -> REPEATED, and the fraction grows with every turn. -> -> leCore already had the pieces and never joined them: the runtime exposes -> prefill/step over an InferenceState with copy(), galvacache memoises pure -> functions at a measured 75% hit rate with bit-identical output, and sessions -> exist. What was missing is the RADIX TREE -- the structure that answers "what is -> the longest prefix of this prompt that I have already computed?" -> -> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three -> turns with turn 3, and a dictionary keyed on the whole prompt misses that -> completely. vLLM and SGLang call this RadixAttention; the idea is the same -> whatever the model: index by prefix, resume from the deepest match, compute only -> the tail. -> -> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no -> cache: resuming from a cached state reproduces a full recompute TO FLOAT -> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the -> difference is real rather than pedantic: resuming STEPS the tail while a fresh -> call PREFILLS it, and the two associate their sums differently. The selftest -> asserts the measured bound against the live runtime rather than assuming it. - -**Public API:** - -- `class PrefixCache` -- A radix tree over token sequences, holding inference states. - ### holographic_shader.py > holographic_shader.py -- N filter passes in ONE evaluation. Two things a GPU structurally cannot do. @@ -27945,6 +27945,48 @@ ### holographic_transform.py +> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). +> +> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's +> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need +> in one place. This gathers the standard, well-known math: +> +> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, +> and what a property panel shows as the T/R/S fields; +> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); +> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector +> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); +> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. +> +> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN +> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit +> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL +> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is +> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. +> Deterministic; NumPy + stdlib only. + +**Public API:** + +- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. +- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). +- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). +- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. +- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear +- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. +- `def quat_normalize(q)` +- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". +- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. +- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. +- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. +- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for +- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). +- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock +- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an +- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. +- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks + +### holographic_transform.py + > TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. > > Everything before this applied leCore's levers uniformly: grow a memory channel @@ -27987,48 +28029,6 @@ - `def plan(weights, cfg, target_tokens, kv_rank, grow_gain)` -- Decide what to do to each layer, from the analysis rather than by rule. - `def apply_plan(weights, cfg, the_plan, progress)` -- Carry out the growth actions. KV compression is a RUNTIME setting and is -### holographic_transform.py - -> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). -> -> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's -> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need -> in one place. This gathers the standard, well-known math: -> -> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, -> and what a property panel shows as the T/R/S fields; -> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); -> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector -> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); -> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. -> -> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN -> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit -> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL -> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is -> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. -> Deterministic; NumPy + stdlib only. - -**Public API:** - -- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. -- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). -- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). -- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. -- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear -- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. -- `def quat_normalize(q)` -- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". -- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. -- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. -- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. -- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for -- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). -- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock -- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an -- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. -- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks - ### holographic_transform_space.py > holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it From 75738ee51fcb29cb60aad47b081a5d03efe9bf3d Mon Sep 17 00:00:00 2001 From: docs-bot Date: Thu, 13 Aug 2026 13:17:26 +0000 Subject: [PATCH 11/17] docs: refresh generated docs (REFERENCE, CAPABILITIES, API_QUICKREF, PIPELINE_MAP, FACULTY_MAP, DOC_MAP) [skip ci] --- REFERENCE.md | 428 +++++++++++++++++++++++------------------------ docs/UNIFIERS.md | 4 +- 2 files changed, 216 insertions(+), 216 deletions(-) diff --git a/REFERENCE.md b/REFERENCE.md index 4781efd..6e32bdf 100644 --- a/REFERENCE.md +++ b/REFERENCE.md @@ -75,9 +75,9 @@ | module | what it is | lines | |---|---|---| -| [`holographic_session.py`](#holographic-session) | SESSION -- never compute the same conversation prefix twice. | 253 | | [`holographic_session.py`](#holographic-session) | holographic_session.py -- ONE render session that ties the disconnected rendering threads together. | 227 | | [`holographic_session.py`](#holographic-session) | SESSION STORE -- contexts that outlive the process. | 360 | +| [`holographic_session.py`](#holographic-session) | SESSION -- never compute the same conversation prefix twice. | 253 | ### `splat*` family (6) @@ -423,8 +423,8 @@ | [`holographic_materialio.py`](#holographic-materialio) | PBR material interchange (G2-IO): the standard-format bridge for materials, and a VSA-native carrier. | 307 | | [`holographic_matlib.py`](#holographic-matlib) | holographic_matlib.py -- a comprehensive RENDER material library: plain diffuse -> a fractal planet. | 801 | | [`holographic_meaning_predict.py`](#holographic-meaning-predict) | Generation with structure: predict a next-MEANING vector and settle it, rather | 198 | -| [`holographic_measure.py`](#holographic-measure) | The variance harness: every headline number gets a mean, a spread, and a confidence | 187 | | [`holographic_measure.py`](#holographic-measure) | MEASURE -- perplexity with error bars, and decisions that respect them. | 239 | +| [`holographic_measure.py`](#holographic-measure) | The variance harness: every headline number gets a mean, a spread, and a confidence | 187 | | [`holographic_memory.py`](#holographic-memory) | MEMORY -- the Galvatron's own store, built on leCore's holographic database. | 241 | | [`holographic_memoryhome.py`](#holographic-memoryhome) | holographic_memoryhome.py -- the MEMORY home (consolidation backlog H6): keep the hot working set where the CP | 165 | | [`holographic_memsearch.py`](#holographic-memsearch) | MEMSEARCH -- searchable memory that lives in the weights and runs in the pass. | 212 | @@ -543,8 +543,8 @@ | [`holographic_realtime.py`](#holographic-realtime) | holographic_realtime.py -- the realtime render loop, and the multi-format payload it pushes. | 404 | | [`holographic_reanchor.py`](#holographic-reanchor) | Re-anchoring is load-bearing for deep traversal -- the audit, and the contrast the other tests don't show. | 97 | | [`holographic_reasoning.py`](#holographic-reasoning) | holographic_reasoning.py | 397 | -| [`holographic_recipe.py`](#holographic-recipe) | A generative recipe-store for constructed holostuff structures. | 206 | | [`holographic_recipe.py`](#holographic-recipe) | RECIPE -- ship what leCore ADDED, not the model it was added to. | 334 | +| [`holographic_recipe.py`](#holographic-recipe) | A generative recipe-store for constructed holostuff structures. | 206 | | [`holographic_recipeops.py`](#holographic-recipeops) | StructureRecipe validator + edit operators (ARCH-1): the recipe equivalent of the mesh Euler operators. | 264 | | [`holographic_reclock.py`](#holographic-reclock) | holographic_reclock.py -- sample when an AXIS moves, not when time passes ("make the boring property the | 313 | | [`holographic_recurrent.py`](#holographic-recurrent) | A gradient-free RECURRENT layer for the holographic engine: reservoir computing. | 456 | @@ -578,8 +578,8 @@ | [`holographic_rolling.py`](#holographic-rolling) | holographic_rolling.py -- the CAUSAL rolling / streaming statistics kit: trailing mean, std, min, max, | 378 | | [`holographic_roomacoustic.py`](#holographic-roomacoustic) | holographic_roomacoustic.py -- A6: GEOMETRIC ROOM ACOUSTICS. How a room echoes -- reflections and reverb. | 161 | | [`holographic_route.py`](#holographic-route) | Representation routing (ARCH-7): route each operation to the representation that supports it -- the policy lay | 183 | -| [`holographic_router.py`](#holographic-router) | ROUTER -- the model DECIDING, inside one forward pass. | 229 | | [`holographic_router.py`](#holographic-router) | Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. | 173 | +| [`holographic_router.py`](#holographic-router) | ROUTER -- the model DECIDING, inside one forward pass. | 229 | | [`holographic_sampler.py`](#holographic-sampler) | holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). | 223 | | [`holographic_sampling.py`](#holographic-sampling) | Blue-noise / Poisson-disk point sampling -- the EXCLUSION principle, done right. | 134 | | [`holographic_samplinghome.py`](#holographic-samplinghome) | holographic_samplinghome.py -- the SAMPLING home (consolidation backlog R4): one place for the Monte-Carlo | 127 | @@ -622,8 +622,8 @@ | [`holographic_skymodel.py`](#holographic-skymodel) | holographic_skymodel.py -- a PARAMETRIC sky: time of day, sun, moon, stars, and HIGH cloud layers, as | 450 | | [`holographic_slime.py`](#holographic-slime) | Slime-mold path-finding over a HOLOGRAPHIC associative graph. | 391 | | [`holographic_smokepresets.py`](#holographic-smokepresets) | holographic_smokepresets.py -- SMOKE PRESETS (fluids/matter backlog, content item 1). | 159 | -| [`holographic_snap.py`](#holographic-snap) | holographic_snap.py (mesh_and_geometry) -- the MODELING-GIZMO snap adapter: it answers 'where does this dragge | 184 | | [`holographic_snap.py`](#holographic-snap) | holographic_snap.py -- SNAPPING = cleanup, applied to geometry (modeling-app feature layer). | 141 | +| [`holographic_snap.py`](#holographic-snap) | holographic_snap.py (mesh_and_geometry) -- the MODELING-GIZMO snap adapter: it answers 'where does this dragge | 184 | | [`holographic_softbody.py`](#holographic-softbody) | Position-Based Dynamics -- softbody & hardbody simulation, exposed to VSA. | 680 | | [`holographic_sparsefield.py`](#holographic-sparsefield) | FS-2 -- the narrow-band sparse field (holographic_sparsefield), array-backed for parallelism. | 530 | | [`holographic_spatial.py`](#holographic-spatial) | holographic_spatial.py -- ONE shared spatial index. Bin points into a uniform grid of cells so radius, | 191 | @@ -648,8 +648,8 @@ | [`holographic_subdivcurve.py`](#holographic-subdivcurve) | Subdivision curves on hypervector sequences (ARCH-5): Loop subdivision (FWD-8), turned inward. | 148 | | [`holographic_substrate.py`](#holographic-substrate) | SUBSTRATE -- the model's weight surface as a storage medium. | 694 | | [`holographic_supermemory.py`](#holographic-supermemory) | Superposed key-value memory with a CLOSED-FORM capacity law, a single-shot allocator, | 634 | -| [`holographic_superposed.py`](#holographic-superposed) | holographic_superposed.py -- parallel computation in superposition (the WIDTH faculty). | 363 | | [`holographic_superposed.py`](#holographic-superposed) | COMPATIBILITY SHIM -- this module moved to `holographic_supermemory`. | 51 | +| [`holographic_superposed.py`](#holographic-superposed) | holographic_superposed.py -- parallel computation in superposition (the WIDTH faculty). | 363 | | [`holographic_superres.py`](#holographic-superres) | holographic_superres.py -- EXAMPLE-BASED SUPER-RESOLUTION / GUIDED UPSAMPLING (inverse-rendering ST3). | 83 | | [`holographic_superschedule.py`](#holographic-superschedule) | holographic_superschedule.py -- Fill 3: AUTO-SUPERPOSITION + SPILL. The latency-hiding move: hold N | 225 | | [`holographic_surface.py`](#holographic-surface) | holographic_surface.py -- the FIRST-CLASS render material: every channel is a Param socket, resolved PER HIT. | 293 | @@ -684,8 +684,8 @@ | [`holographic_toolclient.py`](#holographic-toolclient) | holographic_toolclient.py -- call another node the same way leCore is called. | 98 | | [`holographic_topology.py`](#holographic-topology) | Principled topology by persistent homology (EXP-7). | 288 | | [`holographic_transfer.py`](#holographic-transfer) | holographic_transfer.py -- KERNEL SCATTER / GATHER: the ONE bundle/readout under every particle<->grid transfe | 224 | -| [`holographic_transform.py`](#holographic-transform) | holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). | 297 | | [`holographic_transform.py`](#holographic-transform) | TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. | 189 | +| [`holographic_transform.py`](#holographic-transform) | holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). | 297 | | [`holographic_transform_space.py`](#holographic-transform-space) | holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it | 176 | | [`holographic_transformbank.py`](#holographic-transformbank) | holographic_transformbank.py -- a prebuilt map of hypervector transforms, and what it can and cannot hold. | 358 | | [`holographic_transformhome.py`](#holographic-transformhome) | holographic_transformhome.py -- the TRANSFORM home (consolidation backlog H5): one facade over "move / rotate | 223 | @@ -14849,39 +14849,6 @@ ### holographic_measure.py -> The variance harness: every headline number gets a mean, a spread, and a confidence -> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. -> -> WHY this exists -> --------------- -> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's -> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A -> single-seed score is therefore a sample from a distribution, and reporting it alone hides -> how wide that distribution is. For an engine whose entire pitch is "measured, not -> promised," reporting a number without its noise is the sharpest blind spot -- so this -> points that same discipline at the numbers themselves. -> -> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the -> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no -> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI -> bound clears the floor -- which is what stops a single fortunate seed from passing a -> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". -> -> USE REAL DATA. The point of the harness is to characterise the real distribution of a -> real claim; running it on a toy makes the spread meaningless. The measurements wired -> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). - -**Public API:** - -- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. -- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. -- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. -- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is -- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above -- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a - -### holographic_measure.py - > MEASURE -- perplexity with error bars, and decisions that respect them. > > Moose asked what assimilation is actually doing. The answer, from his own run: @@ -14914,6 +14881,39 @@ - `def better_than(a, b, alpha, seed, resamples)` -- Is model A better than model B, or is the difference undecidable? - `def tokens_needed(reference, effect_pct, alpha)` -- How many tokens would be needed to RESOLVE an effect of this size. +### holographic_measure.py + +> The variance harness: every headline number gets a mean, a spread, and a confidence +> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. +> +> WHY this exists +> --------------- +> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's +> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A +> single-seed score is therefore a sample from a distribution, and reporting it alone hides +> how wide that distribution is. For an engine whose entire pitch is "measured, not +> promised," reporting a number without its noise is the sharpest blind spot -- so this +> points that same discipline at the numbers themselves. +> +> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the +> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no +> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI +> bound clears the floor -- which is what stops a single fortunate seed from passing a +> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". +> +> USE REAL DATA. The point of the harness is to characterise the real distribution of a +> real claim; running it on a toy makes the spread meaningless. The measurements wired +> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). + +**Public API:** + +- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. +- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. +- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. +- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is +- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above +- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a + ### holographic_memory.py > MEMORY -- the Galvatron's own store, built on leCore's holographic database. @@ -21090,42 +21090,6 @@ ### holographic_recipe.py -> A generative recipe-store for constructed holostuff structures. -> -> WHY THIS EXISTS -> --------------- -> The panel's "proven structure has no noise" result: a structure BUILT by a deterministic proof (a -> derivation over a seed) carries no noise, so it serialises to its *generator* losslessly -- you store the -> recipe, not the expanded vectors, and replaying the recipe reproduces the structure BIT-FOR-BIT. This is -> the easy, exact half of generative compression: when we are the builder we already hold the proof, so -> there is nothing to search for and no residual to code. -> -> A `StructureRecipe` is a tiny replayable build-graph. Each op produces one result vector from a seed and -> earlier results: `atom` (a derived atom -- regenerated from the seed, never stored), `bind`, `bundle`, -> `permute`, `normalize`. You build your structure THROUGH the recipe, so you get both the vectors and the -> recipe that regenerates them. Serialising stores only the op list (and the seed) -- a few hundred bytes -> that regenerate megabytes of structure, exactly. -> -> THE ESCAPE HATCH AND THE KEPT NEGATIVE -> The `raw` op stores a literal vector verbatim. It is the honest boundary: data that was NOT constructed -> (a measured or random vector) has no short recipe, so it must be stored as-is and gets no compression. -> The recipe's compression ratio is therefore exactly the *constructed fraction* of the structure -- all -> recipe -> enormous ratio; all raw -> ~1x. That is the constructed-vs-measured partition made literal. -> -> THE CAPACITY-CLIFF POINT -> Reading structure back out of a single bounded encoded vector degrades past the capacity cliff (crosstalk). -> A recipe does not: it names its leaves explicitly and replays the construction, so a deeply nested -> structure is recovered EXACTLY at any depth. The recipe is the right store for deep constructed structure; -> the expanded superposition is bounded. -> -> Pure NumPy + holostuff kernel, deterministic, JSON serialisation (readable), no new dependencies. - -**Public API:** - -- `class StructureRecipe` -- A replayable build-graph: store the deterministic ops that built a structure, not the vectors. - -### holographic_recipe.py - > RECIPE -- ship what leCore ADDED, not the model it was added to. > > Moose: the inflated model size needs looking at holographically -- we should be @@ -21178,6 +21142,42 @@ - `def cost(rules, arrays, installed_weights)` -- What the recipe saves, in bytes. The number is the whole argument. - `def expand(rules, arrays, base_weights)` -- Rebuild the installed model from the recipe. Must be byte-exact. +### holographic_recipe.py + +> A generative recipe-store for constructed holostuff structures. +> +> WHY THIS EXISTS +> --------------- +> The panel's "proven structure has no noise" result: a structure BUILT by a deterministic proof (a +> derivation over a seed) carries no noise, so it serialises to its *generator* losslessly -- you store the +> recipe, not the expanded vectors, and replaying the recipe reproduces the structure BIT-FOR-BIT. This is +> the easy, exact half of generative compression: when we are the builder we already hold the proof, so +> there is nothing to search for and no residual to code. +> +> A `StructureRecipe` is a tiny replayable build-graph. Each op produces one result vector from a seed and +> earlier results: `atom` (a derived atom -- regenerated from the seed, never stored), `bind`, `bundle`, +> `permute`, `normalize`. You build your structure THROUGH the recipe, so you get both the vectors and the +> recipe that regenerates them. Serialising stores only the op list (and the seed) -- a few hundred bytes +> that regenerate megabytes of structure, exactly. +> +> THE ESCAPE HATCH AND THE KEPT NEGATIVE +> The `raw` op stores a literal vector verbatim. It is the honest boundary: data that was NOT constructed +> (a measured or random vector) has no short recipe, so it must be stored as-is and gets no compression. +> The recipe's compression ratio is therefore exactly the *constructed fraction* of the structure -- all +> recipe -> enormous ratio; all raw -> ~1x. That is the constructed-vs-measured partition made literal. +> +> THE CAPACITY-CLIFF POINT +> Reading structure back out of a single bounded encoded vector degrades past the capacity cliff (crosstalk). +> A recipe does not: it names its leaves explicitly and replays the construction, so a deeply nested +> structure is recovered EXACTLY at any depth. The recipe is the right store for deep constructed structure; +> the expanded superposition is bounded. +> +> Pure NumPy + holostuff kernel, deterministic, JSON serialisation (readable), no new dependencies. + +**Public API:** + +- `class StructureRecipe` -- A replayable build-graph: store the deterministic ops that built a structure, not the vectors. + ### holographic_recipeops.py > StructureRecipe validator + edit operators (ARCH-1): the recipe equivalent of the mesh Euler operators. @@ -22711,6 +22711,33 @@ ### holographic_router.py +> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. +> +> WHY THIS EXISTS (measured, backlog N9/N28): +> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, +> but it has no notion of MEANING: "squish a big array down for storage" shares no token with +> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics +> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median +> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. +> +> WHAT SHIPS (and what does NOT): +> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted +> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a +> query it already has a VECTOR for: +> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) +> * a caller that supplies its own query vector (an app that ran the encoder itself) +> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this +> router says so and returns None, so the caller falls back to the token router rather than guessing. That +> honesty is the whole point: silence beats a confident wrong route. +> +> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. + +**Public API:** + +- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. + +### holographic_router.py + > ROUTER -- the model DECIDING, inside one forward pass. > > Moose raised the architecture that dissolves the wall this project kept hitting: @@ -22750,33 +22777,6 @@ - `def route(runtime, router, text, tokenize)` -- Would this model choose to use the capability on this prompt? - `def install_routed(weights, cfg, operator, router, layer, gain, temperature)` -- Install a circuit whose GATE is the router, not a constant. -### holographic_router.py - -> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. -> -> WHY THIS EXISTS (measured, backlog N9/N28): -> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, -> but it has no notion of MEANING: "squish a big array down for storage" shares no token with -> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics -> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median -> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. -> -> WHAT SHIPS (and what does NOT): -> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted -> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a -> query it already has a VECTOR for: -> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) -> * a caller that supplies its own query vector (an app that ran the encoder itself) -> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this -> router says so and returns None, so the caller falls back to the token router rather than guessing. That -> honesty is the whole point: silence beats a confident wrong route. -> -> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. - -**Public API:** - -- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. - ### holographic_sampler.py > holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). @@ -24379,39 +24379,6 @@ ### holographic_session.py -> SESSION -- never compute the same conversation prefix twice. -> -> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a -> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS -> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens -> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS -> REPEATED, and the fraction grows with every turn. -> -> leCore already had the pieces and never joined them: the runtime exposes -> prefill/step over an InferenceState with copy(), galvacache memoises pure -> functions at a measured 75% hit rate with bit-identical output, and sessions -> exist. What was missing is the RADIX TREE -- the structure that answers "what is -> the longest prefix of this prompt that I have already computed?" -> -> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three -> turns with turn 3, and a dictionary keyed on the whole prompt misses that -> completely. vLLM and SGLang call this RadixAttention; the idea is the same -> whatever the model: index by prefix, resume from the deepest match, compute only -> the tail. -> -> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no -> cache: resuming from a cached state reproduces a full recompute TO FLOAT -> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the -> difference is real rather than pedantic: resuming STEPS the tail while a fresh -> call PREFILLS it, and the two associate their sums differently. The selftest -> asserts the measured bound against the live runtime rather than assuming it. - -**Public API:** - -- `class PrefixCache` -- A radix tree over token sequences, holding inference states. - -### holographic_session.py - > holographic_session.py -- ONE render session that ties the disconnected rendering threads together. > > WHY THIS EXISTS (the above/below audit's keystone, CORE_NOTES section 3) @@ -24472,6 +24439,39 @@ - `class SessionStore` -- Named, persistent, independent contexts on disk. - `def runtime_fingerprint(runtime)` -- A cheap, deterministic id for the checkpoint behind a runtime, so a +### holographic_session.py + +> SESSION -- never compute the same conversation prefix twice. +> +> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a +> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS +> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens +> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS +> REPEATED, and the fraction grows with every turn. +> +> leCore already had the pieces and never joined them: the runtime exposes +> prefill/step over an InferenceState with copy(), galvacache memoises pure +> functions at a measured 75% hit rate with bit-identical output, and sessions +> exist. What was missing is the RADIX TREE -- the structure that answers "what is +> the longest prefix of this prompt that I have already computed?" +> +> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three +> turns with turn 3, and a dictionary keyed on the whole prompt misses that +> completely. vLLM and SGLang call this RadixAttention; the idea is the same +> whatever the model: index by prefix, resume from the deepest match, compute only +> the tail. +> +> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no +> cache: resuming from a cached state reproduces a full recompute TO FLOAT +> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the +> difference is real rather than pedantic: resuming STEPS the tail while a fresh +> call PREFILLS it, and the two associate their sums differently. The selftest +> asserts the measured bound against the live runtime rather than assuming it. + +**Public API:** + +- `class PrefixCache` -- A radix tree over token sequences, holding inference states. + ### holographic_shader.py > holographic_shader.py -- N filter passes in ONE evaluation. Two things a GPU structurally cannot do. @@ -25209,6 +25209,28 @@ ### holographic_snap.py +> holographic_snap.py -- SNAPPING = cleanup, applied to geometry (modeling-app feature layer). +> +> Thinking holographically: snapping IS cleanup. VSA cleanup projects a noisy vector onto the nearest CLEAN atom in +> a codebook; snapping projects a dragged, continuous position onto the nearest ALLOWED place -- a grid node, an +> existing vertex, a point on an edge, an angle increment. Same operation, geometric codebook. And just as cleanup +> can REFUSE a weak match (return "no confident atom"), a snap has a TOLERANCE: if nothing allowed is close enough, +> the point is left where it is. That confidence gate is what stops a cursor from teleporting across the screen. +> +> These read raw coordinates (the honest way -- no lossy encoding for something this exact). NumPy + stdlib only; +> deterministic. + +**Public API:** + +- `def snap_to_grid(p, spacing, origin)` -- Snap a point to the nearest grid node -- round each coordinate to the lattice. The simplest cleanup: the +- `def snap_to_points(p, points, tol)` -- Snap to the NEAREST point in a set -- this is literally cleanup (nearest codebook entry). Returns +- `def snap_to_segment(p, a, b)` -- The nearest point on the line SEGMENT a-b (clamped to the endpoints) -- snapping to an edge. +- `def snap_value(x, increment, origin)` -- Snap a scalar to the nearest multiple of `increment` from `origin` -- e.g. a length to 0.25 m steps. +- `def snap_angle(theta, increment)` -- Snap an angle (radians) to the nearest multiple of `increment` -- e.g. rotate in 15-degree steps. +- `class Snapper` -- Snaps a point to the nearest snap target within a tolerance, combining a GRID and a VERTEX set. Whichever + +### holographic_snap.py + > holographic_snap.py (mesh_and_geometry) -- the MODELING-GIZMO snap adapter: it answers 'where does this dragged > point / transform delta actually go?' in the shapes the interactive edit spine wants (dict hit records, a corrected > transform delta), DELEGATING all the actual snap math to the canonical snap primitives in @@ -25234,28 +25256,6 @@ - `def snap_to_edge(point, vertices, edges, max_dist)` -- Snap a point to the nearest point ON any edge, returned as {edge, position, distance, t}, or None if beyond - `def snap_transform_delta(delta, target, increment, moved_point, vertices, edges, origin, max_dist)` -- Snap a TRANSFORM DELTA so the moved point lands on a snap target, and return the corrected delta. This is the -### holographic_snap.py - -> holographic_snap.py -- SNAPPING = cleanup, applied to geometry (modeling-app feature layer). -> -> Thinking holographically: snapping IS cleanup. VSA cleanup projects a noisy vector onto the nearest CLEAN atom in -> a codebook; snapping projects a dragged, continuous position onto the nearest ALLOWED place -- a grid node, an -> existing vertex, a point on an edge, an angle increment. Same operation, geometric codebook. And just as cleanup -> can REFUSE a weak match (return "no confident atom"), a snap has a TOLERANCE: if nothing allowed is close enough, -> the point is left where it is. That confidence gate is what stops a cursor from teleporting across the screen. -> -> These read raw coordinates (the honest way -- no lossy encoding for something this exact). NumPy + stdlib only; -> deterministic. - -**Public API:** - -- `def snap_to_grid(p, spacing, origin)` -- Snap a point to the nearest grid node -- round each coordinate to the lattice. The simplest cleanup: the -- `def snap_to_points(p, points, tol)` -- Snap to the NEAREST point in a set -- this is literally cleanup (nearest codebook entry). Returns -- `def snap_to_segment(p, a, b)` -- The nearest point on the line SEGMENT a-b (clamped to the endpoints) -- snapping to an edge. -- `def snap_value(x, increment, origin)` -- Snap a scalar to the nearest multiple of `increment` from `origin` -- e.g. a length to 0.25 m steps. -- `def snap_angle(theta, increment)` -- Snap an angle (radians) to the nearest multiple of `increment` -- e.g. rotate in 15-degree steps. -- `class Snapper` -- Snaps a point to the nearest snap target within a tolerance, combining a GRID and a VERTEX set. Whichever - ### holographic_softbody.py > Position-Based Dynamics -- softbody & hardbody simulation, exposed to VSA. @@ -26663,6 +26663,20 @@ ### holographic_superposed.py +> COMPATIBILITY SHIM -- this module moved to `holographic_supermemory`. +> +> WHY THE RENAME (Rule-0 lesson, on record in NOTES): a week-old, DIFFERENT module +> already lived at holographic/misc/holographic_superposed.py (leOS-ported "computing +> in superposition"); this one's build audit queried capability phrasings but never +> grepped the basename, so two unrelated modules shared a name