Add model cohere-transcribe-arabic-07-2026 - #127
Open
oaattia wants to merge 2 commits into
Open
Conversation
- scripts/convert-cohere.py: arabic variant support - tests/cohere_real_smoke.cpp: real smoke test for arabic - reports/porting/cohere/cohere-transcribe-arabic-07-2026/intake.json: port intake - scripts/hf_cards/cohere-transcribe-arabic-07-2026.yaml: hf card - tests/golden/cohere/cohere-transcribe-arabic-07-2026.manifest.json: golden - tests/tolerances/cohere-transcribe-arabic-07-2026.json: tolerances
…izer type and manifest fields - Swap hf_repo/target_repo in hf_cards yaml to match repo convention - Pin upstream_commit to c3e911b (verified on HF, replaces invalid 6a1364c) - Pin hf_revision in golden manifest to c3e911b (was 'main') - Fix tokenizer type from 'bpe' to 'sentencepiece' in manifest and intake - Populate vocab_size and special_tokens in manifest from intake data
oaattia
marked this pull request as ready for review
August 13, 2026 20:01
Contributor
|
Thanks for submitting, I'll double check as well, I thought I might have already added this to huggingface but I forget |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Add support for
CohereLabs/cohere-transcribe-arabic-07-2026as requested in #86. This is a new variant of the cohere family (Arabic/English, 2B params, Conformer encoder + Transformer encoder-decoder). It reuses the existingsrc/arch/cohere/loader — no C++ changes needed.Scope
reports/porting/cohere/cohere-transcribe-arabic-07-2026/intake.json— new intake (draft)scripts/convert-cohere.py— variant/version derivation + vocab_size fallbackscripts/hf_cards/cohere-transcribe-arabic-07-2026.yaml— new HF card spectests/cohere_real_smoke.cpp— variant-agnostic checkstests/golden/cohere/cohere-transcribe-arabic-07-2026.manifest.json— new golden manifesttests/tolerances/cohere-transcribe-arabic-07-2026.json— new tolerancesAI Assistance
I used AI to generate code and reviewed it myself, so I am responsible for any diffs.
Validation
scripts/ci/clang-format.sh --check— PASSuv run scripts/preflight.py --family cohere --variant cohere-transcribe-arabic-07-2026 --gate A— WARN (expected: no GGUF yet)upstream_commit c3e911bverified on HuggingFace (CohereLabs/cohere-transcribe-arabic-07-2026, 2026-07-13)