rename the repo, drop -private

.devcontainer/postCreateCommand.sh

@@ -4,7 +4,7 @@
 pip install --no-build-isolation --editable .
 
 # Set up git safe directory
-git config --global --add safe.directory /workspaces/nvSubquadratic-private
+git config --global --add safe.directory /workspaces/nvSubquadratic
 
 # Configure bash history search (up/down arrows)
 echo 'Setting up bash history search...'

.github/workflows/gpu-tests.yml

@@ -47,7 +47,7 @@ jobs:
         uses: int128/docker-build-cache-config-action@v1
         id: cache
         with:
-          image: ghcr.io/nvidia-bionemo/nvsubquadratic-private/ci-cache
+          image: ghcr.io/nvidia-bionemo/nvsubquadratic/ci-cache
 
       - name: Build image
         uses: docker/build-push-action@v5
@@ -61,7 +61,7 @@ jobs:
             nvsubquadratic-ci:latest
           cache-from: |
             ${{ steps.cache.outputs.cache-from }}
-            type=registry,ref=ghcr.io/nvidia-bionemo/nvsubquadratic-private/ci-cache:main
+            type=registry,ref=ghcr.io/nvidia-bionemo/nvsubquadratic/ci-cache:main
           cache-to: ${{ steps.cache.outputs.cache-to }}
 
       - name: Run pytest

Dockerfile

@@ -61,7 +61,7 @@ RUN --mount=type=cache,id=apt-cache,target=/var/cache/apt,sharing=locked \
     ninja-build \
     git
 
-WORKDIR /workspaces/nvSubquadratic-private
+WORKDIR /workspaces/nvSubquadratic
 
 # ── Heavy build: Apex from source (cached until apex commit changes) ──────────
 # This layer is intentionally placed before COPY so code changes do not
@@ -81,7 +81,7 @@ RUN pip install --no-cache-dir -r requirements-dev.txt
 # ── Source: invalidated on every code change (fast — just package install) ────
 COPY . .
 
-RUN git config --global --add safe.directory /workspaces/nvSubquadratic-private
+RUN git config --global --add safe.directory /workspaces/nvSubquadratic
 
 RUN pip install --no-cache-dir wheel-stub \
     && pip install --no-cache-dir --no-build-isolation ".[quack]" \
@@ -97,7 +97,7 @@ RUN chown -R ubuntu:ubuntu /workspaces && \
 USER ubuntu
 
 # Set environment variables for development mode
-ENV PYTHONPATH=/workspaces/nvSubquadratic-private
+ENV PYTHONPATH=/workspaces/nvSubquadratic
 
 # Expose Jupyter port
 EXPOSE 8888

README.md

@@ -50,7 +50,7 @@ Open in VS Code and select "Reopen in Container". The devcontainer extension wil
 ```bash
 # Build and run
 docker build -t nvsubquadratic:dev .
-docker run --gpus all -p 8888:8888 -v $(pwd):/workspaces/nvSubquadratic-private nvsubquadratic:dev
+docker run --gpus all -p 8888:8888 -v $(pwd):/workspaces/nvSubquadratic nvsubquadratic:dev
 ```
 
 The Dockerfile builds NVIDIA Apex from source for a broad set of NVIDIA archs by default (`7.0;7.5;8.0;8.6;8.9;9.0;10.0;12.0` — Volta through Blackwell). Two build-args let you tune the compile:
@@ -83,13 +83,13 @@ PLATFORM=arm64 scripts/slurm/enroot/build_sqsh.sh
 apptainer build nvsubquadratic.sif nvsubquadratic.def
 
 # Interactive shell with GPUs and live code from your checkout
-apptainer shell --nv --bind $(pwd):/workspaces/nvSubquadratic-private nvsubquadratic.sif
+apptainer shell --nv --bind $(pwd):/workspaces/nvSubquadratic nvsubquadratic.sif
 
 # Run a command inside the image (example: tests)
-apptainer exec --nv --bind $(pwd):/workspaces/nvSubquadratic-private nvsubquadratic.sif python -m pytest nvsubquadratic/ tests/
+apptainer exec --nv --bind $(pwd):/workspaces/nvSubquadratic nvsubquadratic.sif python -m pytest nvsubquadratic/ tests/
 