across families -- a +> discoverability tax caught by the fuzzy-ask demo answering 'misc' for this module's +> name. The capacity-law memory now lives under its own name; this shim keeps every +> existing import working forever (additive, backward-compatible only). + +*(no public functions or classes -- internal or data-only)* + +### holographic_superposed.py + > holographic_superposed.py -- parallel computation in superposition (the WIDTH faculty). > > PORTED FROM leOS (`superposed_compute.py`, "one processor, many states simultaneously"). @@ -26700,20 +26714,6 @@ - `def hierarchical_recall(S, group_key, leaf_key, chunk_codebook, item_codebook, min_chunk_similarity)` -- Descend one hierarchical superposition with a CLEANUP at the middle level. - `def flat_recall(S, group_key, leaf_key, item_codebook)` -- The BASELINE hierarchical_recall must beat, and the strongest honest one: unbind both roles from the single -### holographic_superposed.py - -> COMPATIBILITY SHIM -- this module moved to `holographic_supermemory`. -> -> WHY THE RENAME (Rule-0 lesson, on record in NOTES): a week-old, DIFFERENT module -> already lived at holographic/misc/holographic_superposed.py (leOS-ported "computing -> in superposition"); this one's build audit queried capability phrasings but never -> grepped the basename, so two unrelated modules shared a name across families -- a -> discoverability tax caught by the fuzzy-ask demo answering 'misc' for this module's -> name. The capacity-law memory now lives under its own name; this shim keeps every -> existing import working forever (additive, backward-compatible only). - -*(no public functions or classes -- internal or data-only)* - ### holographic_superres.py > holographic_superres.py -- EXAMPLE-BASED SUPER-RESOLUTION / GUIDED UPSAMPLING (inverse-rendering ST3). @@ -28000,48 +28000,6 @@ ### holographic_transform.py -> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). -> -> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's -> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need -> in one place. This gathers the standard, well-known math: -> -> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, -> and what a property panel shows as the T/R/S fields; -> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); -> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector -> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); -> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. -> -> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN -> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit -> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL -> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is -> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. -> Deterministic; NumPy + stdlib only. - -**Public API:** - -- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. -- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). -- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). -- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. -- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear -- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. -- `def quat_normalize(q)` -- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". -- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. -- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. -- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. -- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for -- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). -- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock -- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an -- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. -- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks - -### holographic_transform.py - > TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. > > Everything before this applied leCore's levers uniformly: grow a memory channel @@ -28084,6 +28042,48 @@ - `def plan(weights, cfg, target_tokens, kv_rank, grow_gain)` -- Decide what to do to each layer, from the analysis rather than by rule. - `def apply_plan(weights, cfg, the_plan, progress)` -- Carry out the growth actions. KV compression is a RUNTIME setting and is +### holographic_transform.py + +> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). +> +> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's +> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need +> in one place. This gathers the standard, well-known math: +> +> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, +> and what a property panel shows as the T/R/S fields; +> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); +> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector +> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); +> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. +> +> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN +> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit +> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL +> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is +> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. +> Deterministic; NumPy + stdlib only. + +**Public API:** + +- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. +- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). +- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). +- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. +- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear +- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. +- `def quat_normalize(q)` +- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". +- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. +- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. +- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. +- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for +- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). +- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock +- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an +- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. +- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks + ### holographic_transform_space.py > holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it diff --git a/docs/UNIFIERS.md b/docs/UNIFIERS.md index cb656e0..09732a2 100644 --- a/docs/UNIFIERS.md +++ b/docs/UNIFIERS.md @@ -172,13 +172,13 @@ |---|---| | `holographic_fieldhome` | ✅ wired | -## `cachehome.MarginCache (fat margin for a drifting query)` — 0/1 wired +## `cachehome.MarginCache (fat margin for a drifting query)` — 1/1 wired *a drifting query (a camera, a cursor, a recall neighbourhood) should be served from an ENLARGED baked region, not re-keyed exactly. MEASURED on a unit-step 2-D walk of 400 queries: exact-key caching rebuilds 400/400; margin 6.0 rebuilds 20 at 95% hits. Wired into RenderSession.preview(reuse_margin=), where the drifting query is the CAMERA POSE: 20 drifting frames at margin 0.12 give 19 hits, 1 rebuild. THE GATE a caller must pass is NOT a hit-rate target: a hit serves a STALE value, and on a rendered frame the max error saturates at the FIRST reuse (0.5864, a silhouette edge) while the mean creeps 0.0001 -> 0.0051. `suggest_margin_for_error(..., max_abs_error=)` sizes it against the error you cannot tolerate. Measured: on a value that jumps 0->1, a mean-only budget passes margin 0.1929 and serves a completely wrong answer (max error 1.00); the max-error bound stops at 0.094558, and 0.095158 is already catastrophic. The admissible margin is a CLIFF, not a slope.* | client | status | |---|---| -| `holographic_session` | ❌ not wired | +| `holographic_session` | ✅ wired | ## `denoise.soft_relaxation (stiffness in physical units)` — 2/2 wired From a69352a35777d6c8d0fb9d7888f93f21db4e6b89 Mon Sep 17 00:00:00 2001 From: semantic-bot Date: Thu, 13 Aug 2026 14:39:49 +0000 Subject: [PATCH 12/17] semantic: refresh routing index + seed from embed --- lecore_data/routing/index_128d.npz | Bin 236486 -> 236486 bytes 1 file changed, 0 insertions(+), 0 deletions(-) diff --git a/lecore_data/routing/index_128d.npz b/lecore_data/routing/index_128d.npz index 70be9ae00776e47b2e697b42ae63e1585d809ebd..355b2d35976f713eea8a94f33afae423bc60bfd1 100644 GIT binary patch delta 17972 zcmZ{s2V9iL_V<}*un@(Fjo1NEVHXtaiiNco6dSQ22qK7L2T>_2#Rk}c!A7v4vS2rE zj7iiOQ`Bfov81Uf=H4XU6qB2NeSg1a_enOs`QOjy>zp~$&N*{to@XE9ebwIJReOV# zC&e4L-Cw$P@reD++oGcoUDc28iz!d;*?4;A-uXpq-CiqRdu2hcefkHYp1UYg$#+C6 zQKDAg5ncNZ%gEhpb4$-&9b0*0%bMyuYkB##6La&*R^AwM?A3}(nX%G_gf>{7if)cBn?5i=?K3rKh?V zzVm8E)oTO3+aEvYko{uR(g*eppNrpxZ_56}4=Ov{x-j6i${sI1S}-8#g+tqp#g=!s zk1yEJ+cxv)c~$*|XwjiOIWBo`%&04?=7mN0#>REsUKl%V)N5I1!s0fkzISj`Y`@J1 zs@CP%zxqN9a2ILH{I&3TGW)``?Ym2NEgClIP{Op4^VHO@#q;L;Bhf3C_ey>z;lwu` z_O015W?cG$S64sE8ZvY1)ab;vx%pw6(*{mC*K7HJDLvJI2V#W?vOoVoToB%`uO8=i za&2m_)IEDHY+3lu^Ur<%;fX0Lid6A8!oSh$1w-ds&xGvVGBfW{McjsUWB1!HeIsnr zv-;Nj$}{uRGWQ-Ib+M<)c_Yev-Wq(R3@`oFv=XX*SQ z>C>y~?0kEb&D$+Hso;IDb3?|)PJC3gIBL}VSw~({$A1=P@xJ}zpT&NygECL9$i3Ys zd-|-xb2;XsyrfwXNh_CiT{zl$I6LRqruoImQB!gg?Y7@U9j(sD`8)G3U9zd?{t%{m zX-U)0S*I;3-+6kX{lOn%r|?=>bK_sOyp2nX6UrtPZyY@Ey4wCov=$@m&pi@nWt}%i z`l^WTH7Z+bW^qK>q}H-d>V;O`yYug=D^i;$*4q2HYaiDZ8&vap+HjF!kFTeFFT`v1 zC{L|Ih(_wa4K%YyeD-3svG}E^?#lF}W(ul3cSpD6^tXn_r}?ZaTfJh$q@8;^$}%icq8oxu}Sr7tetOAIkS4x z`M5LjtJ5a_adw}o-$d&!($&BwT8hX~N1AA3M5+2;6U|pxl}}S`>affn`^Hb6S6P~r za@Ss!)GKxT3w5@Bu{1JZ&XE$g(OZ_KC$9F0UG#j&EL)5BmQ0U3HRah;<7*b2>^dfA z$%oHWcYgN#h^>2ODpjD)H`SVnh4x!bwY@@U_DpZA+o3eQdjGSV24@v^Po1$bt7Pxy zysqQ2PsNUTzVPb$4FgJ-=e{yO`bw_4`IPpIcvGb}*Sh$>u*y5L|q*X5MY~P{YY_4?`dzEORg*M3Qe_-~!BT*5_huh6q@UlI; zg*HNX{&3=6?4#H6*Ka?vqO^f3ZK*93HL6Z4t)tj!@7zkeCK^Nrpicm8FM^O#RVW9t5i{d zmeC?GY4E&*JrB>UGcPT_e3tA~kTPaf^2*nfpDt6rfm#>8O`{k5w!O4uYweQtGp6R9 z$TAITu{XOUE#=*oo1dQ9Ah;RcH#vM~@tXm$FUw`b> z`m;M0XY8;ich#N{bynq0DvOPdQ7gJ(d%LOfZknH1q|SHK+CRBJDmi9JYGvp3mx>dM z(yG;u-L%g;MPIFqd6aQs-;Dg_>8F-oF3-+MKUi|_h2eenR&-CAI>;*K=7mkK%*vhK zZDdr%Fjcn)@}!pv>7kh$Mx;lrD_S(WZe+-U^AWk@*QmKYwBzCp)uyM`O{`H9dTKqy z09DWvM|Q9}*Hb&JEqPll?1iW&s`b6JwXJ(B`zkZLdfWD8#~W4@U%dZS;jq-mfEx)# zB~$lINw9b6tu52UJyjj14fd*-FuVM4{_W=$Z5kF_6B)X^X=z`rjaQGsUG@zb8kJtQ zqFw*(9n!P=tImCKBdt(7`(l*>b-J&1QLICR7(p8y3DO=F55C z=VjJZ&${sME$3G5*q(ee{rSc3Cap>yF)44`<+)2{s@KD{Hci(R<_tJ~Dsk4a_omEV zb76Jrj>yF9X!TpT)=3OeCX3d(Rai;si*w=q1ss~*4ev_)G|H9aQm?d+PK;x%r2*DHBuy~qtmsS zO{&j+o4a;rw%1X6#~3ZBR_&s&ic3RRt0^Jj5#drB&KC4>7}@uu&7W5dOqs?G>oaC0f3xm~TJ2M_cA0p0@|qXdP;c z@9blbYbRvw%CwrGl4Hu_6b{_4D&&;b#N+Js_>qz28_E*Z=u=vA?Uhvf>{Hs(Iv9HA zMQx+ISYTi9ver-_I4dvU?hUnncR|}D#2S0?Ma?Wloc-ME+R-QQXk7G;HdzxtsY~x^ zRicARy`cr*$+YW+7Ah+2*KTN!>WO*k+9z6jQEvb76YY^louz}1Y>ZmnLH+hM($!`U zeW10eC0@3V{a%||tG=nSUBvzsuceg~-`K5A{iL<^%slNCzsqM;T1Mr7wO`pk`AN&y zM21TG6}QHI`}4nQgX*}in||=15}FKaHf(7?&C|&Kg z$g;|sq(A# zscBnt$kx<$ygJ(Uv5F0oOQ41$+xqHGHOH#GZQoZvth!q>zG|Sl944C&d80bAszG(vYOC#7_1vmL zbpOKkuj=hJt85X~t84shO{y$afz@3w{8-fq+uo`}wpVRmSAALiT-At!1FAD^OAb!2 zN>qV;W!LT*w#hZ!YMwt-T{XOBsqLkzY}<#nL|gEoXAgdHFsZs{)w*iW>Ic<_s&-Yk zs-9uXuS&J;wmnpv`^vWB3w5lo>?tbMzx&E8@s&#ICr{&0`}UX9aQ$TUm%GFR)pCHm zD7veA1LRHjnACS}s29TJ!Nyxnr-QPBA_LRSFPLrz9t#XMjZ^6s%;}?)MGk7()?^7> zWO^nr(!APq-rUuc8u+?tqx!}ohji3}R+~bD9t4D&J_@Wbrv)}NCzv(!M*)*fYfQfd zzF-b8w=y*eyr8BHl)={jn9rJS1QwdSf_enLX|82TF*OakYzhw?W%@aAsp%clSo3{z zQjlj*f@y)NZs0xBYi2KVp!pNi$3YEDUCdieCFUCQhCpGuYFZXF&~(B)K4_Zxq3L&1 zXrRgTTwo`&oB57ujQMHP;-Gb=X+cpY8KhxU_n?mgttN{p+g@joe1oU^`e0d0h@I-@ z5E&?T*u_wJKy&vki0Q3rhRcKP-`E#V7bj~ ze){mE&f$Kh;FQi|M@jc!H^={c-TeJ4JygO-`BvNr&CRW+3!@>AbioXn1cJ7YF$VPs zF7$`o%Si{6Ix2LWY-Rmw}CtXkmIT?tOj02;zo^Ed85F)_kKPDYRL$o@kR8>g2w8W?9MlcsA(JE!2a>^bCzDXl zc#Qd|lSCCdN;dXPg-8|>vZw+4?lZ#URE>Dx4+>k1-x~5(U!11Tf!) zxsaDS$v|na5b}!0n1Q+z_2dNUWle(OviOUcsFOG_-Gvy)b6i*gd7cXkAd@U7u_z6a zAy51V;@#Yux%I%}TM&W2p)4`dTaAs8b!%-v+Xi)y^>1UO`{-S$mp?`k>SVVIHIPZR z3o9Z24V4pvG}sg)>#L)f+?t0Dq}WLn%FRx4P?mwMAj`=bl+`X2KvqslP(IE!=<>l@ zC;LzuST)>B_~UQvME#Hp&pvH?Tn_rC`%ztVW&eaIzky!C}ax z&`B=J8`!o#QE^jbJGF3%bf5A*^zS*jjq>hee2ls)w@^2@1KHpvwTyjL_i(|HW{czH2d9X z8Q}FUbOy^IFIQe;(dIK2vHk#cgIvh2oJV~lSOoH1xC%MLi47&0?BrRLXTkGo+%%l3 z8PlYDzn4&ta}tB{m{-)XX~3W0si6QLtp!Ax)noFK{zF>GB5(O?YZ5l+^iT%&eRgLy|WOs3C=Dth4+kTiyo~b7V`QUlkPev~CMqXdhN(@_xKKWuCL4NGLs>Et zG8U!59>`fv(omAoPUfQA;KYv7l_=D6T(|&vx|1U)Y~X?yg3aZpgGoXcoB7j zS0KNl;zr9baYmgVEn~%TwR?;J)9(7SR8b>(CiO0t0{-$*=f)bA!yK;2*mWOCGnmmwQG580p$@)Z}dA(Ooz z&q)c&+`s7P_Btx(z%dsNKsGoJ`MjHIFj_X7vP^SxBbne2=nYy6{IdB6x!@Qh@mhCormWkjGjzn~PgI*Y*+zxL$+{j-K&hk7$KlQ9tvDpwXL0hWX06ab z0zqFVVM5K1mo3!FcNfKYlpJA3qM0g`Uq9dZ99s!2Zo@- zlaJ95^~Y%f-9rpExDVNZ+Au@532cD&4lewH_NP#PobObq9+?@^N3BE{m4AY4=;eu? zB-B6&#+Qp1N4C z`mQM7Mwtu}A@dOv2B_y=u+D3hCFf!$2e=r&18~4lEMh@<1-e0y`E8~H zU}Td3ov#Bu$FYxLEEl9fG%NtHehGE@?+=+lqHYGtTf*w-GzSV!NDr7@T!hguERI1N zEguaS8Oo#uNJYOM(5*)K9H8D9_Znm_Mi>=3I07Y8gX<;pa$_U&ieVkAsTqr9heS@y z4xEhTE%f70a`x>5Sk?!;|2U2jG}PH12bnJIh0G&!5i-AH(G0e;ALCDV*P!G+P?v*}W^yj}Wu~(a zpIB&@Zs*)oNRl*Jxw9joq#+W#fd!faZZ1=g*NYqU^HFe9*v>_0*m$Ro*}`~o9G##+ zp8$TZF`=X}10eH6%>ogCM;ifA>*wJM(=ZI;1WZ+K6wh!mWNv5%U}l6v$BpDGCKseJ zY-jXbGmZK{Fc0G&q5K!1K^%J%gh1x2FpZdiI_{5V`m<&el>8=Agob@6>nr;*{EbV# znzUTD9m7qaKN)DPK*=5xAoFw09$@H3L1tfm`B=x9KoNBL;AzNQWsRnOSSI^h6WIoM z4~_>DA5p#MS`v7`K zTX9Es)-P#hB|CA$XvhaB52NHW#c+xr(~^aHYK81Hi;lC-$>dU}# zEIc0NM=0qCGj^)8PYUXhAPa3Hp}URpE_fF@v-)6(>^Pp&&}YVwHxZleq00)Cl`cbA zXFF4Y{d)tBIRF@oTNuN(iEirWCDPA>xlNOPR$)uT>}vK3;4oDl#2k zmpLuZ>lu{6pc5Dht^hvcdAxY^jFZ9kNr0E~Xq4=C3Gi{d!&ztUv;6>g6FMu}*v>R# zTsWRF!jnjI8=)?z$(Gjos1HYf?iF2Nzct{$fI6BShjIY=Fm_BAo{1R15HYnBbexy( zDLfCY(YGh`lL15EgZgZg4B0`*MUZcRPW)>f-nu`d!rQt7xuH#w1*H3Scy7&`ES;W8IrsFe%626#xB z*))!qCY_>z?NM$A?88X&yaaoM#B&931;QO0L7BZp_2fBu0=K4Eos`SJDFM7sYvOfb;$bCFf!wc@*A8 z$y3eEzlu87;MKAX@{=gphjHSYE;pzOI&R)gwO~HJo6$Uni)!Hl*({NX$2yaWPm9Zd zZqi*orMV?EoBPRW>B@1y&>C;{bY&CjoSQcwbMY?v^5#5%y3|z20{qI{vp_cK&ra;X zMJao_)H#^^WhCCx(at?&jE)PHxlp#KJsPr?rW($dEv)OY2>0(S;0ZYk3P2j_{P3{> zeHkww)cJ9XKRM_q-_2+s9jidu5+!YV4l)nN7Svf@0qnyz%GB}W6t6B7F;&of{a<_d6* zqn;K=LiYqpuEPW#sj0F|IllG)^jn~*SF)s^A0t59eub_l`f)7#^AcPM`GBS(u+}X4 zF%Yywzt6#3$iXPt|0>`kmml_gL03TAHBb)(0H(-b!2U+Rk1)PA>O55J-%CYiNbkVY z5Lqn;?EV|NeGT{#i2va%7V_nvK8d_;Q8x z_v0Dz2gc*~CrrS#=phf_Y)vi8lzt`#oljFvz%_jU`#kntX~YK|)eNh=ny{q#7z+qr z9+)3~W4sC!syD;4vs(V5h0SGel5TDG!M!;vh#7dEn>1*07^(Sk2-{h+V#cJ!fe zuFRP@EBjMdnUbTgUbW{SEGh_(9<$U!J>fi)TD7)N#}#3v-WG5_KK#msW0oc6f0^lwQ_ZGZFB#0on!Ryu-w|h zV%{8Np+hu|hF!egJ4^3CZ)iO_>oKJf7q4f9(^7iI0jmaw>e1jpOrk=IWNWJt9&GhZXCT}e|k@2 zxcMlps4zr_^hKBj+#4EjW4yP89>FtMn;KpoWYTq<2yxf@Q@-)TL4Ciu*ECAsCpg1C zoR{;_0nW*i3ty`BKsVTXVe%GT;0Bl8?W21~k7zXqJEFmTbcE0Fv~F6mz8ELO-s^k9 zh#3wv3VUB~rvsb|DXl&#lI>@+60z;3z9>tE-ElVlvXGwB3-+aV+*2Cj>NUqPbJ>r2 z!f>-rXH@x0>C<>zP>sGBj1#pOrCwYqhji7G9#e1{290%2ft1mg>er)zGtYuy3U@nR zh33fKb$|32tlO)W=Exu`Q zH;|dC?~RclJg3ZF%kS~%m#@#ozMnQr)Yo#F>Q4LBp&S`#rIFku&QSJg?{HeiRp^{? zf6!TDWmw^~*s;+}ZETXB_UQL#tZ4-;uDY9^pInI^ms*iti=Z>r&7-XY@Qwt zrf$MRcgyKkhb+t*hf(pmfi#wr^Gq>QIG(PtPPdIvHGj!}V4ah<7$QAP)Dzwqeg~)N z#-MJ6vbBX)avvs6@2`8qebEo)j&7De9{#SUa@nV+b?dpC9K*EbUO5c-MfZxi!!+Re zp^?0^f*S{R$ku{>RjB{zh#Us>ickXHlEVVy3po#S|=a6=genuj)Zu~>TfuoD{@I7hxZ z{NV&VvZ}OL1|{hS85+kFO81@_%8+w9n1_fp&?C*6s2_c8p}uv-!J-Si9Jt+vIdqk| z!!5&YrKW6S~zFBm%R?&xdZ>N`ze7@;N2uj|{(r57};%j1U)NlJZ``MylqUq_34Hj4LpW=qd+rC5OBIdfG7PY@_R(h*`j$C+u%n zXWR&m%xIu{M&r1N+&s2(9vbHKc(8tL8k5j6XQRFsSWv$K8CJ(t60E;i83yv2MSrzu zvm9a#II~0V#1JsQcu#R+uGn*Br-7W9o>Ip?gfb5WJz&1mGwAf|g$dTrS2*?n;bnx_ z-^g#y%gAsnuI!2WZNh^@SJ{znV-Y>>oXoLUsvbs$#h8HJu!6zH?~<*QO1wGQHr8?9BY*nv4xI zLC4cTF4mfprb+;WG7r#V*NUPj7T(QwU(D#vhj?kaN<^XrEJUa4(srDqt8pgRl{ z;RHrf^ZkrAAAE#0u)gTsd9h45(eOy)oBh z7tYV9U_P}wS2njI{`$+JV~_EWKo&4Mu31b284WthOyfQoTR>wO3~mjK)=wmzp`PP7 zFEf)RGm<*PY0Rfe^5r-y_lye|KTmitXaGAKR$~HMufJ>>{pgLWXU5+D70F$CL>YO* z0UE+~M}&EG1doOF1-g=xabaTu8o=|zv%-_ApD{#;gZV%q%vN4vYdeS)*8gFKe@+_5s^>}AFBdboA%6>CB2aM9g%@N!# zdPJEW;fU@Gb<9JybE~)=d@thc)bVQ9^A@Es-f>KKsitN4H*nkndd~&vAr0p}rI(nP zGTveI7YFVuUq3k!VbVJmWgQD6AO3ilsRw|vzAz2LtwbYv_6<)R&#l&~y^@v)MvxZ? zqe#!zKl#)^8uiT!U+ZD_$^Jw?zI^F0B0*zlG!vD^5*nc{DSU0=M)Q4#G2{c8rqETx zWSYXwUAauXJq`0etgZNRmR0k*@!zWG5TS?Kx`E6LdGb_c^(|^oV-u@ zT;RgoB=u{V^l9gq28RPbqiL$XeOOx0J$>t2tCCXb+nC{|UBFs8r~U9 zHpZIk>JA{Zs-Q%+3S(;0ZDUSe4-5y_GIoy^GjxP9cN?;rRU+HBb3Azb^*1XpV}dNxaGirK|;@qgIqpvPI8xWnp@RQQf zQ3LW1m91TA{r5M@AZKNvG(J^4>My%AwRmsQ&XnPk564a&l_67qL z>>JNI9az16Tu|EFD>+rmhQ(|NkBDo#GCLqYb#U0lK1&9L`O87yD@zq$>zVJBON#q@ zIYCbUShKKC>b`xKHqZO$%&8wfeCouC+~LXQv)%S@j>)Rp6P>qq!XfLm zXNpDCt+8dm;1;SFO|m1|777yrG2wq2P)65TD5rD^_^9@g~1zpF3&1m zdn=-|S$^Lc>Bo= zdEs(%l&2ii>#)oes!=&6EkbSSl6t9?``+wF@`g~8l-1UL&gz%1Aq$(=R!1mHtutz? zKPbw(*3i1@9z|&^|65Nry2fPAmm3Sthx$uHebwOOGG}|wgtQNa$E13!-IkL+a`KLS zdFkE96oh#tj^DV>DPs1lDocF$mIAk;q$KO|`s#XBnJ()#Qu`^}>nqCE9Rc) zK5oT=&v%!1{l~?TTlU3BxmI3mqBd3LTJJSc_bZC9X1HVBPBN|h;6FAEUAC^*!s#2A z74FZ^>JgNAGHU#pbvHNU4Jul)@}1;}8(H#hGj+FeL#8!XyLULd)3|oQrQzkR<{zEE ze}Cce+(t!@GKTr?3~#vez4NKawo-2?^(J_&sCUhEYMZqw zPBB$kvQ8T{R!NsBZB(f|vAVWZU#p?)kqtVi2AuhT4r;r)*OH>|r>}^K$}{dgnk{E{ zP&@RPW-T)<>7BkIb9M5Ps>HnD)g6bP-Vn5TL;Cf_Uu-#K*|#|~DQ#`?{OI!0MG-T{ z#+U3$JR>i5P~((b>D5tv*0G?h_mVqZcPFgLKh|%})s9t9ia&fbXU49$A!#!bBj()N z`|g%)Gl$Cloz(U?`qMk91xlR!sS}(Ck*zzcNiM&hJAUbnq!^jsS#7R-B+EOi!<5?R_N1B`Gy~_e{|xH*wsmgyruhOxh+rz18mR^2g5oy;v4*d9856^zf_` z%M3$W?9VJrP5iiJ{-hZHy9W<0mZ{$AxE6B<+$*T~G;05?3+t|(EB?B2_Tht<^7m(z zrUl6_ywzyMM-DQmt(1e-2!q;DRZ8SaAN3I8VenNuDG4&jR}EI)lLvg&1uk)!8&0L3 zT4Zh0Mct|>yQHbK6@?q@{EOMdDVwcilAuDjY^4O%6SbywRWwjXp?J2c$+deok}-g{C$ z2+7(OGVk)W9kNCbwZGSz>rK3-L|J-$5m?4srM!- zGpln!`VMPC54D!!lCyHs)~JXmxwI#?x0l@2Q*Ecrmlu1g9qX5b&YrelVM*5w*9zj+ zr&h?Hda7S{iM&}dtt$P}f$7;x(oQb9zAJM@T4~{#NMjE z5-4+f1ircC=T z>4&V0it?G4zRka|VtaAIv9vSuKaO9OFmiI%w(D~i#L9OA)izDmtXnbY_{q4Lm7j#o z%DI%YaQpbU%&GGCK(&i9R2ocb>sI{=i{6|a9Xl!|p?AfM4(T^G6?BYTcO~MT?(tP6*$=Y_Jy+H;he;i`UQOX^^*iuzNS=Iljf;U6y=OHWWKsaQHEQ;SfJh$${g#-G<71M47HZ3EtCXn z$E9kxsuWw7XQ(4xm2B&^Rq8fZCCVDRQB`W-zKbhR-%)a`#zHl_u99RuwohH`s+_iZ zSE!v{Q=VHV99Q2EuN_FO@F_edT~Fe`{VKbiRNY+Px;LM3ReekBb(OF5h&V5^Lpx_N``6b-F-K;MzsrwWJrQot^6v}k#g?H6swUi8N%17!H zRry6;`$R2MI?9E2R4*k{?!2S=DZ8z=?xO-d{|&cBiS^8H>JS&_+{n^0si-2L>F`Bf6_XZMY>wX@*k)_`n0?FYhRn{* zu?DMRdmX&9{HLB+SxXsVO=&LNg%V?3+e-W*lr7ek9>P>pG0AtlM7A>3I?`KgP?X2k zy9Tk^6+TAw6owkgW|`Ypged#vgT7*~^W5PT17&tUamMps-n(pVdEN5w%MX|B zEN@jl-I86l(6ZO^qs;Fw+A2?EWq;9IIUv97FP16a$@l@{G!C`rKrt29&$5AHr}9*` z93(C){_@cvao4$Mr&CLLE>M&`j|JI#2`aNoPq+t)BbrkSAAQ%aK<(!|4H z@=h`A_8xD{F}!W;VOZ$>t|4DOGl^lH6`xfGKcDZt0u5hy?>44-H!#K;72_9PQw*yO zzk8oEdKp_8+`QkGQwIxQ^E2al!yWH+hK4@=-tQZ080HwRSbkJM!DPs^x(pHTDatwP?V+NE zqHLFUhY4?-Vr95Eq&jatQ`=8gj1Z+AuN|0hXscK6F6(^Vx~}oduJ~v3X_fnY-}G%- zak{L%U!P--EFXD{l1oO4wysb2zUTMfBa17#$-N^*OT}BhIa0KC^=PuyJEU~D;XC=& zNYP_ljU(%h9P}~!Jhf!*aXZ#$M?sg{#=Aa;$~qhwb+p@&t6R(V&pMdrb!UIWgKkH& zN_Sa|#z{x2x(;e*@J;MGVYG1eb+Y}>)2Ywn?zd&^DDi=Lit6O#$PCm;G>8Y0cIKfZ zabP}}`!`HRTPm1djhU$HEP$+&q&msk<3&T+bG!(Yf#YG}LUfaGFsB;RP$!c>1d)~F zMOS%Yyr|qN>oX{G))uka>6v>tXz|okz%U65GkBzZbl!u=|CoA zQUGo^umv(X=D-of$%$M?n@$#FvQI}=Mj#4q(?m1r0Ry6sqa(-8Zj_FkN1beQ;1uMQ z4%~+Pj-5P|WCJ*(BUellja?6+sv#dm!1w17!ns!|+QS!Cxel%%g}qd!WWo{&2^&11Mn1<7yx+>?siLV|J{9BnM`jLEsbmZPJW@eah?)y)uBaae?*o1uNuFguH%ex z)p11|`3d8kQ0laVto#i%(MFosd8RnYtZ>m#X(%s-;WU2$Kb4VOwy(gFxD9P|QhULpTCM${@AD{7isqD`kS?d9@uxB)w&-VYe;1fV4C?KDP7djAc6XzOWb z07_EN&LEU+?KDEEGXgRhYr|g?JW+82CNR>@P?Vq8`4r`6j>x?sJe47QUg&@P?qkH4 z)p&rq&OOLFA4C3FR)&a{b?-oYg^Z@+dC=MLG1?zGB411to#nz1(JZnAIeHo-*%^Uy zik*=t$rf-HOmtuxILL9*ghzjkRS-jE{uI&J?=Zv+FdD1@3qb{#XJ-z| zNIN@F?vP3-yl4=L{hf?DxddYE+(&t}8mmz!(ROyB+zrM6Yc<|Sy*l^MMlOOIf6-m5 zyoZKO4!jL{518-30mvFMXfodRrcV~mb>Bc6i6An6l4xnpf~d0-vIKA1$w4^|Y;_ciXCUW;2@d2!);R&0Bvzvmb)AinH{f#C z$%9M^R9vn2716x6sopO{qD;(crS3AGp zVkGTRuUU=n(C#KvX5c*|I~u{+6ph=|6NCPVj#tRBu{bu)u^7-G7WTdR)3FbRHg}*I zp0R?#Qng-UDsbX@xe@ z$btTl$#0lLp3A*7?HEj}_NPV_x|_<%>3I0wn~qezQEAx7%;UPs0Rkfi}dsZ7;847PMUxxn5R=8bw}!;?(2XFvJ* zO$5~9GKqB)D@M9sD|0U>`u`*?{|e9_Vheovv^ZD$N3%$2!Na2OGgD z;5g?b>PG14g2g|!8%A9SmyApk&HZ!eLZ zPWTDNWZ~;E&jVBW zLzmx5j-$4^hVhQw86L_)S1Gt$Egn5CV-z&j{=m31CHe;@B`2YumyemaDz6p zQ$u@40Ln3705Admmf_b3ca)UZ+ZD2p3a_D!KR{^&l^;O=a?lnsUkCRBzJzdNm}%T( zI?1^qC~0(0)oinp*0)1PDN1_BfxH5Sf)|iS07isS_?BZ)=VOicH0S<=(hq%Hob@&+ zsgIGRafBBJ4W^zBAQYU%`kG}7>$`yINM*d6xBw$Si`xUHP9~rd{u+D_@TV)wWDp1V z$tfOV>Y(IZN4>m4`1Zy-au7=Doe6lObdVQgqJ6tZ z_U129LZ8z^4F4P)LY+tEOGUoBK(q^IAZa&!e}I14#28WK2FM)yBT7CrcoZ26o)u;b zV^W6xV#u7&_M3on{!oe*i5BLUsB$Vxs;moW4u^2c5kR%iNv4ddS-J;_UrSg z?_0pJwb4!&IoBO<9iFyLC=nT1BULn1KU8F&g}9L)FBBeXLsiaNDB2C=3~DXIO4LLx zj#f|$6XY7&eIVaQNz2BfgpNdbC%dN#H?_z~?m^(qyrg~weD|Te4sa2oPu31Q>WnDc zxG)!G2DG3dkUOB{Ld>4opdVm%_<>H~1m@fWT!W#Zj(;n-WvGPL5wEv#C@D_|9DE9} zp9hzdc(_fFIhOAk`!S9Nu%F|&h~rhP9~t)X3C}d9Q9NH<3&&<*yl73&hnTwiDlRs~zsRn0May3LoR87RIsB-_u^}kA7UytR z*#F3`yF|`S#_z2kl0`Ff9>($!!Blhud>7>1G#Su#_S07ely>m$;-$h3;ZC#8!0@5U z8#@c|K4Y1WHtJ!}dCnLoW(%{4>)>aCS-WknqTx2+A~cq%L?ifU`310Vv|TAviz&4p zZ4XiMpwn>%f`ROf`V_#gAEyD+j{VfjbZ!86ozQ?Oe3*{FBzndhg4X?kI-gtgkVY`K zsZ64eWB3}t$vlhPab|#?4fHe`?Od=a+D`($67!Wr02;tP-ay=$#2WZh=p=8Xh<0_^ zz)O+&9xCgn;0qTcM%~m)2WZDEKs((~(zzqxC`hr(rRd{j#Vq2Q>p?JN26Z$_9+N_p zgHSR+2hmUa=m(EU7qs&Z!CH;APlJom$Y&&@j@!|Bp6EX6zi8tD=NkuaJf`|`lqG2M z1-!-h&qBUN`3*>fEHH-tg@H}L0=g*j*nBa($##tKMn8?<=Qiq|Eo0^hqnRb;vx*yu&!|K1%*7;r=iSIgh6CBozSe7h}r()f1u(WM(q!+&j8j0eK}r ztjt=9GZ`3sDk%rt$dP~(^`p!-UY7KcGB@7?${1YwG!4)YCe>=d7&3m;OQZQO3-I#| zkDUQ!Fu18e19xx%h(pJEFdA|N5RgZyazdhLI-G_u0`#1V(`#+N@tjfbRr*ZT${T?2lO@g z&58lwZj1-ZU=!yvs3CyaNqaY=#9b1v8O0-c4$xU10WQu-Tu4{Ub~?cY*mnYbtm*?BduP_!T;zlN(a@;7c z%WP^17$B~t9ci==LA|kk9`#Xf1bE%F1w7i^gr6~v{ruG82~q*;9LJYW`a-=cAh(3f zJ|vU;b*b=;PrFdFmKd*qogB4f#V{SvjG>O>;X8R*BxP?(q#Qi(Oz+p zUDL%N+0-mNWO9a>(1qiCz<)93CHN6AyK4bD_%S#OW@2qGC)p}Pv}wvYy8ufS)COLP zthr3I@ujXhfI8Suy{yv*CI;IWh*_xDkhjwD`RhuC@EFX^pq+eRa4t_5eW9=PnK5xe z$@z5w^|7CGo`UC2^3+n{-Uo>;rP!H#`ZyKO^ z|M#1h%eOsz<#e;SX})}WnaS(PjKE?!>$ZN? zH;`I5hBNnOi@tSe8kZ*WaklWOO@o$rJDZ?F=C2aoZa=vW)#k6X%rS8|b;-X|w3tdDYNLZ2+yK5|>bkrFwcnl;Km1UbnWc)D0YHh&ecl=EHL~7Mn)0ME! zC{iK%O{i-!>zYDsu}aM<|I@=TOl}ox#>O-oVaXu6+D`~)YU;bkHtn#Ai z=hvnTW}R7ke0FW5V}<4}(o)N07_1q{UAGzET3eh3(uj7YQ%yW`l-aMxjU`Wko_CD` z%;9!#8fk-hD77T^&(q?{!_5TeLWD-Kue$tVK;T z1RKc%Lz#w>2($h4JBE}p!$yT1tS?NfDKjH7?!I!p^vxeaDjt1 zBlR2R#-&P=nc?BUR&}f_anE9 zkz*hCpL%IHjk6sKcWv*P2V8GJ&o;IM;Xo>H{#J*LpiwVco&BuqpeUE0bqT$XvnIYe(~9&ygppin$GHi>4Pd9(bo z_NjJ&;iWd3o5oR`#i$^irLqZMpR|T$1&U- zwwLWQmudr+bO$K&nzOkh8sA=JSr zA_AIG%)vWmqz)TtF&E{=u+BFghKh=rxQzlneYMT(XPc$~^JxSPqd_!|>Bv4tNi!HF zjkjIZ9kheXb~@3S6X>NLLMYd~(GB51?3GN&5*_gZ0W7%>RK!GYxKcDR z+m0?@^B!yNaUe}#<}grcq6MJ(KJ9+C-30GxMS!c=B z>5=}Fp;wqAzh5Id%H*~9{>J5K3uiI`nRJA{vYq|(jhYw(`uf*)hokH;2T$`E8w81n??}u{41gU@^(nz|_tH2gNo6#7r z1)qje51rsuzzw#&p+C{|aehgn+2$2J<8IM9F339L#zC|e%BWO-QP$dN6f=cmnJJjB zJqURGwf7Jh$q3L8o+e%uG=Ro3N_6)AYqvJ=DI!O%7oH8+!WF5JR>-_{*daPe{kEqA z--~S525sf3rGq?#T#GxX%lZSJ>oQ;r2H_aqZCruDWE<~429hh&HtM2(UT3$vX#;HA zT1$&O(m}4s6@H^_$C$6xo3%^Mo|>8!T*Y?7x@%iT1=Pq@Lm~%iTf{bkang5#7}kuh z034*7!A|V~_BdB~nYlsQZfiShyVB@9&7iZ~FwWF&5VY&zWS}^2$@ZR_lhmVcHpg?` z08d5RT;5=;(+Zw8>ZQpDm|4qpb~5LwfIf0ZS@K%r-f?%?#xqYpDbqkI(XUEImomq5 zA=YUKca}!mqHjxGy3e`p^URyIt!5`1*1Lac-%jm`q%Fu)<-N*Y+Y#IJC%Yb7uEj)PtQdFd z=Eid@*ZcEjhSv7iE03*@Zv@PH*!e2kIFG@jJ=`<>%Ajpn0~6sCJDBQh)PJZkjTsc4 zsFB-5M?+wpp;7rg?E&`X?%6#U@GKQ!>f@iG9ef{E#5F&UxU0zP1n87aD_D z%MDxc9*qr>hYQ63xp^Bt-+fnve{oGM60I833)_NxuwYT+61lBNw2tA+qAftBw#(C= zMO=Zq$CTsaoh#EaTF48SGTq~@b9Y&%Ck#hOUgNWGv_}^Fekm6n$Lm87*HFh~{$C zR=gTBmOR@W!*kDEVcrsY%JWZ|ukj2CHRz84ddW8Csvdtm0kjhPLL0O_f_^`P^F9o)nKwjG>Uf2FVZe){U(be|f2#+Wh-E98_Y(}w& owf)DR$BOZv#r#$*S~zR}on^D_!rz(y-kS`jq Date: Thu, 13 Aug 2026 14:45:39 +0000 Subject: [PATCH 13/17] docs: refresh generated docs (REFERENCE, CAPABILITIES, API_QUICKREF, PIPELINE_MAP, FACULTY_MAP, DOC_MAP) [skip ci] --- REFERENCE.md | 352 +++++++++++++++++++++++++-------------------------- 1 file changed, 176 insertions(+), 176 deletions(-) diff --git a/REFERENCE.md b/REFERENCE.md index 6669956..ab86e1c 100644 --- a/REFERENCE.md +++ b/REFERENCE.md @@ -76,8 +76,8 @@ | module | what it is | lines | |---|---|---| | [`holographic_session.py`](#holographic-session) | holographic_session.py -- ONE render session that ties the disconnected rendering threads together. | 227 | -| [`holographic_session.py`](#holographic-session) | SESSION -- never compute the same conversation prefix twice. | 253 | | [`holographic_session.py`](#holographic-session) | SESSION STORE -- contexts that outlive the process. | 360 | +| [`holographic_session.py`](#holographic-session) | SESSION -- never compute the same conversation prefix twice. | 253 | ### `splat*` family (6) @@ -423,8 +423,8 @@ | [`holographic_materialio.py`](#holographic-materialio) | PBR material interchange (G2-IO): the standard-format bridge for materials, and a VSA-native carrier. | 307 | | [`holographic_matlib.py`](#holographic-matlib) | holographic_matlib.py -- a comprehensive RENDER material library: plain diffuse -> a fractal planet. | 801 | | [`holographic_meaning_predict.py`](#holographic-meaning-predict) | Generation with structure: predict a next-MEANING vector and settle it, rather | 198 | -| [`holographic_measure.py`](#holographic-measure) | The variance harness: every headline number gets a mean, a spread, and a confidence | 187 | | [`holographic_measure.py`](#holographic-measure) | MEASURE -- perplexity with error bars, and decisions that respect them. | 239 | +| [`holographic_measure.py`](#holographic-measure) | The variance harness: every headline number gets a mean, a spread, and a confidence | 187 | | [`holographic_memory.py`](#holographic-memory) | MEMORY -- the Galvatron's own store, built on leCore's holographic database. | 241 | | [`holographic_memoryhome.py`](#holographic-memoryhome) | holographic_memoryhome.py -- the MEMORY home (consolidation backlog H6): keep the hot working set where the CP | 165 | | [`holographic_memsearch.py`](#holographic-memsearch) | MEMSEARCH -- searchable memory that lives in the weights and runs in the pass. | 212 | @@ -543,8 +543,8 @@ | [`holographic_realtime.py`](#holographic-realtime) | holographic_realtime.py -- the realtime render loop, and the multi-format payload it pushes. | 404 | | [`holographic_reanchor.py`](#holographic-reanchor) | Re-anchoring is load-bearing for deep traversal -- the audit, and the contrast the other tests don't show. | 97 | | [`holographic_reasoning.py`](#holographic-reasoning) | holographic_reasoning.py | 397 | -| [`holographic_recipe.py`](#holographic-recipe) | A generative recipe-store for constructed holostuff structures. | 206 | | [`holographic_recipe.py`](#holographic-recipe) | RECIPE -- ship what leCore ADDED, not the model it was added to. | 334 | +| [`holographic_recipe.py`](#holographic-recipe) | A generative recipe-store for constructed holostuff structures. | 206 | | [`holographic_recipeops.py`](#holographic-recipeops) | StructureRecipe validator + edit operators (ARCH-1): the recipe equivalent of the mesh Euler operators. | 264 | | [`holographic_reclock.py`](#holographic-reclock) | holographic_reclock.py -- sample when an AXIS moves, not when time passes ("make the boring property the | 313 | | [`holographic_recurrent.py`](#holographic-recurrent) | A gradient-free RECURRENT layer for the holographic engine: reservoir computing. | 456 | @@ -578,8 +578,8 @@ | [`holographic_rolling.py`](#holographic-rolling) | holographic_rolling.py -- the CAUSAL rolling / streaming statistics kit: trailing mean, std, min, max, | 378 | | [`holographic_roomacoustic.py`](#holographic-roomacoustic) | holographic_roomacoustic.py -- A6: GEOMETRIC ROOM ACOUSTICS. How a room echoes -- reflections and reverb. | 161 | | [`holographic_route.py`](#holographic-route) | Representation routing (ARCH-7): route each operation to the representation that supports it -- the policy lay | 183 | -| [`holographic_router.py`](#holographic-router) | ROUTER -- the model DECIDING, inside one forward pass. | 229 | | [`holographic_router.py`](#holographic-router) | Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. | 189 | +| [`holographic_router.py`](#holographic-router) | ROUTER -- the model DECIDING, inside one forward pass. | 229 | | [`holographic_sampler.py`](#holographic-sampler) | holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). | 223 | | [`holographic_sampling.py`](#holographic-sampling) | Blue-noise / Poisson-disk point sampling -- the EXCLUSION principle, done right. | 134 | | [`holographic_samplinghome.py`](#holographic-samplinghome) | holographic_samplinghome.py -- the SAMPLING home (consolidation backlog R4): one place for the Monte-Carlo | 127 | @@ -684,8 +684,8 @@ | [`holographic_toolclient.py`](#holographic-toolclient) | holographic_toolclient.py -- call another node the same way leCore is called. | 98 | | [`holographic_topology.py`](#holographic-topology) | Principled topology by persistent homology (EXP-7). | 288 | | [`holographic_transfer.py`](#holographic-transfer) | holographic_transfer.py -- KERNEL SCATTER / GATHER: the ONE bundle/readout under every particle<->grid transfe | 224 | -| [`holographic_transform.py`](#holographic-transform) | holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). | 297 | | [`holographic_transform.py`](#holographic-transform) | TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. | 189 | +| [`holographic_transform.py`](#holographic-transform) | holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). | 297 | | [`holographic_transform_space.py`](#holographic-transform-space) | holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it | 176 | | [`holographic_transformbank.py`](#holographic-transformbank) | holographic_transformbank.py -- a prebuilt map of hypervector transforms, and what it can and cannot hold. | 358 | | [`holographic_transformhome.py`](#holographic-transformhome) | holographic_transformhome.py -- the TRANSFORM home (consolidation backlog H5): one facade over "move / rotate | 223 | @@ -14849,39 +14849,6 @@ ### holographic_measure.py -> The variance harness: every headline number gets a mean, a spread, and a confidence -> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. -> -> WHY this exists -> --------------- -> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's -> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A -> single-seed score is therefore a sample from a distribution, and reporting it alone hides -> how wide that distribution is. For an engine whose entire pitch is "measured, not -> promised," reporting a number without its noise is the sharpest blind spot -- so this -> points that same discipline at the numbers themselves. -> -> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the -> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no -> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI -> bound clears the floor -- which is what stops a single fortunate seed from passing a -> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". -> -> USE REAL DATA. The point of the harness is to characterise the real distribution of a -> real claim; running it on a toy makes the spread meaningless. The measurements wired -> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). - -**Public API:** - -- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. -- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. -- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. -- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is -- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above -- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a - -### holographic_measure.py - > MEASURE -- perplexity with error bars, and decisions that respect them. > > Moose asked what assimilation is actually doing. The answer, from his own run: @@ -14914,6 +14881,39 @@ - `def better_than(a, b, alpha, seed, resamples)` -- Is model A better than model B, or is the difference undecidable? - `def tokens_needed(reference, effect_pct, alpha)` -- How many tokens would be needed to RESOLVE an effect of this size. +### holographic_measure.py + +> The variance harness: every headline number gets a mean, a spread, and a confidence +> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. +> +> WHY this exists +> --------------- +> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's +> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A +> single-seed score is therefore a sample from a distribution, and reporting it alone hides +> how wide that distribution is. For an engine whose entire pitch is "measured, not +> promised," reporting a number without its noise is the sharpest blind spot -- so this +> points that same discipline at the numbers themselves. +> +> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the +> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no +> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI +> bound clears the floor -- which is what stops a single fortunate seed from passing a +> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". +> +> USE REAL DATA. The point of the harness is to characterise the real distribution of a +> real claim; running it on a toy makes the spread meaningless. The measurements wired +> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). + +**Public API:** + +- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. +- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. +- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. +- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is +- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above +- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a + ### holographic_memory.py > MEMORY -- the Galvatron's own store, built on leCore's holographic database. @@ -21091,42 +21091,6 @@ ### holographic_recipe.py -> A generative recipe-store for constructed holostuff structures. -> -> WHY THIS EXISTS -> --------------- -> The panel's "proven structure has no noise" result: a structure BUILT by a deterministic proof (a -> derivation over a seed) carries no noise, so it serialises to its *generator* losslessly -- you store the -> recipe, not the expanded vectors, and replaying the recipe reproduces the structure BIT-FOR-BIT. This is -> the easy, exact half of generative compression: when we are the builder we already hold the proof, so -> there is nothing to search for and no residual to code. -> -> A `StructureRecipe` is a tiny replayable build-graph. Each op produces one result vector from a seed and -> earlier results: `atom` (a derived atom -- regenerated from the seed, never stored), `bind`, `bundle`, -> `permute`, `normalize`. You build your structure THROUGH the recipe, so you get both the vectors and the -> recipe that regenerates them. Serialising stores only the op list (and the seed) -- a few hundred bytes -> that regenerate megabytes of structure, exactly. -> -> THE ESCAPE HATCH AND THE KEPT NEGATIVE -> The `raw` op stores a literal vector verbatim. It is the honest boundary: data that was NOT constructed -> (a measured or random vector) has no short recipe, so it must be stored as-is and gets no compression. -> The recipe's compression ratio is therefore exactly the *constructed fraction* of the structure -- all -> recipe -> enormous ratio; all raw -> ~1x. That is the constructed-vs-measured partition made literal. -> -> THE CAPACITY-CLIFF POINT -> Reading structure back out of a single bounded encoded vector degrades past the capacity cliff (crosstalk). -> A recipe does not: it names its leaves explicitly and replays the construction, so a deeply nested -> structure is recovered EXACTLY at any depth. The recipe is the right store for deep constructed structure; -> the expanded superposition is bounded. -> -> Pure NumPy + holostuff kernel, deterministic, JSON serialisation (readable), no new dependencies. - -**Public API:** - -- `class StructureRecipe` -- A replayable build-graph: store the deterministic ops that built a structure, not the vectors. - -### holographic_recipe.py - > RECIPE -- ship what leCore ADDED, not the model it was added to. > > Moose: the inflated model size needs looking at holographically -- we should be @@ -21179,6 +21143,42 @@ - `def cost(rules, arrays, installed_weights)` -- What the recipe saves, in bytes. The number is the whole argument. - `def expand(rules, arrays, base_weights)` -- Rebuild the installed model from the recipe. Must be byte-exact. +### holographic_recipe.py + +> A generative recipe-store for constructed holostuff structures. +> +> WHY THIS EXISTS +> --------------- +> The panel's "proven structure has no noise" result: a structure BUILT by a deterministic proof (a +> derivation over a seed) carries no noise, so it serialises to its *generator* losslessly -- you store the +> recipe, not the expanded vectors, and replaying the recipe reproduces the structure BIT-FOR-BIT. This is +> the easy, exact half of generative compression: when we are the builder we already hold the proof, so +> there is nothing to search for and no residual to code. +> +> A `StructureRecipe` is a tiny replayable build-graph. Each op produces one result vector from a seed and +> earlier results: `atom` (a derived atom -- regenerated from the seed, never stored), `bind`, `bundle`, +> `permute`, `normalize`. You build your structure THROUGH the recipe, so you get both the vectors and the +> recipe that regenerates them. Serialising stores only the op list (and the seed) -- a few hundred bytes +> that regenerate megabytes of structure, exactly. +> +> THE ESCAPE HATCH AND THE KEPT NEGATIVE +> The `raw` op stores a literal vector verbatim. It is the honest boundary: data that was NOT constructed +> (a measured or random vector) has no short recipe, so it must be stored as-is and gets no compression. +> The recipe's compression ratio is therefore exactly the *constructed fraction* of the structure -- all +> recipe -> enormous ratio; all raw -> ~1x. That is the constructed-vs-measured partition made literal. +> +> THE CAPACITY-CLIFF POINT +> Reading structure back out of a single bounded encoded vector degrades past the capacity cliff (crosstalk). +> A recipe does not: it names its leaves explicitly and replays the construction, so a deeply nested +> structure is recovered EXACTLY at any depth. The recipe is the right store for deep constructed structure; +> the expanded superposition is bounded. +> +> Pure NumPy + holostuff kernel, deterministic, JSON serialisation (readable), no new dependencies. + +**Public API:** + +- `class StructureRecipe` -- A replayable build-graph: store the deterministic ops that built a structure, not the vectors. + ### holographic_recipeops.py > StructureRecipe validator + edit operators (ARCH-1): the recipe equivalent of the mesh Euler operators. @@ -22712,6 +22712,33 @@ ### holographic_router.py +> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. +> +> WHY THIS EXISTS (measured, backlog N9/N28): +> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, +> but it has no notion of MEANING: "squish a big array down for storage" shares no token with +> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics +> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median +> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. +> +> WHAT SHIPS (and what does NOT): +> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted +> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a +> query it already has a VECTOR for: +> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) +> * a caller that supplies its own query vector (an app that ran the encoder itself) +> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this +> router says so and returns None, so the caller falls back to the token router rather than guessing. That +> honesty is the whole point: silence beats a confident wrong route. +> +> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. + +**Public API:** + +- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. + +### holographic_router.py + > ROUTER -- the model DECIDING, inside one forward pass. > > Moose raised the architecture that dissolves the wall this project kept hitting: @@ -22751,33 +22778,6 @@ - `def route(runtime, router, text, tokenize)` -- Would this model choose to use the capability on this prompt? - `def install_routed(weights, cfg, operator, router, layer, gain, temperature)` -- Install a circuit whose GATE is the router, not a constant. -### holographic_router.py - -> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. -> -> WHY THIS EXISTS (measured, backlog N9/N28): -> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, -> but it has no notion of MEANING: "squish a big array down for storage" shares no token with -> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics -> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median -> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. -> -> WHAT SHIPS (and what does NOT): -> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted -> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a -> query it already has a VECTOR for: -> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) -> * a caller that supplies its own query vector (an app that ran the encoder itself) -> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this -> router says so and returns None, so the caller falls back to the token router rather than guessing. That -> honesty is the whole point: silence beats a confident wrong route. -> -> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. - -**Public API:** - -- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. - ### holographic_sampler.py > holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). @@ -24408,39 +24408,6 @@ ### holographic_session.py -> SESSION -- never compute the same conversation prefix twice. -> -> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a -> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS -> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens -> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS -> REPEATED, and the fraction grows with every turn. -> -> leCore already had the pieces and never joined them: the runtime exposes -> prefill/step over an InferenceState with copy(), galvacache memoises pure -> functions at a measured 75% hit rate with bit-identical output, and sessions -> exist. What was missing is the RADIX TREE -- the structure that answers "what is -> the longest prefix of this prompt that I have already computed?" -> -> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three -> turns with turn 3, and a dictionary keyed on the whole prompt misses that -> completely. vLLM and SGLang call this RadixAttention; the idea is the same -> whatever the model: index by prefix, resume from the deepest match, compute only -> the tail. -> -> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no -> cache: resuming from a cached state reproduces a full recompute TO FLOAT -> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the -> difference is real rather than pedantic: resuming STEPS the tail while a fresh -> call PREFILLS it, and the two associate their sums differently. The selftest -> asserts the measured bound against the live runtime rather than assuming it. - -**Public API:** - -- `class PrefixCache` -- A radix tree over token sequences, holding inference states. - -### holographic_session.py - > SESSION STORE -- contexts that outlive the process. > > A Galvatron's context is not a transcript, it is STATE: the GDN recurrent @@ -24473,6 +24440,39 @@ - `class SessionStore` -- Named, persistent, independent contexts on disk. - `def runtime_fingerprint(runtime)` -- A cheap, deterministic id for the checkpoint behind a runtime, so a +### holographic_session.py + +> SESSION -- never compute the same conversation prefix twice. +> +> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a +> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS +> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens +> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS +> REPEATED, and the fraction grows with every turn. +> +> leCore already had the pieces and never joined them: the runtime exposes +> prefill/step over an InferenceState with copy(), galvacache memoises pure +> functions at a measured 75% hit rate with bit-identical output, and sessions +> exist. What was missing is the RADIX TREE -- the structure that answers "what is +> the longest prefix of this prompt that I have already computed?" +> +> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three +> turns with turn 3, and a dictionary keyed on the whole prompt misses that +> completely. vLLM and SGLang call this RadixAttention; the idea is the same +> whatever the model: index by prefix, resume from the deepest match, compute only +> the tail. +> +> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no +> cache: resuming from a cached state reproduces a full recompute TO FLOAT +> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the +> difference is real rather than pedantic: resuming STEPS the tail while a fresh +> call PREFILLS it, and the two associate their sums differently. The selftest +> asserts the measured bound against the live runtime rather than assuming it. + +**Public API:** + +- `class PrefixCache` -- A radix tree over token sequences, holding inference states. + ### holographic_shader.py > holographic_shader.py -- N filter passes in ONE evaluation. Two things a GPU structurally cannot do. @@ -28001,48 +28001,6 @@ ### holographic_transform.py -> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). -> -> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's -> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need -> in one place. This gathers the standard, well-known math: -> -> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, -> and what a property panel shows as the T/R/S fields; -> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); -> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector -> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); -> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. -> -> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN -> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit -> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL -> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is -> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. -> Deterministic; NumPy + stdlib only. - -**Public API:** - -- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. -- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). -- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). -- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. -- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear -- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. -- `def quat_normalize(q)` -- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". -- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. -- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. -- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. -- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for -- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). -- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock -- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an -- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. -- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks - -### holographic_transform.py - > TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. > > Everything before this applied leCore's levers uniformly: grow a memory channel @@ -28085,6 +28043,48 @@ - `def plan(weights, cfg, target_tokens, kv_rank, grow_gain)` -- Decide what to do to each layer, from the analysis rather than by rule. - `def apply_plan(weights, cfg, the_plan, progress)` -- Carry out the growth actions. KV compression is a RUNTIME setting and is +### holographic_transform.py + +> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). +> +> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's +> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need +> in one place. This gathers the standard, well-known math: +> +> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, +> and what a property panel shows as the T/R/S fields; +> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); +> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector +> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); +> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. +> +> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN +> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit +> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL +> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is +> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. +> Deterministic; NumPy + stdlib only. + +**Public API:** + +- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. +- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). +- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). +- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. +- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear +- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. +- `def quat_normalize(q)` +- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". +- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. +- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. +- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. +- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for +- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). +- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock +- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an +- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. +- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks + ### holographic_transform_space.py > holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it From 4f3573d7d520091dddbf4e7dbe1e77d8eae0ee4d Mon Sep 17 00:00:00 2001 From: atimics Date: Thu, 13 Aug 2026 14:37:17 -0700 Subject: [PATCH 14/17] results: publish Qwen v3 terminal evidence --- experiments/qwen35_acceptance/README.md | 33 +- .../qwen35_acceptance/build_publication.py | 22 +- .../qwen35_acceptance/launch-manifest.json | 44 +- .../README.md | 76 +++ .../ledger/config.json | 5 + .../ledger/events.jsonl | 11 + ...be1d17d9c1efea28fbb043c72381a951e5a6b100c1 | 1 + ...08b7d5e51ef5b4044c33c2f7fc30d4a61210c38a2f | 1 + ...cc9147bc5a4c61579a52c0dba4c3d55ee1d03ac5fe | 1 + ...d108e5bedf21a0c390f1bc248cf6c33a1a852f24d9 | 1 + ...e6379d97ada04a5968209fe8eae45fab477d249eea | 1 + ...183c114725032a320e03bcd3cb72a9e12f18569f47 | 1 + ...1b88aae7e74e0609bbded168252ae731eff6a39434 | 1 + ...0ad0c7f30338431c50d99754b938eaeb59f1967d1d | 1 + ...29b24028ecf54721d454811dd9aacf4ff46de4e3b3 | 1 + ...f9c426fb63f33cf7e67ad076a17d93381dc465dc76 | 1 + .../project/engineering-review.json | 21 + .../project/experiment-design.json | 22 + .../project/experiment.json | 247 ++++++++ .../project/forecast-empirical.json | 17 + .../project/forecast-mechanistic.json | 17 + .../project/foundation.json | 20 + .../project/funding.json | 11 + .../project/hypothesis.json | 19 + .../project/project.json | 156 ++++++ .../publication-manifest.json | 459 +++++++++++++++ .../publication-receipt.json | 67 +++ .../result/acceptance-artifacts/install.log | 30 + .../result/acceptance-artifacts/metrics.json | 70 +++ .../result/admission-preflight.json | 24 + .../result/aws-cleanup-receipt.json | 16 + .../result/aws-instance-identity.json | 17 + .../result/cloud-init-output.log | 525 ++++++++++++++++++ .../result/corpus-manifest.json | 12 + .../result/environment.txt | 72 +++ .../result/evidence-bundle.sha256 | 1 + .../result/execution-summary.json | 8 + .../result/generation.json | 1 + .../result/ilxyr-admission.json | 68 +++ .../result/ilxyr-commit.txt | 7 + .../result/ilxyr-compile.json | 3 + .../ilxyr-contribute-engineering-review.json | 3 + .../ilxyr-contribute-experiment-design.json | 3 + .../result/ilxyr-contribute-foundation.json | 3 + .../result/ilxyr-contribute-hypothesis.json | 3 + .../result/ilxyr-forecast-empirical.json | 3 + .../result/ilxyr-forecast-mechanistic.json | 3 + .../result/ilxyr-funding.json | 3 + .../result/ilxyr-init.json | 4 + .../result/ilxyr-run.json | 0 .../result/ilxyr-run.stderr | 1 + .../result/ilxyr-status.json | 92 +++ .../result/ilxyr-status.stderr | 0 .../result/ilxyr-verify.json | 5 + .../result/ilxyr-verify.stderr | 0 .../result/launch-manifest.json | 107 ++++ .../result/lecore-commit.txt | 7 + .../result/lecore-status.txt | 0 .../result/model-manifest.json | 66 +++ .../result/requirements-resolved.txt | 36 ++ .../result/runner-status.json | 7 + .../result/s3-terminal-status.json | 7 + .../result/user-data.log | 462 +++++++++++++++ 63 files changed, 2874 insertions(+), 52 deletions(-) create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/README.md create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/config.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/events.jsonl create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/0067fb1dc8736fd862f156be1d17d9c1efea28fbb043c72381a951e5a6b100c1 create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/08d81fe6c3b0a1d2f6c93d08b7d5e51ef5b4044c33c2f7fc30d4a61210c38a2f create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/09c2a3c1d5060228405fa5cc9147bc5a4c61579a52c0dba4c3d55ee1d03ac5fe create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/0e519bb9ba1feb38105fd2d108e5bedf21a0c390f1bc248cf6c33a1a852f24d9 create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/50266fe71e4f151846f15ce6379d97ada04a5968209fe8eae45fab477d249eea create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/64f1db0c4ea0afdf533b17183c114725032a320e03bcd3cb72a9e12f18569f47 create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/c5399dba347f001bf553941b88aae7e74e0609bbded168252ae731eff6a39434 create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/d1c0d6083c36dc5e277e4a0ad0c7f30338431c50d99754b938eaeb59f1967d1d create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/d3c31ef2f50da0924c719329b24028ecf54721d454811dd9aacf4ff46de4e3b3 create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/e0cf1ad5fb075f3e94ac70f9c426fb63f33cf7e67ad076a17d93381dc465dc76 create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/project/engineering-review.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/project/experiment-design.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/project/experiment.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/project/forecast-empirical.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/project/forecast-mechanistic.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/project/foundation.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/project/funding.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/project/hypothesis.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/project/project.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-receipt.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/acceptance-artifacts/install.log create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/acceptance-artifacts/metrics.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/admission-preflight.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/aws-cleanup-receipt.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/aws-instance-identity.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/cloud-init-output.log create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/corpus-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/environment.txt create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/evidence-bundle.sha256 create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/execution-summary.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/generation.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-admission.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-compile.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-funding.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-init.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-run.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-run.stderr create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-status.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-status.stderr create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-verify.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-verify.stderr create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/launch-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/lecore-commit.txt create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/lecore-status.txt create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/model-manifest.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/requirements-resolved.txt create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/runner-status.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/s3-terminal-status.json create mode 100644 experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/user-data.log diff --git a/experiments/qwen35_acceptance/README.md b/experiments/qwen35_acceptance/README.md index 118ad3a..ba1b26a 100644 --- a/experiments/qwen35_acceptance/README.md +++ b/experiments/qwen35_acceptance/README.md @@ -1,19 +1,24 @@ # Qwen3.5 acceptance experiment The committee-facing design and review questions are in -[`PROPOSAL.md`](PROPOSAL.md). Both authorized runs are preserved as execution -failures. V1 exposed an interpreter-path bug before scientific execution. V2 -fixed that problem and completed both 4,097-token evaluations, but the emitted -checkpoint failed the mandatory official Transformers reload because the two -prepended layers' tensor names and MLP shapes do not match the saved Qwen -architecture. The terminal v2 evidence is in -[`results/v2-20c3330d0b3e-execution-failure/`](results/v2-20c3330d0b3e-execution-failure/). -The permanent evidence index is -[`f35OgXu…ARZTU`](https://arweave.net/f35OgXuOoulk9EcndJBNWIgZ5tmaQ6DRzeINP7ARZTU/). - -The layer-prepending installer therefore remains experimental. No v3 run is -authorized. The model revision, corpora, thresholds, seed, and spending ceiling -for the completed v2 attempt remain frozen in +[`PROPOSAL.md`](PROPOSAL.md). All three authorized runs are preserved as +execution failures. V1 exposed an interpreter-path bug before scientific +execution. V2 completed both evaluation streams but exposed an official +Transformers checkpoint-layout incompatibility. V3 fixed that incompatibility +and completed the full 4,096-position paired evaluation, but human-readable +diagnostics before the metric JSON caused strict ilxyr ingestion to fail. + +The terminal v3 record is in +[`results/v3-20c3330d0b3e-execution-failure/`](results/v3-20c3330d0b3e-execution-failure/). +Its complete evidence bundle, excluding model weights, is permanently indexed +at [`h96JlX…ojYSw`](https://arweave.net/h96JlX-4ttZsPZdYRBuT-_JNO2s0lS18hoFjM4ojYSw/). +The v3 diagnostic artifact completed all required positions and measured a +31.0% perplexity regression, but it was not admitted and is therefore neither +`GO` nor `NO-GO`. + +The layer-prepending installer remains experimental. No retry is authorized +under the v3 identity. The model revision, corpora, thresholds, seed, and +spending ceiling for the completed v3 attempt remain frozen in [`launch-manifest.json`](launch-manifest.json). This directory turns the open Qwen integration questions into a frozen ilxyr @@ -39,7 +44,7 @@ python experiments/qwen35_acceptance/generate.py \ /absolute/path/to/installation-corpus.txt \ /absolute/path/to/evaluation-corpus.txt \ /absolute/path/to/ilxyr-project \ - --experiment-version 2 + --experiment-version 3 ``` The two corpora must have different contents. diff --git a/experiments/qwen35_acceptance/build_publication.py b/experiments/qwen35_acceptance/build_publication.py index 0a61078..859f7e4 100644 --- a/experiments/qwen35_acceptance/build_publication.py +++ b/experiments/qwen35_acceptance/build_publication.py @@ -27,7 +27,7 @@ def read_json(path: Path) -> Any: return json.load(handle) -def evidence_ref(result_dir: Path) -> str: +def evidence_ref(result_dir: Path) -> str | None: events_path = result_dir / "ledger" / "events.jsonl" with events_path.open("r", encoding="utf-8") as handle: events = [json.loads(line) for line in handle if line.strip()] @@ -36,9 +36,9 @@ def evidence_ref(result_dir: Path) -> str: for event in events if event.get("event_type") == "EvidenceRecorded" ] - if len(recorded) != 1: - raise ValueError(f"expected exactly one EvidenceRecorded event, got {len(recorded)}") - return recorded[0] + if len(recorded) > 1: + raise ValueError(f"expected at most one EvidenceRecorded event, got {len(recorded)}") + return recorded[0] if recorded else None def build_manifest(result_dir: Path) -> dict[str, Any]: @@ -46,6 +46,15 @@ def build_manifest(result_dir: Path) -> dict[str, Any]: status = read_json(result_dir / "result" / "ilxyr-status.json") verification = read_json(result_dir / "result" / "ilxyr-verify.json") latest_evidence = status["latest_evidence"] + if latest_evidence is None: + runner_status = read_json(result_dir / "result" / "runner-status.json") + if runner_status.get("stage") != "execution_failure": + raise ValueError("a run without admitted evidence must be an execution failure") + resolved_outcome = "execution_failure" + run_ref = status["latest_run"]["id"] + else: + resolved_outcome = latest_evidence["resolved_outcome"] + run_ref = latest_evidence["run_ref"] files = [] for path in sorted(item for item in result_dir.rglob("*") if item.is_file()): @@ -65,8 +74,9 @@ def build_manifest(result_dir: Path) -> dict[str, Any]: return { "schema": "lecore.qwen35-publication-manifest.v1", "experiment_id": project["experiment_id"], - "resolved_outcome": latest_evidence["resolved_outcome"], - "run_ref": latest_evidence["run_ref"], + "resolved_outcome": resolved_outcome, + "scientific_result_admitted": latest_evidence is not None, + "run_ref": run_ref, "evidence_ref": evidence_ref(result_dir), "ledger_verification": verification, "files": files, diff --git a/experiments/qwen35_acceptance/launch-manifest.json b/experiments/qwen35_acceptance/launch-manifest.json index 466aeca..6accc53 100644 --- a/experiments/qwen35_acceptance/launch-manifest.json +++ b/experiments/qwen35_acceptance/launch-manifest.json @@ -1,22 +1,22 @@ { "schema": "lecore.qwen35-launch.v1", - "state": "terminal_execution_failure", + "state": "authorized", "attempt": { - "experiment_version": 2, + "experiment_version": 3, "maximum_formal_runs": 1, "retry_after_admission": false, "seed": 0, - "authorized_at": "2026-08-13T00:09:30Z", + "authorized_at": "2026-08-13T18:49:00Z", "predecessor": { - "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.a04ab5692be3.v1.acceptance", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.d6905f043e78.a6c9d1135a04.2f32151c6d44.v2.acceptance", "outcome": "execution_failure", "scientific_result": false, - "corrective_change": "preserve the selected virtual-environment interpreter path" + "corrective_change": "emit official Qwen projection names and shapes, preserve fixed MLP width, add strict official reload gate" } }, "source": { "repository": "https://github.com/atimics/holostuff.git", - "commit": "2f32151c6d44c80a7e2ff98c0696650cd1d9c6f1" + "commit": "f399cd6bc7f24383a792fdcf32c43a4e1197d4f1" }, "ilxyr": { "repository": "https://github.com/cenetex/ilXyr.git", @@ -42,10 +42,10 @@ }, "corpora": { "installation": { - "handle": "git:a04ab5692be38f06120aba4b0bc5e2a284eb2c79:REFERENCE.md", + "handle": "git:f399cd6bc7f24383a792fdcf32c43a4e1197d4f1:REFERENCE.md", "license": "MIT", - "bytes": 2300089, - "sha256": "d6905f043e7856b93b2dd72dac5fa0dc593898c55d6c54c51f3153f4317d6b7f" + "bytes": 2303549, + "sha256": "59f551d452cfb0664deeb6dfbe5b98598abb1633cc38a5bb020ec100ac3cd33a" }, "evaluation": { "handle": "https://www.gutenberg.org/ebooks/18", @@ -63,7 +63,8 @@ "confidence": 0.95, "bootstrap": "paired moving block", "spectral_filtering": false, - "installer_experimental": true + "installer_experimental": true, + "official_transformers_strict_reload": true }, "dependencies": { "python": "3.12", @@ -97,31 +98,10 @@ "total_ceiling_usd": 10.0 }, "publication": { - "temporary_staging": "s3://zero-training-022118847419/qwen35-acceptance/2f32151-2fc06364-a6c9d113/v2-formal/", + "temporary_staging": "s3://zero-training-022118847419/qwen35-acceptance/f399cd6-2fc06364-a6c9d113/v3-formal/", "review_index": "https://github.com/atimics/holostuff/pull/3", "permanent_transport": "Arweave", - "permanent_root": "https://arweave.net/f35OgXuOoulk9EcndJBNWIgZ5tmaQ6DRzeINP7ARZTU/", - "permanent_root_transaction": "f35OgXuOoulk9EcndJBNWIgZ5tmaQ6DRzeINP7ARZTU", - "publication_manifest_sha256": "21874555c7f9e73d20b9d11599c79236f8f40267ee975f703ffc053eb6b1f9ab", "publish_all_outcomes": true, "exclude_model_weights": true - }, - "terminal": { - "completed_at": "2026-08-13T04:51:28.293Z", - "formal_duration_seconds": 15891.556, - "runner_exit_code": 1, - "timed_out": false, - "resolved_outcome": "execution_failure", - "scientific_result": false, - "failure_gate": "official_transformers_checkpoint_reload", - "aws_instance": "i-059aab8bf1d3d87c9", - "estimated_compute_usd": 4.71, - "evidence_bundle_sha256": "df2ba31608f9039e9528a43a93d4766c8e8feb023668a1b9ffc1c85553210c1c", - "ledger_verification": { - "valid": true, - "objects_checked": 15, - "events_checked": 16 - }, - "retry_authorized": false } } diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/README.md b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/README.md new file mode 100644 index 0000000..2983c68 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/README.md @@ -0,0 +1,76 @@ +# Qwen3.5 acceptance experiment: v3 terminal record + +- Experiment: `lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance` +- Terminal state: `execution_failure` — neither `GO` nor `NO-GO` +- Scientific result admitted by ilxyr: no +- AWS instance: `i-0f57dba006efd2f32` +- Formal execution: 2026-08-13 18:52:51.758Z–21:21:28.229Z + (`2h 28m 36.471s`), executor exited `0`, metric ingestion failed +- AWS lifecycle: launched 18:50:56Z and self-terminated after evidence upload; + estimated on-demand compute charge `$2.66` plus a few cents of storage +- Ledger verification: valid (`10` objects and `11` events) +- Evidence bundle SHA-256: + `84db563708d77fb13ff93f072dec85f19e96ec7cfe33a22e56e639df78dfcc60` + +## What happened + +The AWS preflight passed every frozen source, model, corpus, version, timeout, +token, interpreter, tokenizer, and disabled-spectral-path check. ilxyr admitted +the project after all twelve admission gates passed. The runner then completed +the full 4,096-position paired evaluation and emitted a metric object, but four +human-readable Qwen diagnostics preceded that JSON on standard output. ilxyr's +strict parser therefore rejected the stream with: + +```text +executor output is not valid metric JSON: expected value at line 1 column 1 +``` + +No `EvidenceRecorded` event was appended, `latest_evidence` remains null, and +the diagnostic metric artifact must not be relabeled as an admitted result. + +## Diagnostic observations + +The unpromoted runner artifact is still operationally useful: + +- all `4,096` required paired positions were evaluated; +- reference-logit relative error was `6.7877e-7` and tokenizer parity passed; +- the emitted 2,112.9 MB checkpoint reloaded in leCore and official + Transformers; +- original perplexity was `7.9062`, installed perplexity was `10.3576`, a + `31.0053%` regression; +- the paired moving-block interval in loss space was + `[0.20260, 0.34247]` nats, wholly on the harmful side of the frozen gate; +- peak resident memory was `24,306.5` MB (23.74 GiB); +- text generation and vision smoke did not pass. The official processor smoke + reported that Torchvision was absent from the environment. + +These observations are not a formal `NO-GO`: the metric envelope was never +admitted. They do, however, make another paid treatment run scientifically +unattractive until the parser and vision dependency blockers are fixed and the +large quality regression is understood with a cheap local or fixture run. + +## Cleanup and retry policy + +No retry was launched. Instance `i-0f57dba006efd2f32` is terminated and +security group `sg-08e840b4749a47406` was deleted after detachment. The frozen +v3 identity is closed. + +## Evidence map + +- `project/` contains the exact ilxyr project admitted on AWS. +- `result/` contains preflight, installer, runtime, status, metric diagnostic, + environment, checksum, and cleanup records. +- `ledger/` is the independently reverified `.ilxyr` event store, renamed for + Git publication. +- `publication-manifest.json` is the canonical hash-and-size index. +- `publication-receipt.json` records the permanent Arweave transactions and + verified downloads. + +The redistributable corpora and the 1.7 GB public model checkpoint are omitted +from Git. Corpus hashes remain frozen in the manifests. The permanent Arweave +publication includes reconstructable chunks of both corpora and every other +evidence file, but excludes model weights. + +The original AWS evidence bundle remains at +`s3://zero-training-022118847419/qwen35-acceptance/f399cd6-2fc06364-a6c9d113/v3-formal/qwen35-evidence.tgz`. + diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/config.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/config.json new file mode 100644 index 0000000..c7becc4 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/config.json @@ -0,0 +1,5 @@ +{ + "ledger_mode": "single_writer", + "object_hash": "sha256", + "schema": "ilxyr.workspace.v1" +} \ No newline at end of file diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/events.jsonl b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/events.jsonl new file mode 100644 index 0000000..acbc5d1 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/events.jsonl @@ -0,0 +1,11 @@ +{"actor":{"id":"model://codex/lecore-qwen/research-director","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.hypothesis","artifact_ref":"artifact://sha256/d3c31ef2f50da0924c719329b24028ecf54721d454811dd9aacf4ff46de4e3b3","event_hash":"b9c536497df105a06ebdb04d9a3c91a20e96626f0c0f7fbf575e9359833cd025","event_type":"ContributionSubmitted","occurred_at_ms":1786647171209,"schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/statistical-reviewer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/statistical-reviewer"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.foundation","artifact_ref":"artifact://sha256/d1c0d6083c36dc5e277e4a0ad0c7f30338431c50d99754b938eaeb59f1967d1d","event_hash":"d8f9662cdee51d3d5f3406ae333d4cd61695313368e74e3748fd18893d62de84","event_type":"ContributionSubmitted","occurred_at_ms":1786647171214,"previous_event":"b9c536497df105a06ebdb04d9a3c91a20e96626f0c0f7fbf575e9359833cd025","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/engineering-reviewer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/engineering-reviewer"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.engineering-review","artifact_ref":"artifact://sha256/08d81fe6c3b0a1d2f6c93d08b7d5e51ef5b4044c33c2f7fc30d4a61210c38a2f","event_hash":"3b74145abf0410641115b1836ae9af6e7017e1e2922b8f341cb31d5d6a7a9060","event_type":"ContributionSubmitted","occurred_at_ms":1786647171218,"previous_event":"d8f9662cdee51d3d5f3406ae333d4cd61695313368e74e3748fd18893d62de84","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/experiment-designer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/experiment-designer"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.experiment-design","artifact_ref":"artifact://sha256/09c2a3c1d5060228405fa5cc9147bc5a4c61579a52c0dba4c3d55ee1d03ac5fe","event_hash":"5336d08c7265c12937ed86b51314b57d70e92e6a56796a36af917073da69cef3","event_type":"ContributionSubmitted","occurred_at_ms":1786647171222,"previous_event":"3b74145abf0410641115b1836ae9af6e7017e1e2922b8f341cb31d5d6a7a9060","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/research-director","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/research-director"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance","artifact_ref":"artifact://sha256/c5399dba347f001bf553941b88aae7e74e0609bbded168252ae731eff6a39434","event_hash":"d122354665a80ec2f1342c45213e12d4def633d2612e0d49eb1c3ade371d5c0e","event_type":"ExperimentCompiled","occurred_at_ms":1786647171227,"previous_event":"5336d08c7265c12937ed86b51314b57d70e92e6a56796a36af917073da69cef3","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/forecaster-empirical","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.forecast.empirical","artifact_ref":"artifact://sha256/0e519bb9ba1feb38105fd2d108e5bedf21a0c390f1bc248cf6c33a1a852f24d9","event_hash":"a3e673a577e645d96b6135ec5d86f234aa8e6b14c2c7be1e8b5ef817beed16dc","event_type":"ForecastSubmitted","occurred_at_ms":1786647171232,"previous_event":"d122354665a80ec2f1342c45213e12d4def633d2612e0d49eb1c3ade371d5c0e","schema":"ilxyr.event.v1"} +{"actor":{"id":"model://codex/lecore-qwen/forecaster-mechanistic","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-mechanistic"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.forecast-mechanistic","artifact_ref":"artifact://sha256/e0cf1ad5fb075f3e94ac70f9c426fb63f33cf7e67ad076a17d93381dc465dc76","event_hash":"0b559494e62aca4fb4a9d58a6853afa159c088ebd37a1fa10e7f896655a4d85d","event_type":"ForecastSubmitted","occurred_at_ms":1786647171236,"previous_event":"a3e673a577e645d96b6135ec5d86f234aa8e6b14c2c7be1e8b5ef817beed16dc","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://lecore/qwen-experiment-generator","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.funding","artifact_ref":"artifact://sha256/0067fb1dc8736fd862f156be1d17d9c1efea28fbb043c72381a951e5a6b100c1","event_hash":"93c8b61dd23e46c0e63e5fa838d8756e2ce1dfb16929b6b7325ab6e5fabfad54","event_type":"FundingCommitted","occurred_at_ms":1786647171241,"previous_event":"0b559494e62aca4fb4a9d58a6853afa159c088ebd37a1fa10e7f896655a4d85d","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/admission-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance","artifact_ref":"artifact://sha256/64f1db0c4ea0afdf533b17183c114725032a320e03bcd3cb72a9e12f18569f47","event_hash":"254d3ba3079ed9bb2e8309533b3b2f6ac3ee9c91b1e26c781039fab9368fa440","event_type":"AdmissionDecided","occurred_at_ms":1786647171245,"previous_event":"93c8b61dd23e46c0e63e5fa838d8756e2ce1dfb16929b6b7325ab6e5fabfad54","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/local-executor-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance","event_hash":"e737172dcf144c29dc1dc01cff23534e441c19b7504c266e9cfab9d4404446c4","event_type":"ExecutionStarted","occurred_at_ms":1786647171731,"previous_event":"254d3ba3079ed9bb2e8309533b3b2f6ac3ee9c91b1e26c781039fab9368fa440","schema":"ilxyr.event.v1"} +{"actor":{"id":"service://ilxyr/local-executor-v1","kind":"service"},"aggregate_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance","artifact_ref":"artifact://sha256/50266fe71e4f151846f15ce6379d97ada04a5968209fe8eae45fab477d249eea","event_hash":"b8d4f3a9d8bf25181788abd3f8fb370614e33b6e24a3e0d5578976cec2008de1","event_type":"ExperimentCompleted","occurred_at_ms":1786656088231,"previous_event":"e737172dcf144c29dc1dc01cff23534e441c19b7504c266e9cfab9d4404446c4","schema":"ilxyr.event.v1"} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/0067fb1dc8736fd862f156be1d17d9c1efea28fbb043c72381a951e5a6b100c1 b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/0067fb1dc8736fd862f156be1d17d9c1efea28fbb043c72381a951e5a6b100c1 new file mode 100644 index 0000000..9ea1723 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/0067fb1dc8736fd862f156be1d17d9c1efea28fbb043c72381a951e5a6b100c1 @@ -0,0 +1 @@ +{"compute_credits":100,"experiment_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance","funder":{"id":"service://lecore/qwen-experiment-generator","kind":"service"},"id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.funding","rationale":"Reserve one complete public-weight Qwen acceptance run under the frozen contract.","schema":"ilxyr.funding.v1"} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/08d81fe6c3b0a1d2f6c93d08b7d5e51ef5b4044c33c2f7fc30d4a61210c38a2f b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/08d81fe6c3b0a1d2f6c93d08b7d5e51ef5b4044c33c2f7fc30d4a61210c38a2f new file mode 100644 index 0000000..c9e8625 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/08d81fe6c3b0a1d2f6c93d08b7d5e51ef5b4044c33c2f7fc30d4a61210c38a2f @@ -0,0 +1 @@ +{"actor":{"id":"model://codex/lecore-qwen/engineering-reviewer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/engineering-reviewer"},"body":"The shell-free runner uses absolute paths, records the exact leCore commit and checker hashes, keeps spectral filtering disabled, invokes the layer-prepending path only with its experimental acknowledgement, records peak memory, reloads the emitted artifact through both leCore and official Transformers, and exercises the official text and image-text interfaces.","claims":["The emitted stdout is exactly the ilxyr metrics/source envelope.","Installer logs and a human-readable metrics artifact are retained beside the output checkpoint."],"confidence":0.85,"id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.engineering-review","input_refs":["lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.hypothesis","lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.foundation"],"schema":"ilxyr.contribution.v1","stage":"engineering_review","title":"Qwen acceptance runner and provenance boundary"} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/09c2a3c1d5060228405fa5cc9147bc5a4c61579a52c0dba4c3d55ee1d03ac5fe b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/09c2a3c1d5060228405fa5cc9147bc5a4c61579a52c0dba4c3d55ee1d03ac5fe new file mode 100644 index 0000000..94e3114 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/09c2a3c1d5060228405fa5cc9147bc5a4c61579a52c0dba4c3d55ee1d03ac5fe @@ -0,0 +1 @@ +{"actor":{"id":"model://codex/lecore-qwen/experiment-designer","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/experiment-designer"},"body":"Execute once against the content-bound public checkpoint, installation corpus, and separate held-out evaluation corpus. Do not tune thresholds or replace either corpus after admission. Resolve accepted only when source cleanliness, tokenizer parity, reference-logit parity, the paired statistical gate, leCore disk reload, official Transformers reload, official text generation, and official vision smoke all pass. An official model capability failure emits zero-valued gates and is preserved as rejected evidence; dependency or runner failure remains execution_failure.","claims":["Accepted and rejected are exhaustive for a valid metrics envelope.","Runtime or dependency failure resolves separately as execution_failure."],"confidence":0.95,"id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.experiment-design","input_refs":["lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.hypothesis","lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.foundation","lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.engineering-review"],"schema":"ilxyr.contribution.v1","stage":"experiment_design","title":"One-shot Qwen3.5 installation acceptance run"} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/0e519bb9ba1feb38105fd2d108e5bedf21a0c390f1bc248cf6c33a1a852f24d9 b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/0e519bb9ba1feb38105fd2d108e5bedf21a0c390f1bc248cf6c33a1a852f24d9 new file mode 100644 index 0000000..bf1b982 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/0e519bb9ba1feb38105fd2d108e5bedf21a0c390f1bc248cf6c33a1a852f24d9 @@ -0,0 +1 @@ +{"experiment_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance","forecaster":{"id":"model://codex/lecore-qwen/forecaster-empirical","kind":"model","model_ref":"model://openai/codex/gpt-5/2026-08-12/qwen-acceptance-design/forecaster-empirical"},"id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.forecast.empirical","probabilities":{"accepted":0.2,"execution_failure":0.15,"rejected":0.65},"rationale":"Structural rehearsal passes, but no complete post-memory-fix real-Qwen install and powered acceptance run has succeeded yet.","schema":"ilxyr.forecast.v1","stake":5} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/50266fe71e4f151846f15ce6379d97ada04a5968209fe8eae45fab477d249eea b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/50266fe71e4f151846f15ce6379d97ada04a5968209fe8eae45fab477d249eea new file mode 100644 index 0000000..cf6abcd --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/ledger/objects/sha256/50266fe71e4f151846f15ce6379d97ada04a5968209fe8eae45fab477d249eea @@ -0,0 +1 @@ +{"completed_at_ms":1786656088229,"exit_code":0,"experiment_id":"lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance","id":"run:lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance:1786647171758","metrics":{},"output_error":"executor output is not valid metric JSON: expected value at line 1 column 1","output_truncated":false,"schema":"ilxyr.run.v1","started_at_ms":1786647171758,"stderr":"[transformers] `torch_dtype` is deprecated! Use `dtype` instead!\n[transformers] The fast path is not available because one of the required library is not installed. Falling back to torch implementation. To install follow https://github.com/fla-org/flash-linear-attention#installation and https://github.com/Dao-AILab/causal-conv1d\n\rLoading weights: 0%| | 0/320 [00:00", + "/opt/qwen35-acceptance-v3/project/hypothesis.json" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance-v3/project/foundation.json" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance-v3/project/engineering-review.json" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "contribute", + "", + "/opt/qwen35-acceptance-v3/project/experiment-design.json" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "compile", + "", + "/opt/qwen35-acceptance-v3/project/experiment.json" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "forecast", + "", + "/opt/qwen35-acceptance-v3/project/forecast-empirical.json" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "forecast", + "", + "/opt/qwen35-acceptance-v3/project/forecast-mechanistic.json" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "fund", + "", + "/opt/qwen35-acceptance-v3/project/funding.json" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "admit", + "", + "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "run", + "", + "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance", + "--execute" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "status", + "", + "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance" + ], + [ + "/opt/qwen35-acceptance-v3/ilxyr/target/release/ilxyr", + "verify", + "" + ] + ], + "corpora": { + "evaluation": { + "bytes": 1213410, + "path": "/opt/qwen35-acceptance-v3/inputs/federalist-papers.txt", + "sha256": "a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + }, + "installation": { + "bytes": 2303549, + "path": "/opt/qwen35-acceptance-v3/lecore/REFERENCE.md", + "sha256": "59f551d452cfb0664deeb6dfbe5b98598abb1633cc38a5bb020ec100ac3cd33a" + } + }, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance", + "experiment_version": 3, + "installed_dir": "/opt/qwen35-acceptance-v3/installed-checkpoint", + "model_digest": "20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "model_files": [ + { + "bytes": 11544, + "path": "LICENSE", + "sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a" + }, + { + "bytes": 61705, + "path": "README.md", + "sha256": "87a163af54f32fa608a0f8d3ac67945c53dd2b4a7c96740b3d7fdc28e8458864" + }, + { + "bytes": 7755, + "path": "chat_template.jinja", + "sha256": "273d8e0e683b885071fb17e08d71e5f2a5ddfb5309756181681de4f5a1822d80" + }, + { + "bytes": 2907, + "path": "config.json", + "sha256": "b90b86f35c8e6925ef74ee04d0e758f0a845c83a42089ad82bbaa948de9b4204" + }, + { + "bytes": 3353259, + "path": "merges.txt", + "sha256": "a9d356d7bdf1ef4949e3e748e95b8e10ad9d4e2e838eddc38a0a7b6b94d1db8d" + }, + { + "bytes": 1746942600, + "path": "model.safetensors-00001-of-00001.safetensors", + "sha256": "04b1c301231dd422b8860db31311ab2721511346a32cb1e079c4c4e5f1fe4696" + }, + { + "bytes": 50900, + "path": "model.safetensors.index.json", + "sha256": "d8a08838a613b025eb7952ed9db11696213e57e76a375661ef5c12f9dd5dcf4e" + }, + { + "bytes": 390, + "path": "preprocessor_config.json", + "sha256": "27225450ac9c6529872ee1924fcb0962ff5634834f817040f444118116f4e516" + }, + { + "bytes": 12807982, + "path": "tokenizer.json", + "sha256": "5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42" + }, + { + "bytes": 16709, + "path": "tokenizer_config.json", + "sha256": "49e2b6e395f959f077f1e992b338919c0d4a9732fc6e613995e06557f843500c" + }, + { + "bytes": 385, + "path": "video_preprocessor_config.json", + "sha256": "7768af27c1fafa9cc9011c1dc20067e03f8915e03b63504550e11d5066986d13" + }, + { + "bytes": 6722759, + "path": "vocab.json", + "sha256": "ce99b4cb2983d118806ce0a8b777a35b093e2000a503ebde25853284c9dfa003" + } + ], + "schema": "lecore.ilxyr-project.v1", + "source_commit": "f399cd6bc7f24383a792fdcf32c43a4e1197d4f1" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-manifest.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-manifest.json new file mode 100644 index 0000000..38ead74 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-manifest.json @@ -0,0 +1,459 @@ +{ + "evidence_ref": null, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance", + "files": [ + { + "bytes": 3590, + "media_type": "text/markdown", + "path": "README.md", + "sha256": "d7b27ab2da53b22f148ba8c9dd13c9cd7b419bff1d45066e803c38ef08846e54" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-000", + "sha256": "b2ce94cfafce4d2de52f6e9703ef1ec65149e5ccabf52f6cbcab8fa8a5f33bc2" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-001", + "sha256": "11873d7727ad50f1c24050d3bc1a60c88cd2e916e295691933f55912f481869c" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-002", + "sha256": "e60749ca3f54ef2ba6d39d8e78fdea7485ee1efb389bdc16ebab14d660dc5f27" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-003", + "sha256": "cfc4c41bef590b6d32e7d54466fd7358e15c38d9a3c90f0b9ff17157e7e3fa00" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-004", + "sha256": "a128b48addf4119dd94d797c202d2caf3cff26c6909a76acd0e20904c90aab6a" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-005", + "sha256": "be54b5ac35e53aedf51e929daf9aa34c6e39457a1b3a1264fcee9c223a41de7e" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-006", + "sha256": "7524cecef88df581cbbd4878e5da1031d0e153d66c69667e8b8d5a3253c8ce39" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-007", + "sha256": "fd7ca72424736d4fae0b5c08c10a82bacb617c717789eb145de25a346a206caa" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-008", + "sha256": "9ef0f71a50fdaa82e541e168576a4c9a500146f77a0ec17c671299b0769e3f8f" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-009", + "sha256": "2e63a20ebe80748249dc46baaaf93715cc13e31f4fc51f271de89510c58ef133" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-010", + "sha256": "b5866ace71ddee8d6497f35e007098d6d5d06223e390972196353662ce613858" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-011", + "sha256": "c8ad9aedc155b10baeb7f98ce73776cba2c993b3525ec704e16be6d9d4cd42a5" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-012", + "sha256": "0fd592c0f5838f83bfed355986fb7d516fb431a0acb8d487540c7806105aebc9" + }, + { + "bytes": 90000, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-013", + "sha256": "af44a54d51781c61cb0c795e6c65fc55521093426f4add926fece19bf0d247f9" + }, + { + "bytes": 65327, + "media_type": "application/octet-stream", + "path": "bundle-parts/qwen35-evidence.tgz.part-014", + "sha256": "d41dfed6add2f7254c16c943c6b3b48b82d37cb73b8da0571da508c746c1a38f" + }, + { + "bytes": 97, + "media_type": "application/json", + "path": "ledger/config.json", + "sha256": "2301b853e1e01b8844970b90e72036bed57c580ded81a857ec7faf27f8deca66" + }, + { + "bytes": 6424, + "media_type": "application/octet-stream", + "path": "ledger/events.jsonl", + "sha256": "ee844391a410ec83c471939d212572b94cc25168597bee3741081335c093ae0f" + }, + { + "bytes": 424, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/0067fb1dc8736fd862f156be1d17d9c1efea28fbb043c72381a951e5a6b100c1", + "sha256": "0a4e375bb3d60adf7471c5df81e7ff352c5f0028311cacec3b99e923dcbb072d" + }, + { + "bytes": 1161, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/08d81fe6c3b0a1d2f6c93d08b7d5e51ef5b4044c33c2f7fc30d4a61210c38a2f", + "sha256": "3f315e74ab91b284cca39b3fbf0eac954df5e52d7d03473c3761281d6e89b5c3" + }, + { + "bytes": 1438, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/09c2a3c1d5060228405fa5cc9147bc5a4c61579a52c0dba4c3d55ee1d03ac5fe", + "sha256": "86bb9e72ca869bd74e2f7fb613721f56c472cc0bb2cd20df215ff173c91e6934" + }, + { + "bytes": 643, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/0e519bb9ba1feb38105fd2d108e5bedf21a0c390f1bc248cf6c33a1a852f24d9", + "sha256": "26aae57ffc4c121a23ffdeda5162875efac2639047cd3a12d0e18beb9e2cb20b" + }, + { + "bytes": 6116, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/50266fe71e4f151846f15ce6379d97ada04a5968209fe8eae45fab477d249eea", + "sha256": "7ef6e8c156e5cdade38156136fab5ef6b364aed68e0ba61dd18be2ede13c9140" + }, + { + "bytes": 1381, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/64f1db0c4ea0afdf533b17183c114725032a320e03bcd3cb72a9e12f18569f47", + "sha256": "8668ba60aecbf1fa3f70c9f2bc1a8d6c9ecb846d64c1793a5c1406d1db7b08a8" + }, + { + "bytes": 8770, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/c5399dba347f001bf553941b88aae7e74e0609bbded168252ae731eff6a39434", + "sha256": "24153c1cf078e0d8e9e9b95e8e9626044fecd757d1eb76d33211baad1f26f899" + }, + { + "bytes": 1125, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/d1c0d6083c36dc5e277e4a0ad0c7f30338431c50d99754b938eaeb59f1967d1d", + "sha256": "ea312bd62a285436586532599ea75abff430f256660564791151f3254a7ac609" + }, + { + "bytes": 917, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/d3c31ef2f50da0924c719329b24028ecf54721d454811dd9aacf4ff46de4e3b3", + "sha256": "a33d8a8fdb0bff9dd61a170a5b4a03452ace5b8ef67129e66eb9d073fc8adf38" + }, + { + "bytes": 680, + "media_type": "application/octet-stream", + "path": "ledger/objects/sha256/e0cf1ad5fb075f3e94ac70f9c426fb63f33cf7e67ad076a17d93381dc465dc76", + "sha256": "9b12a22adc2351843eb4cefd82f8ccb6a62b0c2515391fd9685b8ac12737fe40" + }, + { + "bytes": 1245, + "media_type": "application/json", + "path": "project/engineering-review.json", + "sha256": "abafdac1871b29f72eb1c618df6d04dfcee14962c8d6af36c12e6c9fd73b9788" + }, + { + "bytes": 1527, + "media_type": "application/json", + "path": "project/experiment-design.json", + "sha256": "58f92e84f8e79ecbc5cebea83b7bd34687da88e840564462a8b5c76d9c63def2" + }, + { + "bytes": 9051, + "media_type": "application/json", + "path": "project/experiment.json", + "sha256": "04f3b73c12874b14235e68496e5f01561ed0d8dc6130a7bacbba337b68c76ffb" + }, + { + "bytes": 714, + "media_type": "application/json", + "path": "project/forecast-empirical.json", + "sha256": "b64c576f46bc70bdf0b623609069d3c3e1e825d52c6f8790c781cb3bfc852a12" + }, + { + "bytes": 751, + "media_type": "application/json", + "path": "project/forecast-mechanistic.json", + "sha256": "5cad1dbb4adf5eba4c9f098fd878f1a8c610f22998ec3d005fd1d969e89f366d" + }, + { + "bytes": 1204, + "media_type": "application/json", + "path": "project/foundation.json", + "sha256": "46dae610e7fbead0830d29a28cc636a7fda14c282956ade9ab7fa3752a5e9884" + }, + { + "bytes": 464, + "media_type": "application/json", + "path": "project/funding.json", + "sha256": "a8a24296eb062a8fec4b963404f03ac2b7d3c3b47622e53f9abeda55d9780a80" + }, + { + "bytes": 993, + "media_type": "application/json", + "path": "project/hypothesis.json", + "sha256": "6223bba67ec648fa6c0ed47a8e38ece1820b0700ed2f5c8e7e9317af5989e125" + }, + { + "bytes": 4888, + "media_type": "application/json", + "path": "project/project.json", + "sha256": "ff5a6b22e2ecdb9a5cb946ae51f88240e71628cf80cdd97f1a1ed1a5fb8d47ee" + }, + { + "bytes": 1978, + "media_type": "text/plain", + "path": "result/acceptance-artifacts/install.log", + "sha256": "82de50a6c5defe836080bede9c1d9764cd13286da2a70bbf1f031903ca0d091c" + }, + { + "bytes": 2722, + "media_type": "application/json", + "path": "result/acceptance-artifacts/metrics.json", + "sha256": "34da55d92d3c4e4a7f273eb82d31fc09b106ff077ffb61c1acc68205a1df81dd" + }, + { + "bytes": 686, + "media_type": "application/json", + "path": "result/admission-preflight.json", + "sha256": "e8bd84c73e7d311ed4ea9866e5a13a8985b982a2a41164769178340018bca267" + }, + { + "bytes": 408, + "media_type": "application/json", + "path": "result/aws-cleanup-receipt.json", + "sha256": "e74f5143af8bf797f3e14accf4a44121c1ccf8e5f8dd16bf55e09c6e4dd3967e" + }, + { + "bytes": 479, + "media_type": "application/json", + "path": "result/aws-instance-identity.json", + "sha256": "50c98cc1ab5530329f2649088baab8fb183b0447d7a0f5326b334d0f0d23ba63" + }, + { + "bytes": 42208, + "media_type": "text/plain", + "path": "result/cloud-init-output.log", + "sha256": "81b03125cb347aa2aa563690ae4b0f850bd3cf4e593dddc83a0b04f68974a790" + }, + { + "bytes": 388, + "media_type": "application/json", + "path": "result/corpus-manifest.json", + "sha256": "26b51f1f3aaceaf4b10f4765706e9cebaef8db2cb104b9567ba6cbe84fed9fbf" + }, + { + "bytes": 4829, + "media_type": "text/plain", + "path": "result/environment.txt", + "sha256": "14811419f69523f40b1c7fbe34d37efc2d197a7ef1218215904aee23b41b4f92" + }, + { + "bytes": 112, + "media_type": "application/octet-stream", + "path": "result/evidence-bundle.sha256", + "sha256": "bf043cc7821cbbd8102b4393a80d78153b987cc5be27171a4cccd9fc14cf3528" + }, + { + "bytes": 289, + "media_type": "application/json", + "path": "result/execution-summary.json", + "sha256": "5398bac7aace482f7ac0b80ae8a6ea47d4e1cfca5c9b3abdc33d7f661d04a16b" + }, + { + "bytes": 222, + "media_type": "application/json", + "path": "result/generation.json", + "sha256": "20f68e992c7c47e3738496413039e63c41c41703bacd29f3580e8cc3f3d50d56" + }, + { + "bytes": 1813, + "media_type": "application/json", + "path": "result/ilxyr-admission.json", + "sha256": "350d983369f2a0bcc69e32eb19329c83bc6e49f46a2119e7afc776d695a60b05" + }, + { + "bytes": 259, + "media_type": "text/plain", + "path": "result/ilxyr-commit.txt", + "sha256": "32873b519f7170a47a099930a325f4bb29d04021b822fa7f5387c94a9976add4" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-compile.json", + "sha256": "d87d049e15140c53ecce33e04e171ef9a082576f71d53b79bc24b73d290c7066" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-engineering-review.json", + "sha256": "e23e13f82216d560fbfe1ab8b331667f0c79a0b1fd53b8731f42fb852366f390" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-experiment-design.json", + "sha256": "2218e1405e2b8dbc7f0baa43ee08ffa098e8d14eb3e59ca1713415d28b4a61b4" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-foundation.json", + "sha256": "d168152093262d0223cf7a3ae44dd21969fdc3e8f459c0addd534bbcc04f8733" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-contribute-hypothesis.json", + "sha256": "9946d252fa0e892d6da53330f936e0209ea70d6c868fa104c12a835f4042278a" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-forecast-empirical.json", + "sha256": "c455457466e4c90986146a7c2650020d86f9596592bedadd5772a2ef1f6fc3be" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-forecast-mechanistic.json", + "sha256": "7fef0ea79eaa316a2ee4701ecd3b06b84ab51d816714ba210f79208222528168" + }, + { + "bytes": 107, + "media_type": "application/json", + "path": "result/ilxyr-funding.json", + "sha256": "acb395a1688affcfd30d14c2dcdb188666fb0b2f96aace07e5e6b5d37ea138ed" + }, + { + "bytes": 80, + "media_type": "application/json", + "path": "result/ilxyr-init.json", + "sha256": "ddb1232746d65458dce7ea3b455a00850cd9fbe5caf752e21a346f913a4893c8" + }, + { + "bytes": 0, + "media_type": "application/json", + "path": "result/ilxyr-run.json", + "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" + }, + { + "bytes": 67, + "media_type": "application/octet-stream", + "path": "result/ilxyr-run.stderr", + "sha256": "789f19a75ebec95fedda52b1fa0817eca710391669409ba2f59f83f32ff6117a" + }, + { + "bytes": 8551, + "media_type": "application/json", + "path": "result/ilxyr-status.json", + "sha256": "5410d8b166ea41eaacaa65adb0ec3fbd206f9a859d48f979e9cd7f5bb0ed11de" + }, + { + "bytes": 0, + "media_type": "application/octet-stream", + "path": "result/ilxyr-status.stderr", + "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" + }, + { + "bytes": 69, + "media_type": "application/json", + "path": "result/ilxyr-verify.json", + "sha256": "7220969c8bef798767522784933b4abe4ae820c26d3ea560291545df66cd57c2" + }, + { + "bytes": 0, + "media_type": "application/octet-stream", + "path": "result/ilxyr-verify.stderr", + "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" + }, + { + "bytes": 3576, + "media_type": "application/json", + "path": "result/launch-manifest.json", + "sha256": "3cf9b3914a9d1cca34b6a86f136f6618778012dbb8e502caa779cd59dfa615cf" + }, + { + "bytes": 283, + "media_type": "text/plain", + "path": "result/lecore-commit.txt", + "sha256": "8a47b1fa8b399f480d9d0e6390b86026bcb83ab93563d47f74eb060203d0e270" + }, + { + "bytes": 0, + "media_type": "text/plain", + "path": "result/lecore-status.txt", + "sha256": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855" + }, + { + "bytes": 2023, + "media_type": "application/json", + "path": "result/model-manifest.json", + "sha256": "f352ac2c17ee1bea38f70d7ab1975e3297954423cb74f5330383134e4b6c493a" + }, + { + "bytes": 588, + "media_type": "text/plain", + "path": "result/requirements-resolved.txt", + "sha256": "ffc5f6316f74e8e6c465b3845f6b4feb9e3b534160434299de9b05c809d7f647" + }, + { + "bytes": 181, + "media_type": "application/json", + "path": "result/runner-status.json", + "sha256": "791cd435caaa12f52a83d12f3f2a9c2cfdbb65141a54195739357fc5bcb8b5c9" + }, + { + "bytes": 181, + "media_type": "application/json", + "path": "result/s3-terminal-status.json", + "sha256": "ba6b24bdad77b6d61bb8bee17bf22f5e0c01ade38a0bc5ae437525b3936c95df" + }, + { + "bytes": 38672, + "media_type": "text/plain", + "path": "result/user-data.log", + "sha256": "2bfab64072156905e4a18cf1997ef0b0baa8448fa838edd9dabe3df3f3bd1620" + }, + { + "bytes": 13425, + "media_type": "application/x-sh", + "path": "result/user-data.sh", + "sha256": "1fa0d1e2f06043d7327fed4ed2074963a5f11c27d170d03b1dedd7db91a1f1e9" + } + ], + "ledger_verification": { + "events_checked": 11, + "objects_checked": 10, + "valid": true + }, + "resolved_outcome": "execution_failure", + "run_ref": "run:lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance:1786647171758", + "schema": "lecore.qwen35-publication-manifest.v1", + "scientific_result_admitted": false +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-receipt.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-receipt.json new file mode 100644 index 0000000..a90224e --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-receipt.json @@ -0,0 +1,67 @@ +{ + "schema": "lecore.qwen35-arweave-publication-receipt.v1", + "published_at": "2026-08-13T21:29:26Z", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance", + "resolved_outcome": "execution_failure", + "scientific_result_admitted": false, + "publisher": { + "service": "ArDrive Turbo signed free upload", + "sdk": "@ardrive/turbo-sdk@1.42.0", + "ownership": "The uploads used one-use in-memory Arweave identities. Their private keys were never persisted." + }, + "publication_manifest": { + "path": "publication-manifest.json", + "bytes": 16447, + "sha256": "19ac497b7e55881422e28cc8a118085a48e685a8bd5c5e7ee8abe2839ef8f8bb", + "transaction_id": "1SvpypXJlC2JNqX5QyF08ExECvjiYxm0B-pmWGaNqKY" + }, + "arweave_path_manifest": { + "transaction_id": "h96JlX-4ttZsPZdYRBuT-_JNO2s0lS18hoFjM4ojYSw", + "bytes": 8605, + "sha256": "844a8410e74b91bdc5c15fa4435d0c4ccdab21bec4b8a8fd858431825e880674", + "item_count": 75, + "gateway_url": "https://arweave.net/h96JlX-4ttZsPZdYRBuT-_JNO2s0lS18hoFjM4ojYSw/", + "index_url": "https://arweave.net/h96JlX-4ttZsPZdYRBuT-_JNO2s0lS18hoFjM4ojYSw/publication-manifest.json", + "confirmed_cache_url": "https://turbo-gateway.com/h96JlX-4ttZsPZdYRBuT-_JNO2s0lS18hoFjM4ojYSw/", + "owner": "Tp9rnI0lVaMBpPbhbofJXTBI6yJtfoxrI5KWM4GEXKE" + }, + "evidence_bundle": { + "original_name": "qwen35-evidence.tgz", + "bytes": 1325327, + "sha256": "84db563708d77fb13ff93f072dec85f19e96ec7cfe33a22e56e639df78dfcc60", + "arweave_parts": 15, + "part_size_bytes": 90000, + "reconstruction": "Concatenate bundle-parts/qwen35-evidence.tgz.part-000 through part-014 in lexical order, then verify SHA-256. The bundle contains no model weights." + }, + "selected_items": { + "ledger_events": { + "transaction_id": "REmYCo1Hta3RPPNYipNshvJ9ec8onIVndNGN63Fw2Zo", + "sha256": "ee844391a410ec83c471939d212572b94cc25168597bee3741081335c093ae0f" + }, + "ledger_verification": { + "transaction_id": "-5WEUj_anDpdkxIR_sgUNbpFvWFWnVFmLMlLrNuqqjU", + "sha256": "7220969c8bef798767522784933b4abe4ae820c26d3ea560291545df66cd57c2" + }, + "diagnostic_metrics": { + "transaction_id": "TpXJ4v8jUJuB2E9X_KOIVBXWFcbawfTotnG-iQ0wNh4", + "sha256": "34da55d92d3c4e4a7f273eb82d31fc09b106ff077ffb61c1acc68205a1df81dd", + "admitted": false + }, + "aws_cleanup": { + "transaction_id": "KZlZD7uPtO5CtS5hyDrn5qh1FmQaOZiuKSMUbSX83jU", + "sha256": "e74f5143af8bf797f3e14accf4a44121c1ccf8e5f8dd16bf55e09c6e4dd3967e" + } + }, + "confirmation": { + "status_service": "https://upload.ardrive.io/v1/tx/{transaction_id}/status", + "transactions_confirmed": 76, + "total_winc_charged": 0, + "public_cache_verified_at": "2026-08-13T21:33:10Z", + "public_gateway_verified_at": null, + "verified_downloads": { + "publication-manifest.json": "19ac497b7e55881422e28cc8a118085a48e685a8bd5c5e7ee8abe2839ef8f8bb", + "ledger/events.jsonl": "ee844391a410ec83c471939d212572b94cc25168597bee3741081335c093ae0f", + "reconstructed qwen35-evidence.tgz": "84db563708d77fb13ff93f072dec85f19e96ec7cfe33a22e56e639df78dfcc60" + } + } +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/acceptance-artifacts/install.log b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/acceptance-artifacts/install.log new file mode 100644 index 0000000..9eab37b --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/acceptance-artifacts/install.log @@ -0,0 +1,30 @@ +[load] /opt/qwen35-acceptance-v3/model + sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct + hidden 1024 | 24 layers | vocab 248320 | BF16, F32 + inferred from tensors alone: 24 layers, hidden 1024, tied=True (confidence 1.00) + tokenizer: model tokenizer + corpus: /opt/qwen35-acceptance-v3/lecore/REFERENCE.md (2303 KB) + memory: 128 registers (of 1024 dimensions) and 200 searchable passages + +[install] leCore into the weights + hardware: cpu (disabled) + prepend ok 2 layers added, drift 0.000e+00 (relative 0.000e+00, first at token 0) -- bit-identical + registers ok 128 reserved slots, 896 dims left to the model + hrnn_channel FAIL ValueError: cannot reshape array of size 65536 into shape (64,20,128) [heads k=16 v=16, kdim=128 vdim=128, hidden=1024, 26 layers -- the ladder is optional, continuing without it] + nullspace_guard ok 994 of 1024 dims safe to write (true null space) + router ok layer 0, held-out accuracy 95% + memory_index ok 200 passages in 200 unused rows, perplexity -1.776% + self_write FAIL novelty readout r=0.714, finds 36% of the top decile + state_track ok 4 of 128 registers reserved as no-decay state slots + exit_calibration ok layer 25 of 26 agrees >=100% -- 4% of the depth is free + improvement FAIL skipped to preserve the checkpoint's fixed intermediate_size for official reload + boot_record ok row 248319, perplexity -1.776% + +[wrote] /opt/qwen35-acceptance-v3/installed-checkpoint (2112.9 MB, BF16, F32) + +[verify] reloading from disk + sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct + 26 layers | perplexity 19.9834 (was 20.3446) | boots as 'leCore' + verdict BETTER (-1.776%) | repetition 0.96 -> 0.77 + +Next: assess.bat (or: python assimilation/galvatron.py /opt/qwen35-acceptance-v3/installed-checkpoint --assess out.npz) diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/acceptance-artifacts/metrics.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/acceptance-artifacts/metrics.json new file mode 100644 index 0000000..48ae689 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/acceptance-artifacts/metrics.json @@ -0,0 +1,70 @@ +{ + "diagnostics": { + "official_smokes": [ + "AutoProcessor: ImportError: \nQwen3VLVideoProcessor requires the Torchvision library but it was not found in your environment. Check out the instructions on the\ninstallation page: https://pytorch.org/get-started/locally/ and follow the ones that match your environment.\nPlease note that you may need to restart your runtime after installation.\n" + ] + }, + "metrics": { + "acceptance_pass": 0.0, + "emitted_checkpoint_mb": 2112.911655, + "eval_tokens": 4096.0, + "experimental_installer_used": 1.0, + "installed_perplexity": 10.357574708230848, + "official_reload_pass": 1.0, + "original_perplexity": 7.9062244629535465, + "paired_block_length": 4.0, + "paired_ci_hi_nats": 0.34246740100294987, + "paired_ci_lo_nats": 0.20259645914578694, + "paired_effective_tokens": 2279.0, + "peak_gpu_mb": 0.0, + "peak_rss_mb": 24306.52734375, + "perplexity_delta_pct": 31.005320640258482, + "reference_logit_parity_pass": 1.0, + "reference_logit_relative_error": 6.787656774900957e-07, + "reload_pass": 1.0, + "source_clean": 1.0, + "spectral_filtering_enabled": 0.0, + "statistical_gate_pass": 0.0, + "text_generation_pass": 0.0, + "tokenizer_parity_pass": 1.0, + "vision_smoke_pass": 0.0 + }, + "source": { + "artifacts": [ + { + "path": "experiments/qwen35_acceptance/run.py", + "sha256": "d1c3ecec0fe19a9aea1c30354032919ad25b15092ffd547a3f487dcb1a10f849" + }, + { + "path": "experiments/qwen35_acceptance/contract.py", + "sha256": "3710f24df07431b2be710ee291f6369aaae48c369bbab32e3534cfa5d31a501b" + }, + { + "path": "assimilation/install.py", + "sha256": "a980357333b0820230ec5deba94ce3cb95260a17da95c786997a7bc527a8ddce" + }, + { + "path": "holographic/io_and_interop/holographic_measure.py", + "sha256": "95ceb45ca9e57962cb1e81e6a335f8ff2d46f7f6b7a128a62b8bdd2682a15527" + }, + { + "path": "holographic/io_and_interop/holographic_gdnruntime.py", + "sha256": "82122011e226ee8eee0ce2ffeec946dcd60f4f95c622c627e6cb7522d82a7928" + } + ], + "commit": "f399cd6bc7f24383a792fdcf32c43a4e1197d4f1", + "inputs": { + "evaluation_corpus": { + "bytes": 1213410, + "path": "/opt/qwen35-acceptance-v3/inputs/federalist-papers.txt", + "sha256": "a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94" + }, + "installation_corpus": { + "bytes": 2303549, + "path": "/opt/qwen35-acceptance-v3/lecore/REFERENCE.md", + "sha256": "59f551d452cfb0664deeb6dfbe5b98598abb1633cc38a5bb020ec100ac3cd33a" + } + }, + "repository": "https://github.com/atimics/holostuff.git" + } +} \ No newline at end of file diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/admission-preflight.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/admission-preflight.json new file mode 100644 index 0000000..5d54a3e --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/admission-preflight.json @@ -0,0 +1,24 @@ +{ + "checks": { + "checker_version": true, + "complete_top_level_model_manifest": true, + "corpora_distinct": true, + "evaluation_corpus": true, + "experiment_version": true, + "installation_corpus": true, + "launch_attempt": true, + "launch_corpora": true, + "launch_method": true, + "launch_model": true, + "launch_source": true, + "minimum_tokens": true, + "official_reload_metric": true, + "source_commit": true, + "spectral_disabled": true, + "timeout": true, + "tokenizer": true, + "venv_interpreter": true, + "weights": true + }, + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/aws-cleanup-receipt.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/aws-cleanup-receipt.json new file mode 100644 index 0000000..c694b25 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/aws-cleanup-receipt.json @@ -0,0 +1,16 @@ +{ + "schema": "lecore.qwen35-aws-cleanup-receipt.v1", + "checked_at": "2026-08-13T21:27:39Z", + "region": "us-east-1", + "instance": { + "id": "i-0f57dba006efd2f32", + "state": "terminated" + }, + "security_group": { + "id": "sg-08e840b4749a47406", + "network_interfaces_before_delete": 0, + "delete_return": true, + "post_delete_lookup": "InvalidGroup.NotFound" + }, + "retry_launched": false +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/aws-instance-identity.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/aws-instance-identity.json new file mode 100644 index 0000000..c0eff51 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/aws-instance-identity.json @@ -0,0 +1,17 @@ +{ + "accountId" : "022118847419", + "architecture" : "x86_64", + "availabilityZone" : "us-east-1a", + "billingProducts" : null, + "devpayProductCodes" : null, + "marketplaceProductCodes" : null, + "imageId" : "ami-07a5b367e8dc8bd92", + "instanceId" : "i-0f57dba006efd2f32", + "instanceType" : "r7i.4xlarge", + "kernelId" : null, + "pendingTime" : "2026-08-13T18:50:56Z", + "privateIp" : "172.31.9.226", + "ramdiskId" : null, + "region" : "us-east-1", + "version" : "2017-09-30" +} \ No newline at end of file diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/cloud-init-output.log b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/cloud-init-output.log new file mode 100644 index 0000000..edafb54 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/cloud-init-output.log @@ -0,0 +1,525 @@ +Cloud-init v. 22.2.2 running 'init' at Thu, 13 Aug 2026 18:51:06 +0000. Up 6.49 seconds. +ci-info: ++++++++++++++++++++++++++++++++++++++Net device info+++++++++++++++++++++++++++++++++++++++ +ci-info: +---------+------+----------------------------+---------------+--------+-------------------+ +ci-info: | Device | Up | Address | Mask | Scope | Hw-Address | +ci-info: +---------+------+----------------------------+---------------+--------+-------------------+ +ci-info: | enp39s0 | True | 172.31.9.226 | 255.255.240.0 | global | 02:b8:9e:ec:18:a7 | +ci-info: | enp39s0 | True | fe80::b8:9eff:feec:18a7/64 | . | link | 02:b8:9e:ec:18:a7 | +ci-info: | lo | True | 127.0.0.1 | 255.0.0.0 | host | . | +ci-info: | lo | True | ::1/128 | . | host | . | +ci-info: +---------+------+----------------------------+---------------+--------+-------------------+ +ci-info: +++++++++++++++++++++++++++++Route IPv4 info++++++++++++++++++++++++++++++ +ci-info: +-------+-------------+------------+-----------------+-----------+-------+ +ci-info: | Route | Destination | Gateway | Genmask | Interface | Flags | +ci-info: +-------+-------------+------------+-----------------+-----------+-------+ +ci-info: | 0 | 0.0.0.0 | 172.31.0.1 | 0.0.0.0 | enp39s0 | UG | +ci-info: | 1 | 172.31.0.0 | 0.0.0.0 | 255.255.240.0 | enp39s0 | U | +ci-info: | 2 | 172.31.0.1 | 0.0.0.0 | 255.255.255.255 | enp39s0 | UH | +ci-info: | 3 | 172.31.0.2 | 0.0.0.0 | 255.255.255.255 | enp39s0 | UH | +ci-info: +-------+-------------+------------+-----------------+-----------+-------+ +ci-info: +++++++++++++++++++Route IPv6 info+++++++++++++++++++ +ci-info: +-------+-------------+---------+-----------+-------+ +ci-info: | Route | Destination | Gateway | Interface | Flags | +ci-info: +-------+-------------+---------+-----------+-------+ +ci-info: | 0 | fe80::/64 | :: | enp39s0 | U | +ci-info: | 2 | local | :: | enp39s0 | U | +ci-info: | 3 | multicast | :: | enp39s0 | U | +ci-info: +-------+-------------+---------+-----------+-------+ +Generating public/private ed25519 key pair. +Your identification has been saved in /etc/ssh/ssh_host_ed25519_key +Your public key has been saved in /etc/ssh/ssh_host_ed25519_key.pub +The key fingerprint is: +SHA256:MZkd1yoofGHy/ZOtTvRNMhEkVJyPrAVpZDm2z9zlerM root@ip-172-31-9-226.ec2.internal +The key's randomart image is: ++--[ED25519 256]--+ +| .o*B+. | +| . o+ +B.+. | +| . +=+.o *.o | +| o +oo o +.o| +| oS o.Oooo| +| .=.==o| +| .o...| +| .. ...| +| .. Eo| ++----[SHA256]-----+ +Generating public/private ecdsa key pair. +Your identification has been saved in /etc/ssh/ssh_host_ecdsa_key +Your public key has been saved in /etc/ssh/ssh_host_ecdsa_key.pub +The key fingerprint is: +SHA256:kVs/s/5s+vfOCVmNCXFs8dCjc2AgkWs78sZDuNj+Qa0 root@ip-172-31-9-226.ec2.internal +The key's randomart image is: ++---[ECDSA 256]---+ +| oo.o.+o | +| o. =o+.| +| o o o.o o| +| *.. + +.| +| S...+ =..| +| o.+. +o | +| o *E. .o | +| . o =.. .o.o| +| ..o.. o=++=| ++----[SHA256]-----+ +Cloud-init v. 22.2.2 running 'modules:config' at Thu, 13 Aug 2026 18:51:08 +0000. Up 7.84 seconds. +Cloud-init v. 22.2.2 running 'modules:final' at Thu, 13 Aug 2026 18:51:08 +0000. Up 8.36 seconds. +Shutdown scheduled for Fri 2026-08-14 02:51:08 UTC, use 'shutdown -c' to cancel. +Amazon Linux 2023 repository 80 MB/s | 75 MB 00:00 +Amazon Linux 2023 Kernel Livepatch repository 572 kB/s | 69 kB 00:00 +Package tar-2:1.34-1.amzn2023.0.4.x86_64 is already installed. +Package gzip-1.12-1.amzn2023.0.1.x86_64 is already installed. +Dependencies resolved. +================================================================================ + Package Arch Version Repository Size +================================================================================ +Installing: + cargo x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 8.2 M + git x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 53 k + python3.12 x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 27 k + python3.12-pip noarch 23.2.1-4.amzn2023.0.10 amazonlinux 2.8 M + rust x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 62 M +Installing dependencies: + annobin-docs noarch 12.69-1.amzn2023.0.1 amazonlinux 93 k + annobin-plugin-gcc x86_64 12.69-1.amzn2023.0.1 amazonlinux 974 k + cpp x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 11 M + gc x86_64 8.0.4-5.amzn2023.0.2 amazonlinux 105 k + gcc x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 36 M + gcc-plugin-annobin x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 38 k + git-core x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 4.9 M + git-core-doc noarch 2.50.1-1.amzn2023.0.1 amazonlinux 2.8 M + glibc-devel x86_64 2.34-231.amzn2023.0.5 amazonlinux 33 k + glibc-headers-x86 noarch 2.34-231.amzn2023.0.5 amazonlinux 439 k + guile22 x86_64 2.2.7-2.amzn2023.0.3 amazonlinux 6.4 M + kernel-headers x86_64 1:6.1.177-224.371.amzn2023 amazonlinux 1.4 M + libmpc x86_64 1.2.1-2.amzn2023.0.2 amazonlinux 62 k + libtool-ltdl x86_64 2.4.7-1.amzn2023.0.3 amazonlinux 38 k + libxcrypt-devel x86_64 4.4.33-7.amzn2023 amazonlinux 32 k + make x86_64 1:4.3-5.amzn2023.0.2 amazonlinux 534 k + mpdecimal x86_64 2.5.1-3.amzn2023.0.3 amazonlinux 101 k + perl-Error noarch 1:0.17030-2.amzn2023.0.1 amazonlinux 42 k + perl-File-Find noarch 1.37-477.amzn2023.0.9 amazonlinux 26 k + perl-Git noarch 2.50.1-1.amzn2023.0.1 amazonlinux 41 k + perl-TermReadKey x86_64 2.38-9.amzn2023.0.3 amazonlinux 36 k + perl-lib x86_64 0.65-477.amzn2023.0.9 amazonlinux 15 k + python3.12-libs x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 9.2 M + python3.12-pip-wheel noarch 23.2.1-4.amzn2023.0.10 amazonlinux 1.5 M + rust-std-static x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 39 M +Installing weak dependencies: + python3.12-setuptools noarch 68.2.2-4.amzn2023.0.3 amazonlinux 1.4 M + +Transaction Summary +================================================================================ +Install 31 Packages + +Total download size: 189 M +Installed size: 700 M +Downloading Packages: +(1/31): annobin-plugin-gcc-12.69-1.amzn2023.0.1 23 MB/s | 974 kB 00:00 +(2/31): annobin-docs-12.69-1.amzn2023.0.1.noarc 2.0 MB/s | 93 kB 00:00 +(3/31): gc-8.0.4-5.amzn2023.0.2.x86_64.rpm 3.4 MB/s | 105 kB 00:00 +(4/31): cpp-11.5.0-5.amzn2023.0.5.x86_64.rpm 77 MB/s | 11 MB 00:00 +(5/31): gcc-plugin-annobin-11.5.0-5.amzn2023.0. 1.3 MB/s | 38 kB 00:00 +(6/31): git-2.50.1-1.amzn2023.0.1.x86_64.rpm 2.0 MB/s | 53 kB 00:00 +(7/31): cargo-1.97.0-1.amzn2023.0.1.x86_64.rpm 28 MB/s | 8.2 MB 00:00 +(8/31): git-core-2.50.1-1.amzn2023.0.1.x86_64.r 67 MB/s | 4.9 MB 00:00 +(9/31): git-core-doc-2.50.1-1.amzn2023.0.1.noar 70 MB/s | 2.8 MB 00:00 +(10/31): glibc-devel-2.34-231.amzn2023.0.5.x86_ 1.2 MB/s | 33 kB 00:00 +(11/31): glibc-headers-x86-2.34-231.amzn2023.0. 11 MB/s | 439 kB 00:00 +(12/31): kernel-headers-6.1.177-224.371.amzn202 37 MB/s | 1.4 MB 00:00 +(13/31): guile22-2.2.7-2.amzn2023.0.3.x86_64.rp 73 MB/s | 6.4 MB 00:00 +(14/31): libmpc-1.2.1-2.amzn2023.0.2.x86_64.rpm 2.0 MB/s | 62 kB 00:00 +(15/31): libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_ 1.6 MB/s | 38 kB 00:00 +(16/31): gcc-11.5.0-5.amzn2023.0.5.x86_64.rpm 77 MB/s | 36 MB 00:00 +(17/31): libxcrypt-devel-4.4.33-7.amzn2023.x86_ 300 kB/s | 32 kB 00:00 +(18/31): make-4.3-5.amzn2023.0.2.x86_64.rpm 5.2 MB/s | 534 kB 00:00 +(19/31): perl-Error-0.17030-2.amzn2023.0.1.noar 1.6 MB/s | 42 kB 00:00 +(20/31): mpdecimal-2.5.1-3.amzn2023.0.3.x86_64. 3.4 MB/s | 101 kB 00:00 +(21/31): perl-File-Find-1.37-477.amzn2023.0.9.n 970 kB/s | 26 kB 00:00 +(22/31): perl-Git-2.50.1-1.amzn2023.0.1.noarch. 1.5 MB/s | 41 kB 00:00 +(23/31): perl-TermReadKey-2.38-9.amzn2023.0.3.x 1.3 MB/s | 36 kB 00:00 +(24/31): perl-lib-0.65-477.amzn2023.0.9.x86_64. 535 kB/s | 15 kB 00:00 +(25/31): python3.12-3.12.13-2.amzn2023.0.5.x86_ 1.2 MB/s | 27 kB 00:00 +(26/31): python3.12-pip-23.2.1-4.amzn2023.0.10. 67 MB/s | 2.8 MB 00:00 +(27/31): python3.12-pip-wheel-23.2.1-4.amzn2023 42 MB/s | 1.5 MB 00:00 +(28/31): python3.12-setuptools-68.2.2-4.amzn202 50 MB/s | 1.4 MB 00:00 +(29/31): python3.12-libs-3.12.13-2.amzn2023.0.5 77 MB/s | 9.2 MB 00:00 +(30/31): rust-std-static-1.97.0-1.amzn2023.0.1. 70 MB/s | 39 MB 00:00 +(31/31): rust-1.97.0-1.amzn2023.0.1.x86_64.rpm 78 MB/s | 62 MB 00:00 +-------------------------------------------------------------------------------- +Total 127 MB/s | 189 MB 00:01 +Running transaction check +Transaction check succeeded. +Running transaction test +Transaction test succeeded. +Running transaction + Preparing : 1/1 + Installing : libmpc-1.2.1-2.amzn2023.0.2.x86_64 1/31 + Installing : git-core-2.50.1-1.amzn2023.0.1.x86_64 2/31 + Installing : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 3/31 + Installing : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Installing : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 5/31 + Installing : perl-lib-0.65-477.amzn2023.0.9.x86_64 6/31 + Installing : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 7/31 + Installing : perl-File-Find-1.37-477.amzn2023.0.9.noarch 8/31 + Installing : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 9/31 + Installing : perl-Git-2.50.1-1.amzn2023.0.1.noarch 10/31 + Installing : git-2.50.1-1.amzn2023.0.1.x86_64 11/31 + Installing : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 12/31 + Installing : python3.12-3.12.13-2.amzn2023.0.5.x86_64 13/31 + Installing : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 14/31 + Installing : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 15/31 + Installing : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Installing : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 17/31 + Installing : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 18/31 + Installing : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 19/31 + Installing : glibc-devel-2.34-231.amzn2023.0.5.x86_64 20/31 + Installing : gc-8.0.4-5.amzn2023.0.2.x86_64 21/31 + Installing : guile22-2.2.7-2.amzn2023.0.3.x86_64 22/31 + Installing : make-1:4.3-5.amzn2023.0.2.x86_64 23/31 + Installing : gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Running scriptlet: gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Installing : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 25/31 + Installing : rust-1.97.0-1.amzn2023.0.1.x86_64 26/31 + Installing : annobin-docs-12.69-1.amzn2023.0.1.noarch 27/31 + Installing : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Running scriptlet: annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Installing : cargo-1.97.0-1.amzn2023.0.1.x86_64 29/31 + Installing : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 30/31 + Installing : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Running scriptlet: python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Verifying : annobin-docs-12.69-1.amzn2023.0.1.noarch 1/31 + Verifying : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 2/31 + Verifying : cargo-1.97.0-1.amzn2023.0.1.x86_64 3/31 + Verifying : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Verifying : gc-8.0.4-5.amzn2023.0.2.x86_64 5/31 + Verifying : gcc-11.5.0-5.amzn2023.0.5.x86_64 6/31 + Verifying : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 7/31 + Verifying : git-2.50.1-1.amzn2023.0.1.x86_64 8/31 + Verifying : git-core-2.50.1-1.amzn2023.0.1.x86_64 9/31 + Verifying : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 10/31 + Verifying : glibc-devel-2.34-231.amzn2023.0.5.x86_64 11/31 + Verifying : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 12/31 + Verifying : guile22-2.2.7-2.amzn2023.0.3.x86_64 13/31 + Verifying : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 14/31 + Verifying : libmpc-1.2.1-2.amzn2023.0.2.x86_64 15/31 + Verifying : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Verifying : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 17/31 + Verifying : make-1:4.3-5.amzn2023.0.2.x86_64 18/31 + Verifying : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 19/31 + Verifying : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 20/31 + Verifying : perl-File-Find-1.37-477.amzn2023.0.9.noarch 21/31 + Verifying : perl-Git-2.50.1-1.amzn2023.0.1.noarch 22/31 + Verifying : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 23/31 + Verifying : perl-lib-0.65-477.amzn2023.0.9.x86_64 24/31 + Verifying : python3.12-3.12.13-2.amzn2023.0.5.x86_64 25/31 + Verifying : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 26/31 + Verifying : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 27/31 + Verifying : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 28/31 + Verifying : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 29/31 + Verifying : rust-1.97.0-1.amzn2023.0.1.x86_64 30/31 + Verifying : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 31/31 + +Installed: + annobin-docs-12.69-1.amzn2023.0.1.noarch + annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 + cargo-1.97.0-1.amzn2023.0.1.x86_64 + cpp-11.5.0-5.amzn2023.0.5.x86_64 + gc-8.0.4-5.amzn2023.0.2.x86_64 + gcc-11.5.0-5.amzn2023.0.5.x86_64 + gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 + git-2.50.1-1.amzn2023.0.1.x86_64 + git-core-2.50.1-1.amzn2023.0.1.x86_64 + git-core-doc-2.50.1-1.amzn2023.0.1.noarch + glibc-devel-2.34-231.amzn2023.0.5.x86_64 + glibc-headers-x86-2.34-231.amzn2023.0.5.noarch + guile22-2.2.7-2.amzn2023.0.3.x86_64 + kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 + libmpc-1.2.1-2.amzn2023.0.2.x86_64 + libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 + libxcrypt-devel-4.4.33-7.amzn2023.x86_64 + make-1:4.3-5.amzn2023.0.2.x86_64 + mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 + perl-Error-1:0.17030-2.amzn2023.0.1.noarch + perl-File-Find-1.37-477.amzn2023.0.9.noarch + perl-Git-2.50.1-1.amzn2023.0.1.noarch + perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 + perl-lib-0.65-477.amzn2023.0.9.x86_64 + python3.12-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-pip-23.2.1-4.amzn2023.0.10.noarch + python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch + python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch + rust-1.97.0-1.amzn2023.0.1.x86_64 + rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 + +Complete! +Cloning into '/opt/qwen35-acceptance-v3/lecore'... +HEAD is now at f399cd6 fix(qwen): emit official-compatible installed checkpoints +/opt/qwen35-acceptance-v3/lecore/REFERENCE.md: OK +/opt/qwen35-acceptance-v3/inputs/federalist-papers.txt: OK +Cloning into '/opt/qwen35-acceptance-v3/ilxyr'... +HEAD is now at e92382f Publish EXP-006 experiment evidence + Updating crates.io index + Downloading crates ... + Downloaded digest v0.10.7 + Downloaded ed25519 v2.2.3 + Downloaded block-buffer v0.10.4 + Downloaded serde_json v1.0.150 + Downloaded version_check v0.9.5 + Downloaded cfg-if v1.0.4 + Downloaded rand_core v0.6.4 + Downloaded pkcs8 v0.10.2 + Downloaded crypto-common v0.1.7 + Downloaded generic-array v0.14.7 + Downloaded sha2 v0.10.9 + Downloaded signature v2.2.0 + Downloaded semver v1.0.28 + Downloaded cpufeatures v0.2.17 + Downloaded itoa v1.0.18 + Downloaded curve25519-dalek-derive v0.1.1 + Downloaded proc-macro2 v1.0.106 + Downloaded serde_core v1.0.229 + Downloaded serde v1.0.229 + Downloaded getrandom v0.2.17 + Downloaded rustc_version v0.4.1 + Downloaded serde_derive v1.0.229 + Downloaded subtle v2.6.1 + Downloaded const-oid v0.9.6 + Downloaded spki v0.7.3 + Downloaded base64 v0.22.1 + Downloaded base64ct v1.8.3 + Downloaded quote v1.0.46 + Downloaded zmij v1.0.23 + Downloaded ed25519-dalek v2.2.0 + Downloaded unicode-ident v1.0.24 + Downloaded zeroize v1.9.0 + Downloaded typenum v1.20.1 + Downloaded der v0.7.10 + Downloaded memchr v2.8.3 + Downloaded libc v0.2.186 + Downloaded syn v2.0.119 + Downloaded syn v3.0.0 + Downloaded curve25519-dalek v4.1.3 + Compiling proc-macro2 v1.0.106 + Compiling version_check v0.9.5 + Compiling unicode-ident v1.0.24 + Compiling quote v1.0.46 + Compiling typenum v1.20.1 + Compiling semver v1.0.28 + Compiling serde_core v1.0.229 + Compiling cpufeatures v0.2.17 + Compiling cfg-if v1.0.4 + Compiling zmij v1.0.23 + Compiling serde v1.0.229 + Compiling signature v2.2.0 + Compiling zeroize v1.9.0 + Compiling subtle v2.6.1 + Compiling serde_json v1.0.150 + Compiling itoa v1.0.18 + Compiling memchr v2.8.3 + Compiling base64 v0.22.1 + Compiling ed25519 v2.2.3 + Compiling rustc_version v0.4.1 + Compiling generic-array v0.14.7 + Compiling curve25519-dalek v4.1.3 + Compiling syn v2.0.119 + Compiling syn v3.0.0 + Compiling crypto-common v0.1.7 + Compiling block-buffer v0.10.4 + Compiling digest v0.10.7 + Compiling sha2 v0.10.9 + Compiling serde_derive v1.0.229 + Compiling curve25519-dalek-derive v0.1.1 + Compiling ed25519-dalek v2.2.0 + Compiling ilxyr-core v0.1.0 (/opt/qwen35-acceptance-v3/ilxyr/crates/ilxyr-core) + Compiling ilxyr-cli v0.1.0 (/opt/qwen35-acceptance-v3/ilxyr/crates/ilxyr-cli) + Finished `release` profile [optimized] target(s) in 20.61s +Looking in indexes: https://download.pytorch.org/whl/cpu +Collecting torch==2.11.0 + Obtaining dependency information for torch==2.11.0 from https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata + Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata (29 kB) +Collecting filelock (from torch==2.11.0) + Obtaining dependency information for filelock from https://files.pythonhosted.org/packages/81/47/dd9a212ef6e343a6857485ffe25bba537304f1913bdbed446a23f7f592e1/filelock-3.29.0-py3-none-any.whl.metadata + Downloading filelock-3.29.0-py3-none-any.whl.metadata (2.0 kB) +Collecting typing-extensions>=4.10.0 (from torch==2.11.0) + Obtaining dependency information for typing-extensions>=4.10.0 from https://files.pythonhosted.org/packages/18/67/36e9267722cc04a6b9f15c7f3441c2363321a3ea07da7ae0c0707beb2a9c/typing_extensions-4.15.0-py3-none-any.whl.metadata + Downloading typing_extensions-4.15.0-py3-none-any.whl.metadata (3.3 kB) +Collecting setuptools<82 (from torch==2.11.0) + Obtaining dependency information for setuptools<82 from https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata (6.6 kB) +Collecting sympy>=1.13.3 (from torch==2.11.0) + Obtaining dependency information for sympy>=1.13.3 from https://files.pythonhosted.org/packages/a2/09/77d55d46fd61b4a135c444fc97158ef34a095e5681d0a6c10b75bf356191/sympy-1.14.0-py3-none-any.whl.metadata + Downloading sympy-1.14.0-py3-none-any.whl.metadata (12 kB) +Collecting networkx>=2.5.1 (from torch==2.11.0) + Obtaining dependency information for networkx>=2.5.1 from https://files.pythonhosted.org/packages/9e/c9/b2622292ea83fbb4ec318f5b9ab867d0a28ab43c5717bb85b0a5f6b3b0a4/networkx-3.6.1-py3-none-any.whl.metadata + Downloading networkx-3.6.1-py3-none-any.whl.metadata (6.8 kB) +Collecting jinja2 (from torch==2.11.0) + Obtaining dependency information for jinja2 from https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata (2.9 kB) +Collecting fsspec>=0.8.5 (from torch==2.11.0) + Obtaining dependency information for fsspec>=0.8.5 from https://files.pythonhosted.org/packages/d5/0c/043d5e551459da400957a1395e0febbf771446ff34291afcbe3d8be2a279/fsspec-2026.4.0-py3-none-any.whl.metadata + Downloading fsspec-2026.4.0-py3-none-any.whl.metadata (10 kB) +Collecting mpmath<1.4,>=1.1.0 (from sympy>=1.13.3->torch==2.11.0) + Obtaining dependency information for mpmath<1.4,>=1.1.0 from https://files.pythonhosted.org/packages/43/e3/7d92a15f894aa0c9c4b49b8ee9ac9850d6e63b03c9c32c0367a13ae62209/mpmath-1.3.0-py3-none-any.whl.metadata + Downloading mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) +Collecting MarkupSafe>=2.0 (from jinja2->torch==2.11.0) + Obtaining dependency information for MarkupSafe>=2.0 from https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.7 kB) +Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl (190.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 190.3/190.3 MB 13.7 MB/s eta 0:00:00 +Downloading fsspec-2026.4.0-py3-none-any.whl (203 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 203.4/203.4 kB 20.7 MB/s eta 0:00:00 +Downloading networkx-3.6.1-py3-none-any.whl (2.1 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 2.1/2.1 MB 122.0 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl (1.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.3/1.3 MB 158.8 MB/s eta 0:00:00 +Downloading sympy-1.14.0-py3-none-any.whl (6.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.3/6.3 MB 196.7 MB/s eta 0:00:00 +Downloading typing_extensions-4.15.0-py3-none-any.whl (44 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 44.6/44.6 kB 16.8 MB/s eta 0:00:00 +Downloading filelock-3.29.0-py3-none-any.whl (39 kB) +Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl (134 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 134.9/134.9 kB 38.3 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (22 kB) +Downloading mpmath-1.3.0-py3-none-any.whl (536 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 536.2/536.2 kB 123.6 MB/s eta 0:00:00 +Installing collected packages: mpmath, typing-extensions, sympy, setuptools, networkx, MarkupSafe, fsspec, filelock, jinja2, torch +Successfully installed MarkupSafe-3.0.3 filelock-3.29.0 fsspec-2026.4.0 jinja2-3.1.6 mpmath-1.3.0 networkx-3.6.1 setuptools-78.1.0 sympy-1.14.0 torch-2.11.0+cpu typing-extensions-4.15.0 +Collecting numpy==2.4.6 + Obtaining dependency information for numpy==2.4.6 from https://files.pythonhosted.org/packages/99/60/14115e6364fa676c5397c2ad3004e527e9aa487abf5d0706ec81bbd08529/numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (6.6 kB) +Collecting pillow==12.3.0 + Obtaining dependency information for pillow==12.3.0 from https://files.pythonhosted.org/packages/84/21/a35af28dcc61f37ed850a2d64c65c701321dfbf25085e469d5559360cbbf/pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (9.1 kB) +Collecting safetensors==0.8.0 + Obtaining dependency information for safetensors==0.8.0 from https://files.pythonhosted.org/packages/28/50/f203ff3a3ddfe19308efc83c5a3a29ed02bf786732ec35e68bf9162f3365/safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (4.2 kB) +Collecting huggingface-hub==1.27.0 + Obtaining dependency information for huggingface-hub==1.27.0 from https://files.pythonhosted.org/packages/de/d8/95b735e183957c1f26d94c52977f09d466d55119cbbc1558ea4975e4c216/huggingface_hub-1.27.0-py3-none-any.whl.metadata + Downloading huggingface_hub-1.27.0-py3-none-any.whl.metadata (16 kB) +Collecting transformers==5.14.0 + Obtaining dependency information for transformers==5.14.0 from https://files.pythonhosted.org/packages/19/3b/a0a807b957c5f8f2075a8740f597a021420583bb0aa68ff6432d74a530d4/transformers-5.14.0-py3-none-any.whl.metadata + Downloading transformers-5.14.0-py3-none-any.whl.metadata (32 kB) +Collecting click<9.0.0,>=8.4.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for click<9.0.0,>=8.4.2 from https://files.pythonhosted.org/packages/fb/e2/79c688af8b210d232694e31e59da9f6ec747bae31c3f5946e4e9b98860d5/click-8.4.2-py3-none-any.whl.metadata + Downloading click-8.4.2-py3-none-any.whl.metadata (2.6 kB) +Requirement already satisfied: filelock>=3.10.0 in /opt/qwen35-acceptance-v3/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (3.29.0) +Requirement already satisfied: fsspec>=2023.5.0 in /opt/qwen35-acceptance-v3/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (2026.4.0) +Collecting hf-xet<2.0.0,>=1.5.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for hf-xet<2.0.0,>=1.5.2 from https://files.pythonhosted.org/packages/67/4e/a28359bf1c1ecf11eba22123168c138698f7cb576ac678f5a2e16cd5da08/hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata + Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (4.9 kB) +Collecting httpx<1,>=0.23.0 (from huggingface-hub==1.27.0) + Obtaining dependency information for httpx<1,>=0.23.0 from https://files.pythonhosted.org/packages/2a/39/e50c7c3a983047577ee07d2a9e53faf5a69493943ec3f6a384bdc792deb2/httpx-0.28.1-py3-none-any.whl.metadata + Downloading httpx-0.28.1-py3-none-any.whl.metadata (7.1 kB) +Collecting packaging>=20.9 (from huggingface-hub==1.27.0) + Obtaining dependency information for packaging>=20.9 from https://files.pythonhosted.org/packages/63/34/ba1c580383c9eada3711951fef0795c80b829a078d72188184bcab9dd527/packaging-26.3-py3-none-any.whl.metadata + Downloading packaging-26.3-py3-none-any.whl.metadata (3.5 kB) +Collecting pyyaml>=5.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for pyyaml>=5.1 from https://files.pythonhosted.org/packages/8b/9d/b3589d3877982d4f2329302ef98a8026e7f4443c765c46cfecc8858c6b4b/pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.4 kB) +Collecting tqdm>=4.42.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for tqdm>=4.42.1 from https://files.pythonhosted.org/packages/f9/1c/01bfd571a64e7f270e6bab5e33777debe0edc56759233ce84f27dec92d14/tqdm-4.70.0-py3-none-any.whl.metadata + Downloading tqdm-4.70.0-py3-none-any.whl.metadata (57 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 57.3/57.3 kB 17.0 MB/s eta 0:00:00 +Requirement already satisfied: typing-extensions>=4.1.0 in /opt/qwen35-acceptance-v3/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (4.15.0) +Collecting regex>=2025.10.22 (from transformers==5.14.0) + Obtaining dependency information for regex>=2025.10.22 from https://files.pythonhosted.org/packages/2a/8e/096d00c7c480ef2ff4265349b14e2261d4ab787ba1f74e2e80d1c58079c3/regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (40 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 40.5/40.5 kB 12.2 MB/s eta 0:00:00 +Collecting tokenizers<=0.23.0,>=0.22.0 (from transformers==5.14.0) + Obtaining dependency information for tokenizers<=0.23.0,>=0.22.0 from https://files.pythonhosted.org/packages/2e/76/932be4b50ef6ccedf9d3c6639b056a967a86258c6d9200643f01269211ca/tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (7.3 kB) +Collecting typer (from transformers==5.14.0) + Obtaining dependency information for typer from https://files.pythonhosted.org/packages/43/89/9518bc0c3929bee36b3a4a8e3daddd6e03f92f9961c66d4983b837160543/typer-0.27.1-py3-none-any.whl.metadata + Downloading typer-0.27.1-py3-none-any.whl.metadata (16 kB) +Collecting anyio (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for anyio from https://files.pythonhosted.org/packages/da/35/f2287558c17e29fafc8ef3daf819bb9834061cfa43bff8014f7df7f63bdc/anyio-4.14.2-py3-none-any.whl.metadata + Downloading anyio-4.14.2-py3-none-any.whl.metadata (4.6 kB) +Collecting certifi (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for certifi from https://files.pythonhosted.org/packages/0b/a7/71ac2cff56fec219ed242bb11b8efb69fcc4bec75db06fb7bfe35de520e6/certifi-2026.7.22-py3-none-any.whl.metadata + Downloading certifi-2026.7.22-py3-none-any.whl.metadata (2.5 kB) +Collecting httpcore==1.* (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for httpcore==1.* from https://files.pythonhosted.org/packages/7e/f5/f66802a942d491edb555dd61e3a9961140fd64c90bce1eafd741609d334d/httpcore-1.0.9-py3-none-any.whl.metadata + Downloading httpcore-1.0.9-py3-none-any.whl.metadata (21 kB) +Collecting idna (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for idna from https://files.pythonhosted.org/packages/1e/5e/d4e9f1a599fb8e573b7b87160658329fbf28d19eac2718f51fc3def3aa5a/idna-3.18-py3-none-any.whl.metadata + Downloading idna-3.18-py3-none-any.whl.metadata (6.1 kB) +Collecting h11>=0.16 (from httpcore==1.*->httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for h11>=0.16 from https://files.pythonhosted.org/packages/04/4b/29cac41a4d98d144bf5f6d33995617b185d14b22401f75ca86f384e87ff1/h11-0.16.0-py3-none-any.whl.metadata + Downloading h11-0.16.0-py3-none-any.whl.metadata (8.3 kB) +Collecting shellingham>=1.3.0 (from typer->transformers==5.14.0) + Obtaining dependency information for shellingham>=1.3.0 from https://files.pythonhosted.org/packages/e0/f9/0595336914c5619e5f28a1fb793285925a8cd4b432c9da0a987836c7f822/shellingham-1.5.4-py2.py3-none-any.whl.metadata + Downloading shellingham-1.5.4-py2.py3-none-any.whl.metadata (3.5 kB) +Collecting rich>=13.8.0 (from typer->transformers==5.14.0) + Obtaining dependency information for rich>=13.8.0 from https://files.pythonhosted.org/packages/82/3b/64d4899d73f91ba49a8c18a8ff3f0ea8f1c1d75481760df8c68ef5235bf5/rich-15.0.0-py3-none-any.whl.metadata + Downloading rich-15.0.0-py3-none-any.whl.metadata (18 kB) +Collecting annotated-doc>=0.0.2 (from typer->transformers==5.14.0) + Obtaining dependency information for annotated-doc>=0.0.2 from https://files.pythonhosted.org/packages/3e/30/e900b21425a860e195f32e37657aa1f7c7f2b1bfb26f03ca209b90933c06/annotated_doc-0.0.5-py3-none-any.whl.metadata + Downloading annotated_doc-0.0.5-py3-none-any.whl.metadata (6.5 kB) +Collecting markdown-it-py>=2.2.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for markdown-it-py>=2.2.0 from https://files.pythonhosted.org/packages/b3/81/4da04ced5a082363ecfa159c010d200ecbd959ae410c10c0264a38cac0f5/markdown_it_py-4.2.0-py3-none-any.whl.metadata + Downloading markdown_it_py-4.2.0-py3-none-any.whl.metadata (7.4 kB) +Collecting pygments<3.0.0,>=2.13.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for pygments<3.0.0,>=2.13.0 from https://files.pythonhosted.org/packages/f4/7e/a72dd26f3b0f4f2bf1dd8923c85f7ceb43172af56d63c7383eb62b332364/pygments-2.20.0-py3-none-any.whl.metadata + Downloading pygments-2.20.0-py3-none-any.whl.metadata (2.5 kB) +Collecting mdurl~=0.1 (from markdown-it-py>=2.2.0->rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for mdurl~=0.1 from https://files.pythonhosted.org/packages/b3/38/89ba8ad64ae25be8de66a6d463314cf1eb366222074cfda9ee839c56a4b4/mdurl-0.1.2-py3-none-any.whl.metadata + Downloading mdurl-0.1.2-py3-none-any.whl.metadata (1.6 kB) +Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (16.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 16.6/16.6 MB 137.8 MB/s eta 0:00:00 +Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.9 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.9/6.9 MB 230.9 MB/s eta 0:00:00 +Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (516 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 516.0/516.0 kB 107.2 MB/s eta 0:00:00 +Downloading huggingface_hub-1.27.0-py3-none-any.whl (784 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 784.9/784.9 kB 158.0 MB/s eta 0:00:00 +Downloading transformers-5.14.0-py3-none-any.whl (11.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 11.6/11.6 MB 115.3 MB/s eta 0:00:00 +Downloading click-8.4.2-py3-none-any.whl (119 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 119.2/119.2 kB 26.5 MB/s eta 0:00:00 +Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (4.5 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 4.5/4.5 MB 187.6 MB/s eta 0:00:00 +Downloading httpx-0.28.1-py3-none-any.whl (73 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 73.5/73.5 kB 29.5 MB/s eta 0:00:00 +Downloading httpcore-1.0.9-py3-none-any.whl (78 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 78.8/78.8 kB 31.0 MB/s eta 0:00:00 +Downloading packaging-26.3-py3-none-any.whl (129 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 130.0/130.0 kB 52.6 MB/s eta 0:00:00 +Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (807 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 807.9/807.9 kB 161.6 MB/s eta 0:00:00 +Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (801 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 801.8/801.8 kB 133.2 MB/s eta 0:00:00 +Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (3.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 3.3/3.3 MB 161.6 MB/s eta 0:00:00 +Downloading tqdm-4.70.0-py3-none-any.whl (80 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 80.2/80.2 kB 29.1 MB/s eta 0:00:00 +Downloading typer-0.27.1-py3-none-any.whl (122 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 122.9/122.9 kB 43.5 MB/s eta 0:00:00 +Downloading annotated_doc-0.0.5-py3-none-any.whl (5.3 kB) +Downloading rich-15.0.0-py3-none-any.whl (310 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 310.7/310.7 kB 105.0 MB/s eta 0:00:00 +Downloading shellingham-1.5.4-py2.py3-none-any.whl (9.8 kB) +Downloading anyio-4.14.2-py3-none-any.whl (125 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 125.8/125.8 kB 49.3 MB/s eta 0:00:00 +Downloading idna-3.18-py3-none-any.whl (65 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 65.5/65.5 kB 28.2 MB/s eta 0:00:00 +Downloading certifi-2026.7.22-py3-none-any.whl (136 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 137.0/137.0 kB 49.9 MB/s eta 0:00:00 +Downloading h11-0.16.0-py3-none-any.whl (37 kB) +Downloading markdown_it_py-4.2.0-py3-none-any.whl (91 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 91.7/91.7 kB 39.0 MB/s eta 0:00:00 +Downloading pygments-2.20.0-py3-none-any.whl (1.2 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.2/1.2 MB 204.7 MB/s eta 0:00:00 +Downloading mdurl-0.1.2-py3-none-any.whl (10.0 kB) +Installing collected packages: tqdm, shellingham, safetensors, regex, pyyaml, pygments, pillow, packaging, numpy, mdurl, idna, hf-xet, h11, click, certifi, annotated-doc, markdown-it-py, httpcore, anyio, rich, httpx, typer, huggingface-hub, tokenizers, transformers +Successfully installed annotated-doc-0.0.5 anyio-4.14.2 certifi-2026.7.22 click-8.4.2 h11-0.16.0 hf-xet-1.6.0 httpcore-1.0.9 httpx-0.28.1 huggingface-hub-1.27.0 idna-3.18 markdown-it-py-4.2.0 mdurl-0.1.2 numpy-2.4.6 packaging-26.3 pillow-12.3.0 pygments-2.20.0 pyyaml-6.0.3 regex-2026.7.19 rich-15.0.0 safetensors-0.8.0 shellingham-1.5.4 tokenizers-0.22.2 tqdm-4.70.0 transformers-5.14.0 typer-0.27.1 + +[notice] A new release of pip is available: 23.2.1 -> 26.2.1 +[notice] To update, run: python3.12 -m pip install --upgrade pip +No broken requirements found. +torch 2.11.0+cpu +transformers 5.14.0 +AutoProcessor Qwen3_5ForConditionalGeneration + Fetching 13 files: 0%| | 0/13 [00:00 +ilxyr contribute +ilxyr shared-task-register +ilxyr compile +ilxyr retro --execute +ilxyr forecast +ilxyr fund +ilxyr trust-key +ilxyr budget-payload +ilxyr budget-register +ilxyr allocate ... +ilxyr admit +ilxyr run --execute +ilxyr authorize +ilxyr run-auto +ilxyr loop-cycle +ilxyr sandbox +ilxyr certify +ilxyr calibration +ilxyr status +ilxyr verify diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/evidence-bundle.sha256 b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/evidence-bundle.sha256 new file mode 100644 index 0000000..9845597 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/evidence-bundle.sha256 @@ -0,0 +1 @@ +84db563708d77fb13ff93f072dec85f19e96ec7cfe33a22e56e639df78dfcc60 /opt/qwen35-acceptance-v3/qwen35-evidence.tgz diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/execution-summary.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/execution-summary.json new file mode 100644 index 0000000..744c1fc --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/execution-summary.json @@ -0,0 +1,8 @@ +{ + "completed_at": "2026-08-13T21:21:28.745030+00:00", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance", + "run_exit_code": 1, + "schema": "lecore.qwen35-execution-summary.v3", + "status_exit_code": 0, + "verify_exit_code": 0 +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/generation.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/generation.json new file mode 100644 index 0000000..ec00f2c --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/generation.json @@ -0,0 +1 @@ +{"experiment_id": "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance", "project": "/opt/qwen35-acceptance-v3/project/project.json", "project_dir": "/opt/qwen35-acceptance-v3/project"} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-admission.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-admission.json new file mode 100644 index 0000000..71d5bc8 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-admission.json @@ -0,0 +1,68 @@ +{ + "schema": "ilxyr.admission.v1", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance", + "accepted": true, + "checks": [ + { + "gate": "methodology", + "passed": true, + "detail": "4/4 required research stages resolved" + }, + { + "gate": "outcome_contract", + "passed": true, + "detail": "3 frozen mutually evaluated outcomes declared" + }, + { + "gate": "forecast_participation", + "passed": true, + "detail": "2/2 distinct forecasters" + }, + { + "gate": "forecast_stake", + "passed": true, + "detail": "10/10 forecast credits staked" + }, + { + "gate": "compute_funding", + "passed": true, + "detail": "100/100 compute credits committed" + }, + { + "gate": "executor_available", + "passed": true, + "detail": "local-command adapter is installed" + }, + { + "gate": "weight_protection", + "passed": true, + "detail": "local execution is limited to public-weight handles" + }, + { + "gate": "local_execution_policy", + "passed": true, + "detail": "local execution requires network=open and an absolute executable path" + }, + { + "gate": "code_policy", + "passed": true, + "detail": "local-command directly executes the declared program" + }, + { + "gate": "export_policy", + "passed": true, + "detail": "local-command records stdout, stderr, and metric artifacts" + }, + { + "gate": "reviewer_separation", + "passed": true, + "detail": "proposer and engineering reviewer use distinct handles" + }, + { + "gate": "forecaster_separation", + "passed": true, + "detail": "proposer and forecasters use distinct handles" + } + ], + "decided_at_ms": 1786647171244 +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt new file mode 100644 index 0000000..c29264e --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-commit.txt @@ -0,0 +1,7 @@ +commit e92382ff2a5e8714466533a160f6609b4ef9cee8 +Author: atimics +AuthorDate: Sat Aug 8 12:46:56 2026 -0700 +Commit: atimics +CommitDate: Sat Aug 8 12:46:56 2026 -0700 + + Publish EXP-006 experiment evidence diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-compile.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-compile.json new file mode 100644 index 0000000..3d7fd66 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-compile.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/c5399dba347f001bf553941b88aae7e74e0609bbded168252ae731eff6a39434" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json new file mode 100644 index 0000000..afdcd6a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-engineering-review.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/08d81fe6c3b0a1d2f6c93d08b7d5e51ef5b4044c33c2f7fc30d4a61210c38a2f" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json new file mode 100644 index 0000000..98a60c6 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-experiment-design.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/09c2a3c1d5060228405fa5cc9147bc5a4c61579a52c0dba4c3d55ee1d03ac5fe" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json new file mode 100644 index 0000000..e78cf7e --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-foundation.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/d1c0d6083c36dc5e277e4a0ad0c7f30338431c50d99754b938eaeb59f1967d1d" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json new file mode 100644 index 0000000..7dd22b1 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-contribute-hypothesis.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/d3c31ef2f50da0924c719329b24028ecf54721d454811dd9aacf4ff46de4e3b3" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json new file mode 100644 index 0000000..4397d87 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-forecast-empirical.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/0e519bb9ba1feb38105fd2d108e5bedf21a0c390f1bc248cf6c33a1a852f24d9" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json new file mode 100644 index 0000000..a1f9011 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-forecast-mechanistic.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/e0cf1ad5fb075f3e94ac70f9c426fb63f33cf7e67ad076a17d93381dc465dc76" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-funding.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-funding.json new file mode 100644 index 0000000..b308e1b --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-funding.json @@ -0,0 +1,3 @@ +{ + "artifact_ref": "artifact://sha256/0067fb1dc8736fd862f156be1d17d9c1efea28fbb043c72381a951e5a6b100c1" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-init.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-init.json new file mode 100644 index 0000000..8ba203a --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-init.json @@ -0,0 +1,4 @@ +{ + "initialized": true, + "workspace": "/opt/qwen35-acceptance-v3/workspace" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-run.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-run.json new file mode 100644 index 0000000..e69de29 diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-run.stderr b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-run.stderr new file mode 100644 index 0000000..a0f0a10 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-run.stderr @@ -0,0 +1 @@ +error: execution failed: run did not resolve to any frozen outcome diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-status.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-status.json new file mode 100644 index 0000000..0b48532 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/ilxyr-status.json @@ -0,0 +1,92 @@ +{ + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance", + "compiled_ref": "artifact://sha256/c5399dba347f001bf553941b88aae7e74e0609bbded168252ae731eff6a39434", + "forecasts": 2, + "total_stake": 10, + "funding_commitments": 1, + "funded_compute_credits": 100, + "latest_admission": { + "schema": "ilxyr.admission.v1", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance", + "accepted": true, + "checks": [ + { + "gate": "methodology", + "passed": true, + "detail": "4/4 required research stages resolved" + }, + { + "gate": "outcome_contract", + "passed": true, + "detail": "3 frozen mutually evaluated outcomes declared" + }, + { + "gate": "forecast_participation", + "passed": true, + "detail": "2/2 distinct forecasters" + }, + { + "gate": "forecast_stake", + "passed": true, + "detail": "10/10 forecast credits staked" + }, + { + "gate": "compute_funding", + "passed": true, + "detail": "100/100 compute credits committed" + }, + { + "gate": "executor_available", + "passed": true, + "detail": "local-command adapter is installed" + }, + { + "gate": "weight_protection", + "passed": true, + "detail": "local execution is limited to public-weight handles" + }, + { + "gate": "local_execution_policy", + "passed": true, + "detail": "local execution requires network=open and an absolute executable path" + }, + { + "gate": "code_policy", + "passed": true, + "detail": "local-command directly executes the declared program" + }, + { + "gate": "export_policy", + "passed": true, + "detail": "local-command records stdout, stderr, and metric artifacts" + }, + { + "gate": "reviewer_separation", + "passed": true, + "detail": "proposer and engineering reviewer use distinct handles" + }, + { + "gate": "forecaster_separation", + "passed": true, + "detail": "proposer and forecasters use distinct handles" + } + ], + "decided_at_ms": 1786647171244 + }, + "execution_started": true, + "latest_run": { + "schema": "ilxyr.run.v1", + "id": "run:lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance:1786647171758", + "experiment_id": "lecore.qwen35.install.20c3330d0b3e.59f551d452cf.a6c9d1135a04.f399cd6bc7f2.v3.acceptance", + "started_at_ms": 1786647171758, + "completed_at_ms": 1786656088229, + "exit_code": 0, + "timed_out": false, + "stdout": " qkv layout: flat (ppl 20.34 vs 201031.79 for grouped -- 9881.2x better)\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n sanity: perplexity 16.2 on plain English (chance ~248320) -- looks correct\n{\"metrics\": {\"acceptance_pass\": 0.0, \"emitted_checkpoint_mb\": 2112.911655, \"eval_tokens\": 4096.0, \"experimental_installer_used\": 1.0, \"installed_perplexity\": 10.357574708230848, \"official_reload_pass\": 1.0, \"original_perplexity\": 7.9062244629535465, \"paired_block_length\": 4.0, \"paired_ci_hi_nats\": 0.34246740100294987, \"paired_ci_lo_nats\": 0.20259645914578694, \"paired_effective_tokens\": 2279.0, \"peak_gpu_mb\": 0.0, \"peak_rss_mb\": 24306.52734375, \"perplexity_delta_pct\": 31.005320640258482, \"reference_logit_parity_pass\": 1.0, \"reference_logit_relative_error\": 6.787656774900957e-07, \"reload_pass\": 1.0, \"source_clean\": 1.0, \"spectral_filtering_enabled\": 0.0, \"statistical_gate_pass\": 0.0, \"text_generation_pass\": 0.0, \"tokenizer_parity_pass\": 1.0, \"vision_smoke_pass\": 0.0}, \"source\": {\"artifacts\": [{\"path\": \"experiments/qwen35_acceptance/run.py\", \"sha256\": \"d1c3ecec0fe19a9aea1c30354032919ad25b15092ffd547a3f487dcb1a10f849\"}, {\"path\": \"experiments/qwen35_acceptance/contract.py\", \"sha256\": \"3710f24df07431b2be710ee291f6369aaae48c369bbab32e3534cfa5d31a501b\"}, {\"path\": \"assimilation/install.py\", \"sha256\": \"a980357333b0820230ec5deba94ce3cb95260a17da95c786997a7bc527a8ddce\"}, {\"path\": \"holographic/io_and_interop/holographic_measure.py\", \"sha256\": \"95ceb45ca9e57962cb1e81e6a335f8ff2d46f7f6b7a128a62b8bdd2682a15527\"}, {\"path\": \"holographic/io_and_interop/holographic_gdnruntime.py\", \"sha256\": \"82122011e226ee8eee0ce2ffeec946dcd60f4f95c622c627e6cb7522d82a7928\"}], \"commit\": \"f399cd6bc7f24383a792fdcf32c43a4e1197d4f1\", \"inputs\": {\"evaluation_corpus\": {\"bytes\": 1213410, \"path\": \"/opt/qwen35-acceptance-v3/inputs/federalist-papers.txt\", \"sha256\": \"a6c9d1135a04d10955fe11d210b7f642e1c2341d4f2c8369b9a832cc97839d94\"}, \"installation_corpus\": {\"bytes\": 2303549, \"path\": \"/opt/qwen35-acceptance-v3/lecore/REFERENCE.md\", \"sha256\": \"59f551d452cfb0664deeb6dfbe5b98598abb1633cc38a5bb020ec100ac3cd33a\"}}, \"repository\": \"https://github.com/atimics/holostuff.git\"}}\n", + "stderr": "[transformers] `torch_dtype` is deprecated! Use `dtype` instead!\n[transformers] The fast path is not available because one of the required library is not installed. Falling back to torch implementation. To install follow https://github.com/fla-org/flash-linear-attention#installation and https://github.com/Dao-AILab/causal-conv1d\n\rLoading weights: 0%| | 0/320 [00:00 +AuthorDate: Thu Aug 13 11:44:59 2026 -0700 +Commit: atimics +CommitDate: Thu Aug 13 11:44:59 2026 -0700 + + fix(qwen): emit official-compatible installed checkpoints diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/lecore-status.txt b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/lecore-status.txt new file mode 100644 index 0000000..e69de29 diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/model-manifest.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/model-manifest.json new file mode 100644 index 0000000..65559ac --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/model-manifest.json @@ -0,0 +1,66 @@ +{ + "digest": "20c3330d0b3e00c98b38e7813202b12952f37d6a4415a30da103c9bfe994d680", + "files": [ + { + "bytes": 11544, + "path": "LICENSE", + "sha256": "bbedc3fda3305820b977265f01b8619d87570a6739de3a5582c3464840f1e57a" + }, + { + "bytes": 61705, + "path": "README.md", + "sha256": "87a163af54f32fa608a0f8d3ac67945c53dd2b4a7c96740b3d7fdc28e8458864" + }, + { + "bytes": 7755, + "path": "chat_template.jinja", + "sha256": "273d8e0e683b885071fb17e08d71e5f2a5ddfb5309756181681de4f5a1822d80" + }, + { + "bytes": 2907, + "path": "config.json", + "sha256": "b90b86f35c8e6925ef74ee04d0e758f0a845c83a42089ad82bbaa948de9b4204" + }, + { + "bytes": 3353259, + "path": "merges.txt", + "sha256": "a9d356d7bdf1ef4949e3e748e95b8e10ad9d4e2e838eddc38a0a7b6b94d1db8d" + }, + { + "bytes": 1746942600, + "path": "model.safetensors-00001-of-00001.safetensors", + "sha256": "04b1c301231dd422b8860db31311ab2721511346a32cb1e079c4c4e5f1fe4696" + }, + { + "bytes": 50900, + "path": "model.safetensors.index.json", + "sha256": "d8a08838a613b025eb7952ed9db11696213e57e76a375661ef5c12f9dd5dcf4e" + }, + { + "bytes": 390, + "path": "preprocessor_config.json", + "sha256": "27225450ac9c6529872ee1924fcb0962ff5634834f817040f444118116f4e516" + }, + { + "bytes": 12807982, + "path": "tokenizer.json", + "sha256": "5f9e4d4901a92b997e463c1f46055088b6cca5ca61a6522d1b9f64c4bb81cb42" + }, + { + "bytes": 16709, + "path": "tokenizer_config.json", + "sha256": "49e2b6e395f959f077f1e992b338919c0d4a9732fc6e613995e06557f843500c" + }, + { + "bytes": 385, + "path": "video_preprocessor_config.json", + "sha256": "7768af27c1fafa9cc9011c1dc20067e03f8915e03b63504550e11d5066986d13" + }, + { + "bytes": 6722759, + "path": "vocab.json", + "sha256": "ce99b4cb2983d118806ce0a8b777a35b093e2000a503ebde25853284c9dfa003" + } + ], + "revision": "2fc06364715b967f1860aea9cf38778875588b17" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/requirements-resolved.txt b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/requirements-resolved.txt new file mode 100644 index 0000000..7bfa4f7 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/requirements-resolved.txt @@ -0,0 +1,36 @@ +Jinja2==3.1.6 +MarkupSafe==3.0.3 +PyYAML==6.0.3 +Pygments==2.20.0 +annotated-doc==0.0.5 +anyio==4.14.2 +certifi==2026.7.22 +click==8.4.2 +filelock==3.29.0 +fsspec==2026.4.0 +h11==0.16.0 +hf-xet==1.6.0 +httpcore==1.0.9 +httpx==0.28.1 +huggingface_hub==1.27.0 +idna==3.18 +markdown-it-py==4.2.0 +mdurl==0.1.2 +mpmath==1.3.0 +networkx==3.6.1 +numpy==2.4.6 +packaging==26.3 +pillow==12.3.0 +pip==23.2.1 +regex==2026.7.19 +rich==15.0.0 +safetensors==0.8.0 +setuptools==78.1.0 +shellingham==1.5.4 +sympy==1.14.0 +tokenizers==0.22.2 +torch==2.11.0+cpu +tqdm==4.70.0 +transformers==5.14.0 +typer==0.27.1 +typing_extensions==4.15.0 diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/runner-status.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/runner-status.json new file mode 100644 index 0000000..8be43cc --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/runner-status.json @@ -0,0 +1,7 @@ +{ + "exit_code": 1, + "formal_run_started": true, + "schema": "lecore.qwen35-runner-status.v3", + "stage": "execution_failure", + "updated_at": "2026-08-13T21:21:28.763310+00:00" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/s3-terminal-status.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/s3-terminal-status.json new file mode 100644 index 0000000..6011215 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/s3-terminal-status.json @@ -0,0 +1,7 @@ +{ + "exit_code": 1, + "formal_run_started": true, + "schema": "lecore.qwen35-runner-status.v3", + "stage": "execution_failure", + "updated_at": "2026-08-13T21:21:31.540759+00:00" +} diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/user-data.log b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/user-data.log new file mode 100644 index 0000000..8bef394 --- /dev/null +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/result/user-data.log @@ -0,0 +1,462 @@ +Shutdown scheduled for Fri 2026-08-14 02:51:08 UTC, use 'shutdown -c' to cancel. +Amazon Linux 2023 repository 80 MB/s | 75 MB 00:00 +Amazon Linux 2023 Kernel Livepatch repository 572 kB/s | 69 kB 00:00 +Package tar-2:1.34-1.amzn2023.0.4.x86_64 is already installed. +Package gzip-1.12-1.amzn2023.0.1.x86_64 is already installed. +Dependencies resolved. +================================================================================ + Package Arch Version Repository Size +================================================================================ +Installing: + cargo x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 8.2 M + git x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 53 k + python3.12 x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 27 k + python3.12-pip noarch 23.2.1-4.amzn2023.0.10 amazonlinux 2.8 M + rust x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 62 M +Installing dependencies: + annobin-docs noarch 12.69-1.amzn2023.0.1 amazonlinux 93 k + annobin-plugin-gcc x86_64 12.69-1.amzn2023.0.1 amazonlinux 974 k + cpp x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 11 M + gc x86_64 8.0.4-5.amzn2023.0.2 amazonlinux 105 k + gcc x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 36 M + gcc-plugin-annobin x86_64 11.5.0-5.amzn2023.0.5 amazonlinux 38 k + git-core x86_64 2.50.1-1.amzn2023.0.1 amazonlinux 4.9 M + git-core-doc noarch 2.50.1-1.amzn2023.0.1 amazonlinux 2.8 M + glibc-devel x86_64 2.34-231.amzn2023.0.5 amazonlinux 33 k + glibc-headers-x86 noarch 2.34-231.amzn2023.0.5 amazonlinux 439 k + guile22 x86_64 2.2.7-2.amzn2023.0.3 amazonlinux 6.4 M + kernel-headers x86_64 1:6.1.177-224.371.amzn2023 amazonlinux 1.4 M + libmpc x86_64 1.2.1-2.amzn2023.0.2 amazonlinux 62 k + libtool-ltdl x86_64 2.4.7-1.amzn2023.0.3 amazonlinux 38 k + libxcrypt-devel x86_64 4.4.33-7.amzn2023 amazonlinux 32 k + make x86_64 1:4.3-5.amzn2023.0.2 amazonlinux 534 k + mpdecimal x86_64 2.5.1-3.amzn2023.0.3 amazonlinux 101 k + perl-Error noarch 1:0.17030-2.amzn2023.0.1 amazonlinux 42 k + perl-File-Find noarch 1.37-477.amzn2023.0.9 amazonlinux 26 k + perl-Git noarch 2.50.1-1.amzn2023.0.1 amazonlinux 41 k + perl-TermReadKey x86_64 2.38-9.amzn2023.0.3 amazonlinux 36 k + perl-lib x86_64 0.65-477.amzn2023.0.9 amazonlinux 15 k + python3.12-libs x86_64 3.12.13-2.amzn2023.0.5 amazonlinux 9.2 M + python3.12-pip-wheel noarch 23.2.1-4.amzn2023.0.10 amazonlinux 1.5 M + rust-std-static x86_64 1.97.0-1.amzn2023.0.1 amazonlinux 39 M +Installing weak dependencies: + python3.12-setuptools noarch 68.2.2-4.amzn2023.0.3 amazonlinux 1.4 M + +Transaction Summary +================================================================================ +Install 31 Packages + +Total download size: 189 M +Installed size: 700 M +Downloading Packages: +(1/31): annobin-plugin-gcc-12.69-1.amzn2023.0.1 23 MB/s | 974 kB 00:00 +(2/31): annobin-docs-12.69-1.amzn2023.0.1.noarc 2.0 MB/s | 93 kB 00:00 +(3/31): gc-8.0.4-5.amzn2023.0.2.x86_64.rpm 3.4 MB/s | 105 kB 00:00 +(4/31): cpp-11.5.0-5.amzn2023.0.5.x86_64.rpm 77 MB/s | 11 MB 00:00 +(5/31): gcc-plugin-annobin-11.5.0-5.amzn2023.0. 1.3 MB/s | 38 kB 00:00 +(6/31): git-2.50.1-1.amzn2023.0.1.x86_64.rpm 2.0 MB/s | 53 kB 00:00 +(7/31): cargo-1.97.0-1.amzn2023.0.1.x86_64.rpm 28 MB/s | 8.2 MB 00:00 +(8/31): git-core-2.50.1-1.amzn2023.0.1.x86_64.r 67 MB/s | 4.9 MB 00:00 +(9/31): git-core-doc-2.50.1-1.amzn2023.0.1.noar 70 MB/s | 2.8 MB 00:00 +(10/31): glibc-devel-2.34-231.amzn2023.0.5.x86_ 1.2 MB/s | 33 kB 00:00 +(11/31): glibc-headers-x86-2.34-231.amzn2023.0. 11 MB/s | 439 kB 00:00 +(12/31): kernel-headers-6.1.177-224.371.amzn202 37 MB/s | 1.4 MB 00:00 +(13/31): guile22-2.2.7-2.amzn2023.0.3.x86_64.rp 73 MB/s | 6.4 MB 00:00 +(14/31): libmpc-1.2.1-2.amzn2023.0.2.x86_64.rpm 2.0 MB/s | 62 kB 00:00 +(15/31): libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_ 1.6 MB/s | 38 kB 00:00 +(16/31): gcc-11.5.0-5.amzn2023.0.5.x86_64.rpm 77 MB/s | 36 MB 00:00 +(17/31): libxcrypt-devel-4.4.33-7.amzn2023.x86_ 300 kB/s | 32 kB 00:00 +(18/31): make-4.3-5.amzn2023.0.2.x86_64.rpm 5.2 MB/s | 534 kB 00:00 +(19/31): perl-Error-0.17030-2.amzn2023.0.1.noar 1.6 MB/s | 42 kB 00:00 +(20/31): mpdecimal-2.5.1-3.amzn2023.0.3.x86_64. 3.4 MB/s | 101 kB 00:00 +(21/31): perl-File-Find-1.37-477.amzn2023.0.9.n 970 kB/s | 26 kB 00:00 +(22/31): perl-Git-2.50.1-1.amzn2023.0.1.noarch. 1.5 MB/s | 41 kB 00:00 +(23/31): perl-TermReadKey-2.38-9.amzn2023.0.3.x 1.3 MB/s | 36 kB 00:00 +(24/31): perl-lib-0.65-477.amzn2023.0.9.x86_64. 535 kB/s | 15 kB 00:00 +(25/31): python3.12-3.12.13-2.amzn2023.0.5.x86_ 1.2 MB/s | 27 kB 00:00 +(26/31): python3.12-pip-23.2.1-4.amzn2023.0.10. 67 MB/s | 2.8 MB 00:00 +(27/31): python3.12-pip-wheel-23.2.1-4.amzn2023 42 MB/s | 1.5 MB 00:00 +(28/31): python3.12-setuptools-68.2.2-4.amzn202 50 MB/s | 1.4 MB 00:00 +(29/31): python3.12-libs-3.12.13-2.amzn2023.0.5 77 MB/s | 9.2 MB 00:00 +(30/31): rust-std-static-1.97.0-1.amzn2023.0.1. 70 MB/s | 39 MB 00:00 +(31/31): rust-1.97.0-1.amzn2023.0.1.x86_64.rpm 78 MB/s | 62 MB 00:00 +-------------------------------------------------------------------------------- +Total 127 MB/s | 189 MB 00:01 +Running transaction check +Transaction check succeeded. +Running transaction test +Transaction test succeeded. +Running transaction + Preparing : 1/1 + Installing : libmpc-1.2.1-2.amzn2023.0.2.x86_64 1/31 + Installing : git-core-2.50.1-1.amzn2023.0.1.x86_64 2/31 + Installing : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 3/31 + Installing : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Installing : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 5/31 + Installing : perl-lib-0.65-477.amzn2023.0.9.x86_64 6/31 + Installing : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 7/31 + Installing : perl-File-Find-1.37-477.amzn2023.0.9.noarch 8/31 + Installing : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 9/31 + Installing : perl-Git-2.50.1-1.amzn2023.0.1.noarch 10/31 + Installing : git-2.50.1-1.amzn2023.0.1.x86_64 11/31 + Installing : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 12/31 + Installing : python3.12-3.12.13-2.amzn2023.0.5.x86_64 13/31 + Installing : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 14/31 + Installing : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 15/31 + Installing : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Installing : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 17/31 + Installing : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 18/31 + Installing : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 19/31 + Installing : glibc-devel-2.34-231.amzn2023.0.5.x86_64 20/31 + Installing : gc-8.0.4-5.amzn2023.0.2.x86_64 21/31 + Installing : guile22-2.2.7-2.amzn2023.0.3.x86_64 22/31 + Installing : make-1:4.3-5.amzn2023.0.2.x86_64 23/31 + Installing : gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Running scriptlet: gcc-11.5.0-5.amzn2023.0.5.x86_64 24/31 + Installing : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 25/31 + Installing : rust-1.97.0-1.amzn2023.0.1.x86_64 26/31 + Installing : annobin-docs-12.69-1.amzn2023.0.1.noarch 27/31 + Installing : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Running scriptlet: annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 28/31 + Installing : cargo-1.97.0-1.amzn2023.0.1.x86_64 29/31 + Installing : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 30/31 + Installing : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Running scriptlet: python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 31/31 + Verifying : annobin-docs-12.69-1.amzn2023.0.1.noarch 1/31 + Verifying : annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 2/31 + Verifying : cargo-1.97.0-1.amzn2023.0.1.x86_64 3/31 + Verifying : cpp-11.5.0-5.amzn2023.0.5.x86_64 4/31 + Verifying : gc-8.0.4-5.amzn2023.0.2.x86_64 5/31 + Verifying : gcc-11.5.0-5.amzn2023.0.5.x86_64 6/31 + Verifying : gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 7/31 + Verifying : git-2.50.1-1.amzn2023.0.1.x86_64 8/31 + Verifying : git-core-2.50.1-1.amzn2023.0.1.x86_64 9/31 + Verifying : git-core-doc-2.50.1-1.amzn2023.0.1.noarch 10/31 + Verifying : glibc-devel-2.34-231.amzn2023.0.5.x86_64 11/31 + Verifying : glibc-headers-x86-2.34-231.amzn2023.0.5.noarch 12/31 + Verifying : guile22-2.2.7-2.amzn2023.0.3.x86_64 13/31 + Verifying : kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 14/31 + Verifying : libmpc-1.2.1-2.amzn2023.0.2.x86_64 15/31 + Verifying : libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 16/31 + Verifying : libxcrypt-devel-4.4.33-7.amzn2023.x86_64 17/31 + Verifying : make-1:4.3-5.amzn2023.0.2.x86_64 18/31 + Verifying : mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 19/31 + Verifying : perl-Error-1:0.17030-2.amzn2023.0.1.noarch 20/31 + Verifying : perl-File-Find-1.37-477.amzn2023.0.9.noarch 21/31 + Verifying : perl-Git-2.50.1-1.amzn2023.0.1.noarch 22/31 + Verifying : perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 23/31 + Verifying : perl-lib-0.65-477.amzn2023.0.9.x86_64 24/31 + Verifying : python3.12-3.12.13-2.amzn2023.0.5.x86_64 25/31 + Verifying : python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 26/31 + Verifying : python3.12-pip-23.2.1-4.amzn2023.0.10.noarch 27/31 + Verifying : python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch 28/31 + Verifying : python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch 29/31 + Verifying : rust-1.97.0-1.amzn2023.0.1.x86_64 30/31 + Verifying : rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 31/31 + +Installed: + annobin-docs-12.69-1.amzn2023.0.1.noarch + annobin-plugin-gcc-12.69-1.amzn2023.0.1.x86_64 + cargo-1.97.0-1.amzn2023.0.1.x86_64 + cpp-11.5.0-5.amzn2023.0.5.x86_64 + gc-8.0.4-5.amzn2023.0.2.x86_64 + gcc-11.5.0-5.amzn2023.0.5.x86_64 + gcc-plugin-annobin-11.5.0-5.amzn2023.0.5.x86_64 + git-2.50.1-1.amzn2023.0.1.x86_64 + git-core-2.50.1-1.amzn2023.0.1.x86_64 + git-core-doc-2.50.1-1.amzn2023.0.1.noarch + glibc-devel-2.34-231.amzn2023.0.5.x86_64 + glibc-headers-x86-2.34-231.amzn2023.0.5.noarch + guile22-2.2.7-2.amzn2023.0.3.x86_64 + kernel-headers-1:6.1.177-224.371.amzn2023.x86_64 + libmpc-1.2.1-2.amzn2023.0.2.x86_64 + libtool-ltdl-2.4.7-1.amzn2023.0.3.x86_64 + libxcrypt-devel-4.4.33-7.amzn2023.x86_64 + make-1:4.3-5.amzn2023.0.2.x86_64 + mpdecimal-2.5.1-3.amzn2023.0.3.x86_64 + perl-Error-1:0.17030-2.amzn2023.0.1.noarch + perl-File-Find-1.37-477.amzn2023.0.9.noarch + perl-Git-2.50.1-1.amzn2023.0.1.noarch + perl-TermReadKey-2.38-9.amzn2023.0.3.x86_64 + perl-lib-0.65-477.amzn2023.0.9.x86_64 + python3.12-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-libs-3.12.13-2.amzn2023.0.5.x86_64 + python3.12-pip-23.2.1-4.amzn2023.0.10.noarch + python3.12-pip-wheel-23.2.1-4.amzn2023.0.10.noarch + python3.12-setuptools-68.2.2-4.amzn2023.0.3.noarch + rust-1.97.0-1.amzn2023.0.1.x86_64 + rust-std-static-1.97.0-1.amzn2023.0.1.x86_64 + +Complete! +Cloning into '/opt/qwen35-acceptance-v3/lecore'... +HEAD is now at f399cd6 fix(qwen): emit official-compatible installed checkpoints +/opt/qwen35-acceptance-v3/lecore/REFERENCE.md: OK +/opt/qwen35-acceptance-v3/inputs/federalist-papers.txt: OK +Cloning into '/opt/qwen35-acceptance-v3/ilxyr'... +HEAD is now at e92382f Publish EXP-006 experiment evidence + Updating crates.io index + Downloading crates ... + Downloaded digest v0.10.7 + Downloaded ed25519 v2.2.3 + Downloaded block-buffer v0.10.4 + Downloaded serde_json v1.0.150 + Downloaded version_check v0.9.5 + Downloaded cfg-if v1.0.4 + Downloaded rand_core v0.6.4 + Downloaded pkcs8 v0.10.2 + Downloaded crypto-common v0.1.7 + Downloaded generic-array v0.14.7 + Downloaded sha2 v0.10.9 + Downloaded signature v2.2.0 + Downloaded semver v1.0.28 + Downloaded cpufeatures v0.2.17 + Downloaded itoa v1.0.18 + Downloaded curve25519-dalek-derive v0.1.1 + Downloaded proc-macro2 v1.0.106 + Downloaded serde_core v1.0.229 + Downloaded serde v1.0.229 + Downloaded getrandom v0.2.17 + Downloaded rustc_version v0.4.1 + Downloaded serde_derive v1.0.229 + Downloaded subtle v2.6.1 + Downloaded const-oid v0.9.6 + Downloaded spki v0.7.3 + Downloaded base64 v0.22.1 + Downloaded base64ct v1.8.3 + Downloaded quote v1.0.46 + Downloaded zmij v1.0.23 + Downloaded ed25519-dalek v2.2.0 + Downloaded unicode-ident v1.0.24 + Downloaded zeroize v1.9.0 + Downloaded typenum v1.20.1 + Downloaded der v0.7.10 + Downloaded memchr v2.8.3 + Downloaded libc v0.2.186 + Downloaded syn v2.0.119 + Downloaded syn v3.0.0 + Downloaded curve25519-dalek v4.1.3 + Compiling proc-macro2 v1.0.106 + Compiling version_check v0.9.5 + Compiling unicode-ident v1.0.24 + Compiling quote v1.0.46 + Compiling typenum v1.20.1 + Compiling semver v1.0.28 + Compiling serde_core v1.0.229 + Compiling cpufeatures v0.2.17 + Compiling cfg-if v1.0.4 + Compiling zmij v1.0.23 + Compiling serde v1.0.229 + Compiling signature v2.2.0 + Compiling zeroize v1.9.0 + Compiling subtle v2.6.1 + Compiling serde_json v1.0.150 + Compiling itoa v1.0.18 + Compiling memchr v2.8.3 + Compiling base64 v0.22.1 + Compiling ed25519 v2.2.3 + Compiling rustc_version v0.4.1 + Compiling generic-array v0.14.7 + Compiling curve25519-dalek v4.1.3 + Compiling syn v2.0.119 + Compiling syn v3.0.0 + Compiling crypto-common v0.1.7 + Compiling block-buffer v0.10.4 + Compiling digest v0.10.7 + Compiling sha2 v0.10.9 + Compiling serde_derive v1.0.229 + Compiling curve25519-dalek-derive v0.1.1 + Compiling ed25519-dalek v2.2.0 + Compiling ilxyr-core v0.1.0 (/opt/qwen35-acceptance-v3/ilxyr/crates/ilxyr-core) + Compiling ilxyr-cli v0.1.0 (/opt/qwen35-acceptance-v3/ilxyr/crates/ilxyr-cli) + Finished `release` profile [optimized] target(s) in 20.61s +Looking in indexes: https://download.pytorch.org/whl/cpu +Collecting torch==2.11.0 + Obtaining dependency information for torch==2.11.0 from https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata + Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl.metadata (29 kB) +Collecting filelock (from torch==2.11.0) + Obtaining dependency information for filelock from https://files.pythonhosted.org/packages/81/47/dd9a212ef6e343a6857485ffe25bba537304f1913bdbed446a23f7f592e1/filelock-3.29.0-py3-none-any.whl.metadata + Downloading filelock-3.29.0-py3-none-any.whl.metadata (2.0 kB) +Collecting typing-extensions>=4.10.0 (from torch==2.11.0) + Obtaining dependency information for typing-extensions>=4.10.0 from https://files.pythonhosted.org/packages/18/67/36e9267722cc04a6b9f15c7f3441c2363321a3ea07da7ae0c0707beb2a9c/typing_extensions-4.15.0-py3-none-any.whl.metadata + Downloading typing_extensions-4.15.0-py3-none-any.whl.metadata (3.3 kB) +Collecting setuptools<82 (from torch==2.11.0) + Obtaining dependency information for setuptools<82 from https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl.metadata (6.6 kB) +Collecting sympy>=1.13.3 (from torch==2.11.0) + Obtaining dependency information for sympy>=1.13.3 from https://files.pythonhosted.org/packages/a2/09/77d55d46fd61b4a135c444fc97158ef34a095e5681d0a6c10b75bf356191/sympy-1.14.0-py3-none-any.whl.metadata + Downloading sympy-1.14.0-py3-none-any.whl.metadata (12 kB) +Collecting networkx>=2.5.1 (from torch==2.11.0) + Obtaining dependency information for networkx>=2.5.1 from https://files.pythonhosted.org/packages/9e/c9/b2622292ea83fbb4ec318f5b9ab867d0a28ab43c5717bb85b0a5f6b3b0a4/networkx-3.6.1-py3-none-any.whl.metadata + Downloading networkx-3.6.1-py3-none-any.whl.metadata (6.8 kB) +Collecting jinja2 (from torch==2.11.0) + Obtaining dependency information for jinja2 from https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata + Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl.metadata (2.9 kB) +Collecting fsspec>=0.8.5 (from torch==2.11.0) + Obtaining dependency information for fsspec>=0.8.5 from https://files.pythonhosted.org/packages/d5/0c/043d5e551459da400957a1395e0febbf771446ff34291afcbe3d8be2a279/fsspec-2026.4.0-py3-none-any.whl.metadata + Downloading fsspec-2026.4.0-py3-none-any.whl.metadata (10 kB) +Collecting mpmath<1.4,>=1.1.0 (from sympy>=1.13.3->torch==2.11.0) + Obtaining dependency information for mpmath<1.4,>=1.1.0 from https://files.pythonhosted.org/packages/43/e3/7d92a15f894aa0c9c4b49b8ee9ac9850d6e63b03c9c32c0367a13ae62209/mpmath-1.3.0-py3-none-any.whl.metadata + Downloading mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) +Collecting MarkupSafe>=2.0 (from jinja2->torch==2.11.0) + Obtaining dependency information for MarkupSafe>=2.0 from https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.7 kB) +Downloading https://download-r2.pytorch.org/whl/cpu/torch-2.11.0%2Bcpu-cp312-cp312-manylinux_2_28_x86_64.whl (190.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 190.3/190.3 MB 13.7 MB/s eta 0:00:00 +Downloading fsspec-2026.4.0-py3-none-any.whl (203 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 203.4/203.4 kB 20.7 MB/s eta 0:00:00 +Downloading networkx-3.6.1-py3-none-any.whl (2.1 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 2.1/2.1 MB 122.0 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/setuptools-78.1.0-py3-none-any.whl (1.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.3/1.3 MB 158.8 MB/s eta 0:00:00 +Downloading sympy-1.14.0-py3-none-any.whl (6.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.3/6.3 MB 196.7 MB/s eta 0:00:00 +Downloading typing_extensions-4.15.0-py3-none-any.whl (44 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 44.6/44.6 kB 16.8 MB/s eta 0:00:00 +Downloading filelock-3.29.0-py3-none-any.whl (39 kB) +Downloading https://download.pytorch.org/whl/jinja2-3.1.6-py3-none-any.whl (134 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 134.9/134.9 kB 38.3 MB/s eta 0:00:00 +Downloading https://download.pytorch.org/whl/markupsafe-3.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (22 kB) +Downloading mpmath-1.3.0-py3-none-any.whl (536 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 536.2/536.2 kB 123.6 MB/s eta 0:00:00 +Installing collected packages: mpmath, typing-extensions, sympy, setuptools, networkx, MarkupSafe, fsspec, filelock, jinja2, torch +Successfully installed MarkupSafe-3.0.3 filelock-3.29.0 fsspec-2026.4.0 jinja2-3.1.6 mpmath-1.3.0 networkx-3.6.1 setuptools-78.1.0 sympy-1.14.0 torch-2.11.0+cpu typing-extensions-4.15.0 +Collecting numpy==2.4.6 + Obtaining dependency information for numpy==2.4.6 from https://files.pythonhosted.org/packages/99/60/14115e6364fa676c5397c2ad3004e527e9aa487abf5d0706ec81bbd08529/numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (6.6 kB) +Collecting pillow==12.3.0 + Obtaining dependency information for pillow==12.3.0 from https://files.pythonhosted.org/packages/84/21/a35af28dcc61f37ed850a2d64c65c701321dfbf25085e469d5559360cbbf/pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (9.1 kB) +Collecting safetensors==0.8.0 + Obtaining dependency information for safetensors==0.8.0 from https://files.pythonhosted.org/packages/28/50/f203ff3a3ddfe19308efc83c5a3a29ed02bf786732ec35e68bf9162f3365/safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (4.2 kB) +Collecting huggingface-hub==1.27.0 + Obtaining dependency information for huggingface-hub==1.27.0 from https://files.pythonhosted.org/packages/de/d8/95b735e183957c1f26d94c52977f09d466d55119cbbc1558ea4975e4c216/huggingface_hub-1.27.0-py3-none-any.whl.metadata + Downloading huggingface_hub-1.27.0-py3-none-any.whl.metadata (16 kB) +Collecting transformers==5.14.0 + Obtaining dependency information for transformers==5.14.0 from https://files.pythonhosted.org/packages/19/3b/a0a807b957c5f8f2075a8740f597a021420583bb0aa68ff6432d74a530d4/transformers-5.14.0-py3-none-any.whl.metadata + Downloading transformers-5.14.0-py3-none-any.whl.metadata (32 kB) +Collecting click<9.0.0,>=8.4.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for click<9.0.0,>=8.4.2 from https://files.pythonhosted.org/packages/fb/e2/79c688af8b210d232694e31e59da9f6ec747bae31c3f5946e4e9b98860d5/click-8.4.2-py3-none-any.whl.metadata + Downloading click-8.4.2-py3-none-any.whl.metadata (2.6 kB) +Requirement already satisfied: filelock>=3.10.0 in /opt/qwen35-acceptance-v3/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (3.29.0) +Requirement already satisfied: fsspec>=2023.5.0 in /opt/qwen35-acceptance-v3/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (2026.4.0) +Collecting hf-xet<2.0.0,>=1.5.2 (from huggingface-hub==1.27.0) + Obtaining dependency information for hf-xet<2.0.0,>=1.5.2 from https://files.pythonhosted.org/packages/67/4e/a28359bf1c1ecf11eba22123168c138698f7cb576ac678f5a2e16cd5da08/hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata + Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (4.9 kB) +Collecting httpx<1,>=0.23.0 (from huggingface-hub==1.27.0) + Obtaining dependency information for httpx<1,>=0.23.0 from https://files.pythonhosted.org/packages/2a/39/e50c7c3a983047577ee07d2a9e53faf5a69493943ec3f6a384bdc792deb2/httpx-0.28.1-py3-none-any.whl.metadata + Downloading httpx-0.28.1-py3-none-any.whl.metadata (7.1 kB) +Collecting packaging>=20.9 (from huggingface-hub==1.27.0) + Obtaining dependency information for packaging>=20.9 from https://files.pythonhosted.org/packages/63/34/ba1c580383c9eada3711951fef0795c80b829a078d72188184bcab9dd527/packaging-26.3-py3-none-any.whl.metadata + Downloading packaging-26.3-py3-none-any.whl.metadata (3.5 kB) +Collecting pyyaml>=5.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for pyyaml>=5.1 from https://files.pythonhosted.org/packages/8b/9d/b3589d3877982d4f2329302ef98a8026e7f4443c765c46cfecc8858c6b4b/pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (2.4 kB) +Collecting tqdm>=4.42.1 (from huggingface-hub==1.27.0) + Obtaining dependency information for tqdm>=4.42.1 from https://files.pythonhosted.org/packages/f9/1c/01bfd571a64e7f270e6bab5e33777debe0edc56759233ce84f27dec92d14/tqdm-4.70.0-py3-none-any.whl.metadata + Downloading tqdm-4.70.0-py3-none-any.whl.metadata (57 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 57.3/57.3 kB 17.0 MB/s eta 0:00:00 +Requirement already satisfied: typing-extensions>=4.1.0 in /opt/qwen35-acceptance-v3/venv/lib64/python3.12/site-packages (from huggingface-hub==1.27.0) (4.15.0) +Collecting regex>=2025.10.22 (from transformers==5.14.0) + Obtaining dependency information for regex>=2025.10.22 from https://files.pythonhosted.org/packages/2a/8e/096d00c7c480ef2ff4265349b14e2261d4ab787ba1f74e2e80d1c58079c3/regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata + Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl.metadata (40 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 40.5/40.5 kB 12.2 MB/s eta 0:00:00 +Collecting tokenizers<=0.23.0,>=0.22.0 (from transformers==5.14.0) + Obtaining dependency information for tokenizers<=0.23.0,>=0.22.0 from https://files.pythonhosted.org/packages/2e/76/932be4b50ef6ccedf9d3c6639b056a967a86258c6d9200643f01269211ca/tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata + Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (7.3 kB) +Collecting typer (from transformers==5.14.0) + Obtaining dependency information for typer from https://files.pythonhosted.org/packages/43/89/9518bc0c3929bee36b3a4a8e3daddd6e03f92f9961c66d4983b837160543/typer-0.27.1-py3-none-any.whl.metadata + Downloading typer-0.27.1-py3-none-any.whl.metadata (16 kB) +Collecting anyio (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for anyio from https://files.pythonhosted.org/packages/da/35/f2287558c17e29fafc8ef3daf819bb9834061cfa43bff8014f7df7f63bdc/anyio-4.14.2-py3-none-any.whl.metadata + Downloading anyio-4.14.2-py3-none-any.whl.metadata (4.6 kB) +Collecting certifi (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for certifi from https://files.pythonhosted.org/packages/0b/a7/71ac2cff56fec219ed242bb11b8efb69fcc4bec75db06fb7bfe35de520e6/certifi-2026.7.22-py3-none-any.whl.metadata + Downloading certifi-2026.7.22-py3-none-any.whl.metadata (2.5 kB) +Collecting httpcore==1.* (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for httpcore==1.* from https://files.pythonhosted.org/packages/7e/f5/f66802a942d491edb555dd61e3a9961140fd64c90bce1eafd741609d334d/httpcore-1.0.9-py3-none-any.whl.metadata + Downloading httpcore-1.0.9-py3-none-any.whl.metadata (21 kB) +Collecting idna (from httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for idna from https://files.pythonhosted.org/packages/1e/5e/d4e9f1a599fb8e573b7b87160658329fbf28d19eac2718f51fc3def3aa5a/idna-3.18-py3-none-any.whl.metadata + Downloading idna-3.18-py3-none-any.whl.metadata (6.1 kB) +Collecting h11>=0.16 (from httpcore==1.*->httpx<1,>=0.23.0->huggingface-hub==1.27.0) + Obtaining dependency information for h11>=0.16 from https://files.pythonhosted.org/packages/04/4b/29cac41a4d98d144bf5f6d33995617b185d14b22401f75ca86f384e87ff1/h11-0.16.0-py3-none-any.whl.metadata + Downloading h11-0.16.0-py3-none-any.whl.metadata (8.3 kB) +Collecting shellingham>=1.3.0 (from typer->transformers==5.14.0) + Obtaining dependency information for shellingham>=1.3.0 from https://files.pythonhosted.org/packages/e0/f9/0595336914c5619e5f28a1fb793285925a8cd4b432c9da0a987836c7f822/shellingham-1.5.4-py2.py3-none-any.whl.metadata + Downloading shellingham-1.5.4-py2.py3-none-any.whl.metadata (3.5 kB) +Collecting rich>=13.8.0 (from typer->transformers==5.14.0) + Obtaining dependency information for rich>=13.8.0 from https://files.pythonhosted.org/packages/82/3b/64d4899d73f91ba49a8c18a8ff3f0ea8f1c1d75481760df8c68ef5235bf5/rich-15.0.0-py3-none-any.whl.metadata + Downloading rich-15.0.0-py3-none-any.whl.metadata (18 kB) +Collecting annotated-doc>=0.0.2 (from typer->transformers==5.14.0) + Obtaining dependency information for annotated-doc>=0.0.2 from https://files.pythonhosted.org/packages/3e/30/e900b21425a860e195f32e37657aa1f7c7f2b1bfb26f03ca209b90933c06/annotated_doc-0.0.5-py3-none-any.whl.metadata + Downloading annotated_doc-0.0.5-py3-none-any.whl.metadata (6.5 kB) +Collecting markdown-it-py>=2.2.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for markdown-it-py>=2.2.0 from https://files.pythonhosted.org/packages/b3/81/4da04ced5a082363ecfa159c010d200ecbd959ae410c10c0264a38cac0f5/markdown_it_py-4.2.0-py3-none-any.whl.metadata + Downloading markdown_it_py-4.2.0-py3-none-any.whl.metadata (7.4 kB) +Collecting pygments<3.0.0,>=2.13.0 (from rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for pygments<3.0.0,>=2.13.0 from https://files.pythonhosted.org/packages/f4/7e/a72dd26f3b0f4f2bf1dd8923c85f7ceb43172af56d63c7383eb62b332364/pygments-2.20.0-py3-none-any.whl.metadata + Downloading pygments-2.20.0-py3-none-any.whl.metadata (2.5 kB) +Collecting mdurl~=0.1 (from markdown-it-py>=2.2.0->rich>=13.8.0->typer->transformers==5.14.0) + Obtaining dependency information for mdurl~=0.1 from https://files.pythonhosted.org/packages/b3/38/89ba8ad64ae25be8de66a6d463314cf1eb366222074cfda9ee839c56a4b4/mdurl-0.1.2-py3-none-any.whl.metadata + Downloading mdurl-0.1.2-py3-none-any.whl.metadata (1.6 kB) +Downloading numpy-2.4.6-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (16.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 16.6/16.6 MB 137.8 MB/s eta 0:00:00 +Downloading pillow-12.3.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (6.9 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.9/6.9 MB 230.9 MB/s eta 0:00:00 +Downloading safetensors-0.8.0-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (516 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 516.0/516.0 kB 107.2 MB/s eta 0:00:00 +Downloading huggingface_hub-1.27.0-py3-none-any.whl (784 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 784.9/784.9 kB 158.0 MB/s eta 0:00:00 +Downloading transformers-5.14.0-py3-none-any.whl (11.6 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 11.6/11.6 MB 115.3 MB/s eta 0:00:00 +Downloading click-8.4.2-py3-none-any.whl (119 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 119.2/119.2 kB 26.5 MB/s eta 0:00:00 +Downloading hf_xet-1.6.0-cp38-abi3-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (4.5 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 4.5/4.5 MB 187.6 MB/s eta 0:00:00 +Downloading httpx-0.28.1-py3-none-any.whl (73 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 73.5/73.5 kB 29.5 MB/s eta 0:00:00 +Downloading httpcore-1.0.9-py3-none-any.whl (78 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 78.8/78.8 kB 31.0 MB/s eta 0:00:00 +Downloading packaging-26.3-py3-none-any.whl (129 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 130.0/130.0 kB 52.6 MB/s eta 0:00:00 +Downloading pyyaml-6.0.3-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (807 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 807.9/807.9 kB 161.6 MB/s eta 0:00:00 +Downloading regex-2026.7.19-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.manylinux_2_28_x86_64.whl (801 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 801.8/801.8 kB 133.2 MB/s eta 0:00:00 +Downloading tokenizers-0.22.2-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (3.3 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 3.3/3.3 MB 161.6 MB/s eta 0:00:00 +Downloading tqdm-4.70.0-py3-none-any.whl (80 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 80.2/80.2 kB 29.1 MB/s eta 0:00:00 +Downloading typer-0.27.1-py3-none-any.whl (122 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 122.9/122.9 kB 43.5 MB/s eta 0:00:00 +Downloading annotated_doc-0.0.5-py3-none-any.whl (5.3 kB) +Downloading rich-15.0.0-py3-none-any.whl (310 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 310.7/310.7 kB 105.0 MB/s eta 0:00:00 +Downloading shellingham-1.5.4-py2.py3-none-any.whl (9.8 kB) +Downloading anyio-4.14.2-py3-none-any.whl (125 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 125.8/125.8 kB 49.3 MB/s eta 0:00:00 +Downloading idna-3.18-py3-none-any.whl (65 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 65.5/65.5 kB 28.2 MB/s eta 0:00:00 +Downloading certifi-2026.7.22-py3-none-any.whl (136 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 137.0/137.0 kB 49.9 MB/s eta 0:00:00 +Downloading h11-0.16.0-py3-none-any.whl (37 kB) +Downloading markdown_it_py-4.2.0-py3-none-any.whl (91 kB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 91.7/91.7 kB 39.0 MB/s eta 0:00:00 +Downloading pygments-2.20.0-py3-none-any.whl (1.2 MB) + ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.2/1.2 MB 204.7 MB/s eta 0:00:00 +Downloading mdurl-0.1.2-py3-none-any.whl (10.0 kB) +Installing collected packages: tqdm, shellingham, safetensors, regex, pyyaml, pygments, pillow, packaging, numpy, mdurl, idna, hf-xet, h11, click, certifi, annotated-doc, markdown-it-py, httpcore, anyio, rich, httpx, typer, huggingface-hub, tokenizers, transformers +Successfully installed annotated-doc-0.0.5 anyio-4.14.2 certifi-2026.7.22 click-8.4.2 h11-0.16.0 hf-xet-1.6.0 httpcore-1.0.9 httpx-0.28.1 huggingface-hub-1.27.0 idna-3.18 markdown-it-py-4.2.0 mdurl-0.1.2 numpy-2.4.6 packaging-26.3 pillow-12.3.0 pygments-2.20.0 pyyaml-6.0.3 regex-2026.7.19 rich-15.0.0 safetensors-0.8.0 shellingham-1.5.4 tokenizers-0.22.2 tqdm-4.70.0 transformers-5.14.0 typer-0.27.1 + +[notice] A new release of pip is available: 23.2.1 -> 26.2.1 +[notice] To update, run: python3.12 -m pip install --upgrade pip +No broken requirements found. +torch 2.11.0+cpu +transformers 5.14.0 +AutoProcessor Qwen3_5ForConditionalGeneration + Fetching 13 files: 0%| | 0/13 [00:00 Date: Thu, 13 Aug 2026 21:37:52 +0000 Subject: [PATCH 15/17] docs: refresh generated docs (REFERENCE, CAPABILITIES, API_QUICKREF, PIPELINE_MAP, FACULTY_MAP, DOC_MAP) [skip ci] --- REFERENCE.md | 352 +++++++++++++++++++++++++-------------------------- 1 file changed, 176 insertions(+), 176 deletions(-) diff --git a/REFERENCE.md b/REFERENCE.md index ab86e1c..6669956 100644 --- a/REFERENCE.md +++ b/REFERENCE.md @@ -76,8 +76,8 @@ | module | what it is | lines | |---|---|---| | [`holographic_session.py`](#holographic-session) | holographic_session.py -- ONE render session that ties the disconnected rendering threads together. | 227 | -| [`holographic_session.py`](#holographic-session) | SESSION STORE -- contexts that outlive the process. | 360 | | [`holographic_session.py`](#holographic-session) | SESSION -- never compute the same conversation prefix twice. | 253 | +| [`holographic_session.py`](#holographic-session) | SESSION STORE -- contexts that outlive the process. | 360 | ### `splat*` family (6) @@ -423,8 +423,8 @@ | [`holographic_materialio.py`](#holographic-materialio) | PBR material interchange (G2-IO): the standard-format bridge for materials, and a VSA-native carrier. | 307 | | [`holographic_matlib.py`](#holographic-matlib) | holographic_matlib.py -- a comprehensive RENDER material library: plain diffuse -> a fractal planet. | 801 | | [`holographic_meaning_predict.py`](#holographic-meaning-predict) | Generation with structure: predict a next-MEANING vector and settle it, rather | 198 | -| [`holographic_measure.py`](#holographic-measure) | MEASURE -- perplexity with error bars, and decisions that respect them. | 239 | | [`holographic_measure.py`](#holographic-measure) | The variance harness: every headline number gets a mean, a spread, and a confidence | 187 | +| [`holographic_measure.py`](#holographic-measure) | MEASURE -- perplexity with error bars, and decisions that respect them. | 239 | | [`holographic_memory.py`](#holographic-memory) | MEMORY -- the Galvatron's own store, built on leCore's holographic database. | 241 | | [`holographic_memoryhome.py`](#holographic-memoryhome) | holographic_memoryhome.py -- the MEMORY home (consolidation backlog H6): keep the hot working set where the CP | 165 | | [`holographic_memsearch.py`](#holographic-memsearch) | MEMSEARCH -- searchable memory that lives in the weights and runs in the pass. | 212 | @@ -543,8 +543,8 @@ | [`holographic_realtime.py`](#holographic-realtime) | holographic_realtime.py -- the realtime render loop, and the multi-format payload it pushes. | 404 | | [`holographic_reanchor.py`](#holographic-reanchor) | Re-anchoring is load-bearing for deep traversal -- the audit, and the contrast the other tests don't show. | 97 | | [`holographic_reasoning.py`](#holographic-reasoning) | holographic_reasoning.py | 397 | -| [`holographic_recipe.py`](#holographic-recipe) | RECIPE -- ship what leCore ADDED, not the model it was added to. | 334 | | [`holographic_recipe.py`](#holographic-recipe) | A generative recipe-store for constructed holostuff structures. | 206 | +| [`holographic_recipe.py`](#holographic-recipe) | RECIPE -- ship what leCore ADDED, not the model it was added to. | 334 | | [`holographic_recipeops.py`](#holographic-recipeops) | StructureRecipe validator + edit operators (ARCH-1): the recipe equivalent of the mesh Euler operators. | 264 | | [`holographic_reclock.py`](#holographic-reclock) | holographic_reclock.py -- sample when an AXIS moves, not when time passes ("make the boring property the | 313 | | [`holographic_recurrent.py`](#holographic-recurrent) | A gradient-free RECURRENT layer for the holographic engine: reservoir computing. | 456 | @@ -578,8 +578,8 @@ | [`holographic_rolling.py`](#holographic-rolling) | holographic_rolling.py -- the CAUSAL rolling / streaming statistics kit: trailing mean, std, min, max, | 378 | | [`holographic_roomacoustic.py`](#holographic-roomacoustic) | holographic_roomacoustic.py -- A6: GEOMETRIC ROOM ACOUSTICS. How a room echoes -- reflections and reverb. | 161 | | [`holographic_route.py`](#holographic-route) | Representation routing (ARCH-7): route each operation to the representation that supports it -- the policy lay | 183 | -| [`holographic_router.py`](#holographic-router) | Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. | 189 | | [`holographic_router.py`](#holographic-router) | ROUTER -- the model DECIDING, inside one forward pass. | 229 | +| [`holographic_router.py`](#holographic-router) | Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. | 189 | | [`holographic_sampler.py`](#holographic-sampler) | holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). | 223 | | [`holographic_sampling.py`](#holographic-sampling) | Blue-noise / Poisson-disk point sampling -- the EXCLUSION principle, done right. | 134 | | [`holographic_samplinghome.py`](#holographic-samplinghome) | holographic_samplinghome.py -- the SAMPLING home (consolidation backlog R4): one place for the Monte-Carlo | 127 | @@ -684,8 +684,8 @@ | [`holographic_toolclient.py`](#holographic-toolclient) | holographic_toolclient.py -- call another node the same way leCore is called. | 98 | | [`holographic_topology.py`](#holographic-topology) | Principled topology by persistent homology (EXP-7). | 288 | | [`holographic_transfer.py`](#holographic-transfer) | holographic_transfer.py -- KERNEL SCATTER / GATHER: the ONE bundle/readout under every particle<->grid transfe | 224 | -| [`holographic_transform.py`](#holographic-transform) | TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. | 189 | | [`holographic_transform.py`](#holographic-transform) | holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). | 297 | +| [`holographic_transform.py`](#holographic-transform) | TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. | 189 | | [`holographic_transform_space.py`](#holographic-transform-space) | holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it | 176 | | [`holographic_transformbank.py`](#holographic-transformbank) | holographic_transformbank.py -- a prebuilt map of hypervector transforms, and what it can and cannot hold. | 358 | | [`holographic_transformhome.py`](#holographic-transformhome) | holographic_transformhome.py -- the TRANSFORM home (consolidation backlog H5): one facade over "move / rotate | 223 | @@ -14849,6 +14849,39 @@ ### holographic_measure.py +> The variance harness: every headline number gets a mean, a spread, and a confidence +> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. +> +> WHY this exists +> --------------- +> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's +> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A +> single-seed score is therefore a sample from a distribution, and reporting it alone hides +> how wide that distribution is. For an engine whose entire pitch is "measured, not +> promised," reporting a number without its noise is the sharpest blind spot -- so this +> points that same discipline at the numbers themselves. +> +> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the +> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no +> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI +> bound clears the floor -- which is what stops a single fortunate seed from passing a +> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". +> +> USE REAL DATA. The point of the harness is to characterise the real distribution of a +> real claim; running it on a toy makes the spread meaningless. The measurements wired +> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). + +**Public API:** + +- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. +- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. +- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. +- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is +- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above +- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a + +### holographic_measure.py + > MEASURE -- perplexity with error bars, and decisions that respect them. > > Moose asked what assimilation is actually doing. The answer, from his own run: @@ -14881,39 +14914,6 @@ - `def better_than(a, b, alpha, seed, resamples)` -- Is model A better than model B, or is the difference undecidable? - `def tokens_needed(reference, effect_pct, alpha)` -- How many tokens would be needed to RESOLVE an effect of this size. -### holographic_measure.py - -> The variance harness: every headline number gets a mean, a spread, and a confidence -> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. -> -> WHY this exists -> --------------- -> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's -> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A -> single-seed score is therefore a sample from a distribution, and reporting it alone hides -> how wide that distribution is. For an engine whose entire pitch is "measured, not -> promised," reporting a number without its noise is the sharpest blind spot -- so this -> points that same discipline at the numbers themselves. -> -> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the -> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no -> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI -> bound clears the floor -- which is what stops a single fortunate seed from passing a -> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". -> -> USE REAL DATA. The point of the harness is to characterise the real distribution of a -> real claim; running it on a toy makes the spread meaningless. The measurements wired -> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). - -**Public API:** - -- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. -- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. -- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. -- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is -- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above -- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a - ### holographic_memory.py > MEMORY -- the Galvatron's own store, built on leCore's holographic database. @@ -21091,6 +21091,42 @@ ### holographic_recipe.py +> A generative recipe-store for constructed holostuff structures. +> +> WHY THIS EXISTS +> --------------- +> The panel's "proven structure has no noise" result: a structure BUILT by a deterministic proof (a +> derivation over a seed) carries no noise, so it serialises to its *generator* losslessly -- you store the +> recipe, not the expanded vectors, and replaying the recipe reproduces the structure BIT-FOR-BIT. This is +> the easy, exact half of generative compression: when we are the builder we already hold the proof, so +> there is nothing to search for and no residual to code. +> +> A `StructureRecipe` is a tiny replayable build-graph. Each op produces one result vector from a seed and +> earlier results: `atom` (a derived atom -- regenerated from the seed, never stored), `bind`, `bundle`, +> `permute`, `normalize`. You build your structure THROUGH the recipe, so you get both the vectors and the +> recipe that regenerates them. Serialising stores only the op list (and the seed) -- a few hundred bytes +> that regenerate megabytes of structure, exactly. +> +> THE ESCAPE HATCH AND THE KEPT NEGATIVE +> The `raw` op stores a literal vector verbatim. It is the honest boundary: data that was NOT constructed +> (a measured or random vector) has no short recipe, so it must be stored as-is and gets no compression. +> The recipe's compression ratio is therefore exactly the *constructed fraction* of the structure -- all +> recipe -> enormous ratio; all raw -> ~1x. That is the constructed-vs-measured partition made literal. +> +> THE CAPACITY-CLIFF POINT +> Reading structure back out of a single bounded encoded vector degrades past the capacity cliff (crosstalk). +> A recipe does not: it names its leaves explicitly and replays the construction, so a deeply nested +> structure is recovered EXACTLY at any depth. The recipe is the right store for deep constructed structure; +> the expanded superposition is bounded. +> +> Pure NumPy + holostuff kernel, deterministic, JSON serialisation (readable), no new dependencies. + +**Public API:** + +- `class StructureRecipe` -- A replayable build-graph: store the deterministic ops that built a structure, not the vectors. + +### holographic_recipe.py + > RECIPE -- ship what leCore ADDED, not the model it was added to. > > Moose: the inflated model size needs looking at holographically -- we should be @@ -21143,42 +21179,6 @@ - `def cost(rules, arrays, installed_weights)` -- What the recipe saves, in bytes. The number is the whole argument. - `def expand(rules, arrays, base_weights)` -- Rebuild the installed model from the recipe. Must be byte-exact. -### holographic_recipe.py - -> A generative recipe-store for constructed holostuff structures. -> -> WHY THIS EXISTS -> --------------- -> The panel's "proven structure has no noise" result: a structure BUILT by a deterministic proof (a -> derivation over a seed) carries no noise, so it serialises to its *generator* losslessly -- you store the -> recipe, not the expanded vectors, and replaying the recipe reproduces the structure BIT-FOR-BIT. This is -> the easy, exact half of generative compression: when we are the builder we already hold the proof, so -> there is nothing to search for and no residual to code. -> -> A `StructureRecipe` is a tiny replayable build-graph. Each op produces one result vector from a seed and -> earlier results: `atom` (a derived atom -- regenerated from the seed, never stored), `bind`, `bundle`, -> `permute`, `normalize`. You build your structure THROUGH the recipe, so you get both the vectors and the -> recipe that regenerates them. Serialising stores only the op list (and the seed) -- a few hundred bytes -> that regenerate megabytes of structure, exactly. -> -> THE ESCAPE HATCH AND THE KEPT NEGATIVE -> The `raw` op stores a literal vector verbatim. It is the honest boundary: data that was NOT constructed -> (a measured or random vector) has no short recipe, so it must be stored as-is and gets no compression. -> The recipe's compression ratio is therefore exactly the *constructed fraction* of the structure -- all -> recipe -> enormous ratio; all raw -> ~1x. That is the constructed-vs-measured partition made literal. -> -> THE CAPACITY-CLIFF POINT -> Reading structure back out of a single bounded encoded vector degrades past the capacity cliff (crosstalk). -> A recipe does not: it names its leaves explicitly and replays the construction, so a deeply nested -> structure is recovered EXACTLY at any depth. The recipe is the right store for deep constructed structure; -> the expanded superposition is bounded. -> -> Pure NumPy + holostuff kernel, deterministic, JSON serialisation (readable), no new dependencies. - -**Public API:** - -- `class StructureRecipe` -- A replayable build-graph: store the deterministic ops that built a structure, not the vectors. - ### holographic_recipeops.py > StructureRecipe validator + edit operators (ARCH-1): the recipe equivalent of the mesh Euler operators. @@ -22712,33 +22712,6 @@ ### holographic_router.py -> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. -> -> WHY THIS EXISTS (measured, backlog N9/N28): -> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, -> but it has no notion of MEANING: "squish a big array down for storage" shares no token with -> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics -> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median -> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. -> -> WHAT SHIPS (and what does NOT): -> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted -> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a -> query it already has a VECTOR for: -> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) -> * a caller that supplies its own query vector (an app that ran the encoder itself) -> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this -> router says so and returns None, so the caller falls back to the token router rather than guessing. That -> honesty is the whole point: silence beats a confident wrong route. -> -> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. - -**Public API:** - -- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. - -### holographic_router.py - > ROUTER -- the model DECIDING, inside one forward pass. > > Moose raised the architecture that dissolves the wall this project kept hitting: @@ -22778,6 +22751,33 @@ - `def route(runtime, router, text, tokenize)` -- Would this model choose to use the capability on this prompt? - `def install_routed(weights, cfg, operator, router, layer, gain, temperature)` -- Install a circuit whose GATE is the router, not a constant. +### holographic_router.py + +> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. +> +> WHY THIS EXISTS (measured, backlog N9/N28): +> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, +> but it has no notion of MEANING: "squish a big array down for storage" shares no token with +> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics +> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median +> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. +> +> WHAT SHIPS (and what does NOT): +> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted +> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a +> query it already has a VECTOR for: +> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) +> * a caller that supplies its own query vector (an app that ran the encoder itself) +> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this +> router says so and returns None, so the caller falls back to the token router rather than guessing. That +> honesty is the whole point: silence beats a confident wrong route. +> +> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. + +**Public API:** + +- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. + ### holographic_sampler.py > holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). @@ -24408,6 +24408,39 @@ ### holographic_session.py +> SESSION -- never compute the same conversation prefix twice. +> +> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a +> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS +> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens +> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS +> REPEATED, and the fraction grows with every turn. +> +> leCore already had the pieces and never joined them: the runtime exposes +> prefill/step over an InferenceState with copy(), galvacache memoises pure +> functions at a measured 75% hit rate with bit-identical output, and sessions +> exist. What was missing is the RADIX TREE -- the structure that answers "what is +> the longest prefix of this prompt that I have already computed?" +> +> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three +> turns with turn 3, and a dictionary keyed on the whole prompt misses that +> completely. vLLM and SGLang call this RadixAttention; the idea is the same +> whatever the model: index by prefix, resume from the deepest match, compute only +> the tail. +> +> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no +> cache: resuming from a cached state reproduces a full recompute TO FLOAT +> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the +> difference is real rather than pedantic: resuming STEPS the tail while a fresh +> call PREFILLS it, and the two associate their sums differently. The selftest +> asserts the measured bound against the live runtime rather than assuming it. + +**Public API:** + +- `class PrefixCache` -- A radix tree over token sequences, holding inference states. + +### holographic_session.py + > SESSION STORE -- contexts that outlive the process. > > A Galvatron's context is not a transcript, it is STATE: the GDN recurrent @@ -24440,39 +24473,6 @@ - `class SessionStore` -- Named, persistent, independent contexts on disk. - `def runtime_fingerprint(runtime)` -- A cheap, deterministic id for the checkpoint behind a runtime, so a -### holographic_session.py - -> SESSION -- never compute the same conversation prefix twice. -> -> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a -> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS -> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens -> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS -> REPEATED, and the fraction grows with every turn. -> -> leCore already had the pieces and never joined them: the runtime exposes -> prefill/step over an InferenceState with copy(), galvacache memoises pure -> functions at a measured 75% hit rate with bit-identical output, and sessions -> exist. What was missing is the RADIX TREE -- the structure that answers "what is -> the longest prefix of this prompt that I have already computed?" -> -> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three -> turns with turn 3, and a dictionary keyed on the whole prompt misses that -> completely. vLLM and SGLang call this RadixAttention; the idea is the same -> whatever the model: index by prefix, resume from the deepest match, compute only -> the tail. -> -> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no -> cache: resuming from a cached state reproduces a full recompute TO FLOAT -> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the -> difference is real rather than pedantic: resuming STEPS the tail while a fresh -> call PREFILLS it, and the two associate their sums differently. The selftest -> asserts the measured bound against the live runtime rather than assuming it. - -**Public API:** - -- `class PrefixCache` -- A radix tree over token sequences, holding inference states. - ### holographic_shader.py > holographic_shader.py -- N filter passes in ONE evaluation. Two things a GPU structurally cannot do. @@ -28001,6 +28001,48 @@ ### holographic_transform.py +> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). +> +> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's +> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need +> in one place. This gathers the standard, well-known math: +> +> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, +> and what a property panel shows as the T/R/S fields; +> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); +> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector +> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); +> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. +> +> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN +> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit +> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL +> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is +> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. +> Deterministic; NumPy + stdlib only. + +**Public API:** + +- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. +- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). +- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). +- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. +- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear +- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. +- `def quat_normalize(q)` +- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". +- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. +- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. +- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. +- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for +- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). +- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock +- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an +- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. +- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks + +### holographic_transform.py + > TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. > > Everything before this applied leCore's levers uniformly: grow a memory channel @@ -28043,48 +28085,6 @@ - `def plan(weights, cfg, target_tokens, kv_rank, grow_gain)` -- Decide what to do to each layer, from the analysis rather than by rule. - `def apply_plan(weights, cfg, the_plan, progress)` -- Carry out the growth actions. KV compression is a RUNTIME setting and is -### holographic_transform.py - -> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). -> -> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's -> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need -> in one place. This gathers the standard, well-known math: -> -> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, -> and what a property panel shows as the T/R/S fields; -> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); -> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector -> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); -> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. -> -> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN -> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit -> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL -> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is -> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. -> Deterministic; NumPy + stdlib only. - -**Public API:** - -- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. -- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). -- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). -- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. -- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear -- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. -- `def quat_normalize(q)` -- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". -- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. -- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. -- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. -- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for -- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). -- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock -- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an -- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. -- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks - ### holographic_transform_space.py > holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it From c39416fadd97c43775862942da563e3a3b027bfc Mon Sep 17 00:00:00 2001 From: atimics Date: Thu, 13 Aug 2026 14:45:13 -0700 Subject: [PATCH 16/17] results: confirm Qwen v3 Arweave gateway --- .../v3-20c3330d0b3e-execution-failure/publication-receipt.json | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-receipt.json b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-receipt.json index a90224e..7c05847 100644 --- a/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-receipt.json +++ b/experiments/qwen35_acceptance/results/v3-20c3330d0b3e-execution-failure/publication-receipt.json @@ -57,7 +57,7 @@ "transactions_confirmed": 76, "total_winc_charged": 0, "public_cache_verified_at": "2026-08-13T21:33:10Z", - "public_gateway_verified_at": null, + "public_gateway_verified_at": "2026-08-13T21:44:56Z", "verified_downloads": { "publication-manifest.json": "19ac497b7e55881422e28cc8a118085a48e685a8bd5c5e7ee8abe2839ef8f8bb", "ledger/events.jsonl": "ee844391a410ec83c471939d212572b94cc25168597bee3741081335c093ae0f", From 10790736f0a000c0663eca18a2cce5288cd2b0ec Mon Sep 17 00:00:00 2001 From: docs-bot Date: Thu, 13 Aug 2026 21:46:08 +0000 Subject: [PATCH 17/17] docs: refresh generated docs (REFERENCE, CAPABILITIES, API_QUICKREF, PIPELINE_MAP, FACULTY_MAP, DOC_MAP) [skip ci] --- REFERENCE.md | 352 +++++++++++++++++++++++++-------------------------- 1 file changed, 176 insertions(+), 176 deletions(-) diff --git a/REFERENCE.md b/REFERENCE.md index 6669956..ab86e1c 100644 --- a/REFERENCE.md +++ b/REFERENCE.md @@ -76,8 +76,8 @@ | module | what it is | lines | |---|---|---| | [`holographic_session.py`](#holographic-session) | holographic_session.py -- ONE render session that ties the disconnected rendering threads together. | 227 | -| [`holographic_session.py`](#holographic-session) | SESSION -- never compute the same conversation prefix twice. | 253 | | [`holographic_session.py`](#holographic-session) | SESSION STORE -- contexts that outlive the process. | 360 | +| [`holographic_session.py`](#holographic-session) | SESSION -- never compute the same conversation prefix twice. | 253 | ### `splat*` family (6) @@ -423,8 +423,8 @@ | [`holographic_materialio.py`](#holographic-materialio) | PBR material interchange (G2-IO): the standard-format bridge for materials, and a VSA-native carrier. | 307 | | [`holographic_matlib.py`](#holographic-matlib) | holographic_matlib.py -- a comprehensive RENDER material library: plain diffuse -> a fractal planet. | 801 | | [`holographic_meaning_predict.py`](#holographic-meaning-predict) | Generation with structure: predict a next-MEANING vector and settle it, rather | 198 | -| [`holographic_measure.py`](#holographic-measure) | The variance harness: every headline number gets a mean, a spread, and a confidence | 187 | | [`holographic_measure.py`](#holographic-measure) | MEASURE -- perplexity with error bars, and decisions that respect them. | 239 | +| [`holographic_measure.py`](#holographic-measure) | The variance harness: every headline number gets a mean, a spread, and a confidence | 187 | | [`holographic_memory.py`](#holographic-memory) | MEMORY -- the Galvatron's own store, built on leCore's holographic database. | 241 | | [`holographic_memoryhome.py`](#holographic-memoryhome) | holographic_memoryhome.py -- the MEMORY home (consolidation backlog H6): keep the hot working set where the CP | 165 | | [`holographic_memsearch.py`](#holographic-memsearch) | MEMSEARCH -- searchable memory that lives in the weights and runs in the pass. | 212 | @@ -543,8 +543,8 @@ | [`holographic_realtime.py`](#holographic-realtime) | holographic_realtime.py -- the realtime render loop, and the multi-format payload it pushes. | 404 | | [`holographic_reanchor.py`](#holographic-reanchor) | Re-anchoring is load-bearing for deep traversal -- the audit, and the contrast the other tests don't show. | 97 | | [`holographic_reasoning.py`](#holographic-reasoning) | holographic_reasoning.py | 397 | -| [`holographic_recipe.py`](#holographic-recipe) | A generative recipe-store for constructed holostuff structures. | 206 | | [`holographic_recipe.py`](#holographic-recipe) | RECIPE -- ship what leCore ADDED, not the model it was added to. | 334 | +| [`holographic_recipe.py`](#holographic-recipe) | A generative recipe-store for constructed holostuff structures. | 206 | | [`holographic_recipeops.py`](#holographic-recipeops) | StructureRecipe validator + edit operators (ARCH-1): the recipe equivalent of the mesh Euler operators. | 264 | | [`holographic_reclock.py`](#holographic-reclock) | holographic_reclock.py -- sample when an AXIS moves, not when time passes ("make the boring property the | 313 | | [`holographic_recurrent.py`](#holographic-recurrent) | A gradient-free RECURRENT layer for the holographic engine: reservoir computing. | 456 | @@ -578,8 +578,8 @@ | [`holographic_rolling.py`](#holographic-rolling) | holographic_rolling.py -- the CAUSAL rolling / streaming statistics kit: trailing mean, std, min, max, | 378 | | [`holographic_roomacoustic.py`](#holographic-roomacoustic) | holographic_roomacoustic.py -- A6: GEOMETRIC ROOM ACOUSTICS. How a room echoes -- reflections and reverb. | 161 | | [`holographic_route.py`](#holographic-route) | Representation routing (ARCH-7): route each operation to the representation that supports it -- the policy lay | 183 | -| [`holographic_router.py`](#holographic-router) | ROUTER -- the model DECIDING, inside one forward pass. | 229 | | [`holographic_router.py`](#holographic-router) | Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. | 189 | +| [`holographic_router.py`](#holographic-router) | ROUTER -- the model DECIDING, inside one forward pass. | 229 | | [`holographic_sampler.py`](#holographic-sampler) | holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). | 223 | | [`holographic_sampling.py`](#holographic-sampling) | Blue-noise / Poisson-disk point sampling -- the EXCLUSION principle, done right. | 134 | | [`holographic_samplinghome.py`](#holographic-samplinghome) | holographic_samplinghome.py -- the SAMPLING home (consolidation backlog R4): one place for the Monte-Carlo | 127 | @@ -684,8 +684,8 @@ | [`holographic_toolclient.py`](#holographic-toolclient) | holographic_toolclient.py -- call another node the same way leCore is called. | 98 | | [`holographic_topology.py`](#holographic-topology) | Principled topology by persistent homology (EXP-7). | 288 | | [`holographic_transfer.py`](#holographic-transfer) | holographic_transfer.py -- KERNEL SCATTER / GATHER: the ONE bundle/readout under every particle<->grid transfe | 224 | -| [`holographic_transform.py`](#holographic-transform) | holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). | 297 | | [`holographic_transform.py`](#holographic-transform) | TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. | 189 | +| [`holographic_transform.py`](#holographic-transform) | holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). | 297 | | [`holographic_transform_space.py`](#holographic-transform-space) | holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it | 176 | | [`holographic_transformbank.py`](#holographic-transformbank) | holographic_transformbank.py -- a prebuilt map of hypervector transforms, and what it can and cannot hold. | 358 | | [`holographic_transformhome.py`](#holographic-transformhome) | holographic_transformhome.py -- the TRANSFORM home (consolidation backlog H5): one facade over "move / rotate | 223 | @@ -14849,39 +14849,6 @@ ### holographic_measure.py -> The variance harness: every headline number gets a mean, a spread, and a confidence -> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. -> -> WHY this exists -> --------------- -> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's -> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A -> single-seed score is therefore a sample from a distribution, and reporting it alone hides -> how wide that distribution is. For an engine whose entire pitch is "measured, not -> promised," reporting a number without its noise is the sharpest blind spot -- so this -> points that same discipline at the numbers themselves. -> -> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the -> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no -> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI -> bound clears the floor -- which is what stops a single fortunate seed from passing a -> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". -> -> USE REAL DATA. The point of the harness is to characterise the real distribution of a -> real claim; running it on a toy makes the spread meaningless. The measurements wired -> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). - -**Public API:** - -- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. -- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. -- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. -- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is -- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above -- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a - -### holographic_measure.py - > MEASURE -- perplexity with error bars, and decisions that respect them. > > Moose asked what assimilation is actually doing. The answer, from his own run: @@ -14914,6 +14881,39 @@ - `def better_than(a, b, alpha, seed, resamples)` -- Is model A better than model B, or is the difference undecidable? - `def tokens_needed(reference, effect_pct, alpha)` -- How many tokens would be needed to RESOLVE an effect of this size. +### holographic_measure.py + +> The variance harness: every headline number gets a mean, a spread, and a confidence +> interval across seeds -- so a lucky-seed point estimate can't pass as a real result. +> +> WHY this exists +> --------------- +> This whole engine is built on RANDOM vectors. Atoms are random, the RP-tree's +> hyperplanes are random, the reservoir is random, train/test splits are shuffled. A +> single-seed score is therefore a sample from a distribution, and reporting it alone hides +> how wide that distribution is. For an engine whose entire pitch is "measured, not +> promised," reporting a number without its noise is the sharpest blind spot -- so this +> points that same discipline at the numbers themselves. +> +> measure(run_once, seeds) runs a scored experiment once per seed and returns the mean, the +> sample standard deviation, and a 95% percentile-bootstrap confidence interval (no +> distributional assumptions). assert_robust(stats, floor) passes only if the LOWER CI +> bound clears the floor -- which is what stops a single fortunate seed from passing a +> test the typical seed would fail. report() formats "mean +/- std (95% CI [lo, hi], n)". +> +> USE REAL DATA. The point of the harness is to characterise the real distribution of a +> real claim; running it on a toy makes the spread meaningless. The measurements wired +> through it here all run on real corpora (Gutenberg Alice, UDHR, Reuters, Brown). + +**Public API:** + +- `def time_call(fn, repeats, warmup)` -- Time `fn()` honestly: warm up first, then take the MEDIAN of `repeats` timings in microseconds. +- `def measure(run_once, seeds, n_boot, boot_seed)` -- Run a scored experiment across seeds; return mean, std, and a 95% bootstrap CI. +- `def fdr_gate(rows, alpha)` -- P9 -- false-discovery control across a whole ABLATION TABLE, from the measurement home. +- `def assert_robust(stats, floor)` -- Pass only if the LOWER CI bound clears the floor -- not just the mean. This is +- `def is_fragile(stats, margin_floor)` -- A claim is FRAGILE if its spread is large relative to how far its mean sits above +- `def report(name, stats, floor)` -- Format a stats dict as 'name: mean +/- std (95% CI [lo, hi], n)', with a + ### holographic_memory.py > MEMORY -- the Galvatron's own store, built on leCore's holographic database. @@ -21091,42 +21091,6 @@ ### holographic_recipe.py -> A generative recipe-store for constructed holostuff structures. -> -> WHY THIS EXISTS -> --------------- -> The panel's "proven structure has no noise" result: a structure BUILT by a deterministic proof (a -> derivation over a seed) carries no noise, so it serialises to its *generator* losslessly -- you store the -> recipe, not the expanded vectors, and replaying the recipe reproduces the structure BIT-FOR-BIT. This is -> the easy, exact half of generative compression: when we are the builder we already hold the proof, so -> there is nothing to search for and no residual to code. -> -> A `StructureRecipe` is a tiny replayable build-graph. Each op produces one result vector from a seed and -> earlier results: `atom` (a derived atom -- regenerated from the seed, never stored), `bind`, `bundle`, -> `permute`, `normalize`. You build your structure THROUGH the recipe, so you get both the vectors and the -> recipe that regenerates them. Serialising stores only the op list (and the seed) -- a few hundred bytes -> that regenerate megabytes of structure, exactly. -> -> THE ESCAPE HATCH AND THE KEPT NEGATIVE -> The `raw` op stores a literal vector verbatim. It is the honest boundary: data that was NOT constructed -> (a measured or random vector) has no short recipe, so it must be stored as-is and gets no compression. -> The recipe's compression ratio is therefore exactly the *constructed fraction* of the structure -- all -> recipe -> enormous ratio; all raw -> ~1x. That is the constructed-vs-measured partition made literal. -> -> THE CAPACITY-CLIFF POINT -> Reading structure back out of a single bounded encoded vector degrades past the capacity cliff (crosstalk). -> A recipe does not: it names its leaves explicitly and replays the construction, so a deeply nested -> structure is recovered EXACTLY at any depth. The recipe is the right store for deep constructed structure; -> the expanded superposition is bounded. -> -> Pure NumPy + holostuff kernel, deterministic, JSON serialisation (readable), no new dependencies. - -**Public API:** - -- `class StructureRecipe` -- A replayable build-graph: store the deterministic ops that built a structure, not the vectors. - -### holographic_recipe.py - > RECIPE -- ship what leCore ADDED, not the model it was added to. > > Moose: the inflated model size needs looking at holographically -- we should be @@ -21179,6 +21143,42 @@ - `def cost(rules, arrays, installed_weights)` -- What the recipe saves, in bytes. The number is the whole argument. - `def expand(rules, arrays, base_weights)` -- Rebuild the installed model from the recipe. Must be byte-exact. +### holographic_recipe.py + +> A generative recipe-store for constructed holostuff structures. +> +> WHY THIS EXISTS +> --------------- +> The panel's "proven structure has no noise" result: a structure BUILT by a deterministic proof (a +> derivation over a seed) carries no noise, so it serialises to its *generator* losslessly -- you store the +> recipe, not the expanded vectors, and replaying the recipe reproduces the structure BIT-FOR-BIT. This is +> the easy, exact half of generative compression: when we are the builder we already hold the proof, so +> there is nothing to search for and no residual to code. +> +> A `StructureRecipe` is a tiny replayable build-graph. Each op produces one result vector from a seed and +> earlier results: `atom` (a derived atom -- regenerated from the seed, never stored), `bind`, `bundle`, +> `permute`, `normalize`. You build your structure THROUGH the recipe, so you get both the vectors and the +> recipe that regenerates them. Serialising stores only the op list (and the seed) -- a few hundred bytes +> that regenerate megabytes of structure, exactly. +> +> THE ESCAPE HATCH AND THE KEPT NEGATIVE +> The `raw` op stores a literal vector verbatim. It is the honest boundary: data that was NOT constructed +> (a measured or random vector) has no short recipe, so it must be stored as-is and gets no compression. +> The recipe's compression ratio is therefore exactly the *constructed fraction* of the structure -- all +> recipe -> enormous ratio; all raw -> ~1x. That is the constructed-vs-measured partition made literal. +> +> THE CAPACITY-CLIFF POINT +> Reading structure back out of a single bounded encoded vector degrades past the capacity cliff (crosstalk). +> A recipe does not: it names its leaves explicitly and replays the construction, so a deeply nested +> structure is recovered EXACTLY at any depth. The recipe is the right store for deep constructed structure; +> the expanded superposition is bounded. +> +> Pure NumPy + holostuff kernel, deterministic, JSON serialisation (readable), no new dependencies. + +**Public API:** + +- `class StructureRecipe` -- A replayable build-graph: store the deterministic ops that built a structure, not the vectors. + ### holographic_recipeops.py > StructureRecipe validator + edit operators (ARCH-1): the recipe equivalent of the mesh Euler operators. @@ -22712,6 +22712,33 @@ ### holographic_router.py +> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. +> +> WHY THIS EXISTS (measured, backlog N9/N28): +> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, +> but it has no notion of MEANING: "squish a big array down for storage" shares no token with +> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics +> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median +> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. +> +> WHAT SHIPS (and what does NOT): +> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted +> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a +> query it already has a VECTOR for: +> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) +> * a caller that supplies its own query vector (an app that ran the encoder itself) +> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this +> router says so and returns None, so the caller falls back to the token router rather than guessing. That +> honesty is the whole point: silence beats a confident wrong route. +> +> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. + +**Public API:** + +- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. + +### holographic_router.py + > ROUTER -- the model DECIDING, inside one forward pass. > > Moose raised the architecture that dissolves the wall this project kept hitting: @@ -22751,33 +22778,6 @@ - `def route(runtime, router, text, tokenize)` -- Would this model choose to use the capability on this prompt? - `def install_routed(weights, cfg, operator, router, layer, gain, temperature)` -- Install a circuit whose GATE is the router, not a constant. -### holographic_router.py - -> Embedding router -- route a request to the right module by COSINE in nomic's space, not token overlap. -> -> WHY THIS EXISTS (measured, backlog N9/N28): -> The catalog's find_capability scores by shared content words. That is deterministic and needs no model, -> but it has no notion of MEANING: "squish a big array down for storage" shares no token with -> `holographic_coldstore`, and "airspeed velocity of an unladen swallow" confidently matched a physics -> module on the single word "velocity". Measured on the 12-ask suite: token overlap ~2/12 top-1, median -> rank 13 of 503; the nomic embedding router hit 7/12 top-1, median rank 1. -> -> WHAT SHIPS (and what does NOT): -> A 96 KB index -- 503 module vectors at 64d q8, plus the ABTT correction (mu, pc) baked in -- extracted -> from the build cache by tools/semantic/export_index.py. NO model ships. So this router can only score a -> query it already has a VECTOR for: -> * queries embedded at build time and cached (the exam's asks, an app's fixed vocabulary) -> * a caller that supplies its own query vector (an app that ran the encoder itself) -> A brand-new free-text query with no vector and no model present CANNOT be embedded here -- and this -> router says so and returns None, so the caller falls back to the token router rather than guessing. That -> honesty is the whole point: silence beats a confident wrong route. -> -> DETERMINISM: q8 dequant + fixed ABTT transform + argsort with a name tie-break. No RNG, no model. - -**Public API:** - -- `class EmbeddingRouter` -- Loads the shipped q8 index (any dim; 128d is the measured champion) and routes a query VECTOR (or a cached query text) to modules. - ### holographic_sampler.py > holographic_sampler.py -- the SAMPLER: a placeable read-probe (modeling-app backlog, the capstone item). @@ -24408,39 +24408,6 @@ ### holographic_session.py -> SESSION -- never compute the same conversation prefix twice. -> -> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a -> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS -> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens -> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS -> REPEATED, and the fraction grows with every turn. -> -> leCore already had the pieces and never joined them: the runtime exposes -> prefill/step over an InferenceState with copy(), galvacache memoises pure -> functions at a measured 75% hit rate with bit-identical output, and sessions -> exist. What was missing is the RADIX TREE -- the structure that answers "what is -> the longest prefix of this prompt that I have already computed?" -> -> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three -> turns with turn 3, and a dictionary keyed on the whole prompt misses that -> completely. vLLM and SGLang call this RadixAttention; the idea is the same -> whatever the model: index by prefix, resume from the deepest match, compute only -> the tail. -> -> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no -> cache: resuming from a cached state reproduces a full recompute TO FLOAT -> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the -> difference is real rather than pedantic: resuming STEPS the tail while a fresh -> call PREFILLS it, and the two associate their sums differently. The selftest -> asserts the measured bound against the live runtime rather than assuming it. - -**Public API:** - -- `class PrefixCache` -- A radix tree over token sequences, holding inference states. - -### holographic_session.py - > SESSION STORE -- contexts that outlive the process. > > A Galvatron's context is not a transcript, it is STATE: the GDN recurrent @@ -24473,6 +24440,39 @@ - `class SessionStore` -- Named, persistent, independent contexts on disk. - `def runtime_fingerprint(runtime)` -- A cheap, deterministic id for the checkpoint behind a runtime, so a +### holographic_session.py + +> SESSION -- never compute the same conversation prefix twice. +> +> Moose runs a 0.8B on a CPU laptop and it is slow. The single largest waste in a +> conversation is not the model's arithmetic -- it is that every turn RE-PREFILLS +> the entire history. MEASURED on a realistic six-turn exchange: 489 tokens +> processed, of which only 137 were new. SEVENTY-TWO PERCENT OF THE WORK WAS +> REPEATED, and the fraction grows with every turn. +> +> leCore already had the pieces and never joined them: the runtime exposes +> prefill/step over an InferenceState with copy(), galvacache memoises pure +> functions at a measured 75% hit rate with bit-identical output, and sessions +> exist. What was missing is the RADIX TREE -- the structure that answers "what is +> the longest prefix of this prompt that I have already computed?" +> +> WHY A TREE AND NOT A DICTIONARY: turn 4 of a conversation shares its first three +> turns with turn 3, and a dictionary keyed on the whole prompt misses that +> completely. vLLM and SGLang call this RadixAttention; the idea is the same +> whatever the model: index by prefix, resume from the deepest match, compute only +> the tail. +> +> THE GUARANTEE THIS KEEPS, because a cache that changes answers is worse than no +> cache: resuming from a cached state reproduces a full recompute TO FLOAT +> ROUNDING -- measured 7.1e-15, machine epsilon. Not bit-identical, and the +> difference is real rather than pedantic: resuming STEPS the tail while a fresh +> call PREFILLS it, and the two associate their sums differently. The selftest +> asserts the measured bound against the live runtime rather than assuming it. + +**Public API:** + +- `class PrefixCache` -- A radix tree over token sequences, holding inference states. + ### holographic_shader.py > holographic_shader.py -- N filter passes in ONE evaluation. Two things a GPU structurally cannot do. @@ -28001,48 +28001,6 @@ ### holographic_transform.py -> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). -> -> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's -> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need -> in one place. This gathers the standard, well-known math: -> -> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, -> and what a property panel shows as the T/R/S fields; -> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); -> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector -> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); -> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. -> -> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN -> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit -> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL -> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is -> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. -> Deterministic; NumPy + stdlib only. - -**Public API:** - -- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. -- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). -- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). -- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. -- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear -- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. -- `def quat_normalize(q)` -- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". -- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. -- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. -- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. -- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for -- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). -- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock -- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an -- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. -- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks - -### holographic_transform.py - > TRANSFORM -- rebuild a model where the MEASUREMENT says it needs rebuilding. > > Everything before this applied leCore's levers uniformly: grow a memory channel @@ -28085,6 +28043,48 @@ - `def plan(weights, cfg, target_tokens, kv_rank, grow_gain)` -- Decide what to do to each layer, from the analysis rather than by rule. - `def apply_plan(weights, cfg, the_plan, progress)` -- Carry out the growth actions. KV compression is a RUNTIME setting and is +### holographic_transform.py + +> holographic_transform.py -- TRANSFORM UTILITIES for a modeling app (modeling-app backlog, item G). +> +> The engine has scattered transform bits (scenegraph.translation/rotation/compose_transforms; cosserat's +> quaternion helpers; splatexport's rotation<->quaternion), but not the full kit a gizmo and a property panel need +> in one place. This gathers the standard, well-known math: +> +> * decompose(M) -> (translate, rotation-quaternion, scale) -- what a gizmo reads off a matrix to show handles, +> and what a property panel shows as the T/R/S fields; +> * compose_trs(translate, quat, scale) -> 4x4 -- the inverse (build a matrix from panel values); +> * a quaternion kit -- from/to matrix, from/to axis-angle, from/to euler, multiply, SLERP, rotate a vector +> (quaternions are what rotation UI and animation want: no gimbal lock, and slerp gives smooth interpolation); +> * look_at(eye, target, up) -> 4x4 view matrix -- for a camera or an object aimed at a point. +> +> Conventions, stated ONCE and held (the backlog's coordinate-convention note): matrices are 4x4 and act on COLUMN +> vectors, p' = M @ [x, y, z, 1]; compose(A, B) = A @ B means "apply B, then A"; quaternions are (w, x, y, z), unit +> length; euler angles are (rx, ry, rz) applied X then Y then Z, i.e. R = Rz @ Ry @ Rx; look_at returns an OpenGL +> view matrix (the camera looks down -z, y is up) to match the engine's Camera. Nothing here is holographic -- it is +> plain linear algebra a modeling app needs -- so it is kept as a small, readable utility, not dressed up as a bind. +> Deterministic; NumPy + stdlib only. + +**Public API:** + +- `def translation(t)` -- A 4x4 translation matrix from a 3-vector. +- `def scaling(s)` -- A 4x4 scale matrix. `s` is a scalar (uniform) or a 3-vector (per-axis). +- `def rotation_axis_angle(axis, angle)` -- A 4x4 rotation of `angle` radians about `axis` (Rodrigues' formula). +- `def compose(*mats)` -- Matrix product M0 @ M1 @ ... -- with the column-vector convention this applies the RIGHTMOST first. +- `def decompose(M)` -- Split a 4x4 affine transform into (translate (3,), rotation quaternion (4,), scale (3,)). Assumes no shear +- `def compose_trs(translate, quat, scale)` -- Build a 4x4 from translate (3,), a rotation quaternion (4,), and scale (3,) -- the inverse of decompose. +- `def quat_normalize(q)` +- `def quat_mul(a, b)` -- The Hamilton product a*b: the rotation "apply b, then a". +- `def quat_from_axis_angle(axis, angle)` -- A quaternion for a rotation of `angle` radians about `axis`. +- `def quat_to_axis_angle(q)` -- Recover (axis, angle) from a quaternion. +- `def quat_to_matrix(q)` -- The 3x3 rotation matrix for a quaternion. +- `def quat_from_matrix(R)` -- The quaternion for a 3x3 rotation matrix (Shepperd's method: branch on the largest diagonal term for +- `def quat_from_euler(rx, ry, rz)` -- A quaternion from euler angles applied X then Y then Z (R = Rz @ Ry @ Rx). +- `def quat_to_euler(q)` -- Recover euler angles (rx, ry, rz) from a quaternion, inverting R = Rz @ Ry @ Rx. Handles gimbal lock +- `def quat_slerp(a, b, t)` -- Spherical linear interpolation between two rotations -- constant angular speed, the smooth in-between an +- `def quat_rotate(q, v)` -- Rotate a 3-vector by a quaternion. +- `def look_at(eye, target, up)` -- An OpenGL view matrix for a camera at `eye` looking at `target` (the engine's convention: the camera looks + ### holographic_transform_space.py > holographic_transform_space.py -- the TRANSFORM + SPACE model behind a gizmo. A gizmo is a UI; the backend it