 # Use the default runscript (starts Jupyter Lab as defined in the .def)
-apptainer run --nv --bind $(pwd):/workspaces/nvSubquadratic-private nvsubquadratic.sif --no-browser
+apptainer run --nv --bind $(pwd):/workspaces/nvSubquadratic nvsubquadratic.sif --no-browser
 ```
 
 ### Conda (recommended for local development)

docs/benchmarks.md

@@ -2,14 +2,14 @@
 
 Throughput numbers, FLOP scaling, and FP16 op-level results.  The
 tables below are included verbatim from the
-[`benchmarks/README.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/benchmarks/README.md)
+[`benchmarks/README.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/benchmarks/README.md)
 single-source — edits should land there, not here.
 
 ## FLOP scaling
 
 ![FLOP scaling for Hyena / attention / CKConv mixers](_static/flop_scaling.png)
 
-See [`benchmarks/compare_flops.py`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/benchmarks/compare_flops.py)
+See [`benchmarks/compare_flops.py`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/benchmarks/compare_flops.py)
 for the script that produced the plot.
 
 ## ViT-5-Small throughput
@@ -22,7 +22,7 @@ start-after: '# ViT-5-Small Throughput Benchmarks'
 
 ## Op-level results
 
-- [FP16 FFT convolution results](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/benchmarks/ops/FP16_FFTCONV_RESULTS.md)
+- [FP16 FFT convolution results](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/benchmarks/ops/FP16_FFTCONV_RESULTS.md)
   — accuracy and throughput of the FP16 path against the FP32 reference,
   with the dual-mean-centering derivation summarised in
   [FP16 Circular FFT Convolution: Derivation](ops/FP16_FFTCONV_DERIVATION.md).

docs/conf.py

@@ -123,7 +123,7 @@ def _read_version():
     ),
 }
 
-_gh_repo = "https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private"
+_gh_repo = "https://github.com/NVIDIA-BioNeMo/nvSubquadratic"
 _gh_blob_base = f"{_gh_repo}/blob/{github_version}"
 
 extlinks = {
@@ -172,7 +172,7 @@ def _read_version():
 ]
 html_context = {
     "github_user": "NVIDIA-BioNeMo",
-    "github_repo": "nvSubquadratic-private",
+    "github_repo": "nvSubquadratic",
     "github_version": github_version,
     "doc_path": "docs",
 }

docs/examples/index.md

@@ -1,80 +1,80 @@
 # Examples
 
-Each subdirectory of [`examples/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples)
+Each subdirectory of [`examples/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples)
 is a self-contained training recipe.  Recipes are
 {class}`nvsubquadratic.lazy_config.LazyConfig` trees describing the
 network, datamodule, Lightning wrapper, and trainer; running them is
 `python -m experiments.run --config <path>`.
 
 The active experimental roadmap (priorities, owners, status) lives at
-[`examples/overview_tracker.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/examples/overview_tracker.md).
+[`examples/overview_tracker.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/examples/overview_tracker.md).
 
 ## Classification
 
 ### MNIST / SMNIST
 
-[`examples/mnist_classification/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/mnist_classification)
+[`examples/mnist_classification/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/mnist_classification)
 covers MNIST with both attention and Hyena baselines, plus a small CCNN
-backbone.  [`examples/smnist_classification/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/smnist_classification)
+backbone.  [`examples/smnist_classification/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/smnist_classification)
 covers sequential MNIST (1D input).
 
 ### ImageNet
 
-[`examples/imagenet_classification/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/imagenet_classification)
+[`examples/imagenet_classification/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/imagenet_classification)
 ships seven CCNN configs (Hyena / Hyena-circular / attention, with and
 without augmentation, plus tiny variants for laptop sanity checks).
 Representative entry points: `ccnn_7_512_hyena.py`,
 `ccnn_7_512_attention.py`.
 
 ### TinyImageNet — ViT-5
 
-[`examples/vit5_imagenet/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/vit5_imagenet)
+[`examples/vit5_imagenet/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/vit5_imagenet)
 is the ViT-5 baseline suite (v1–v5) with its own
-[`TRACKER.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/examples/vit5_imagenet/TRACKER.md).
+[`TRACKER.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/examples/vit5_imagenet/TRACKER.md).
 
 ### UCF101
 
-[`examples/ucf101_classification/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/ucf101_classification)
+[`examples/ucf101_classification/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/ucf101_classification)
 covers video classification with both sequence- and clip-mode datamodules.
 
 ## Diffusion
 
 ### MNIST
 
-[`examples/mnist_diffusion/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/mnist_diffusion)
+[`examples/mnist_diffusion/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/mnist_diffusion)
 is a small DDPM/JiT diffusion sanity-check.
 
 ### ImageNet
 
-[`examples/imagenet_diffusion/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/imagenet_diffusion)
+[`examples/imagenet_diffusion/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/imagenet_diffusion)
 is the full ImageNet diffusion setup.  See its
-[README](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/examples/imagenet_diffusion/README.md)
+[README](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/examples/imagenet_diffusion/README.md)
 for the JiT vs Hyena-vs-attention comparison.
 
 ## Spatial recall
 
-[`examples/spatial_recall_1d/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/spatial_recall_1d),
-[`spatial_recall_2d/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/spatial_recall_2d),
-[`spatial_recall_3d/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/spatial_recall_3d),
+[`examples/spatial_recall_1d/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/spatial_recall_1d),
+[`spatial_recall_2d/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/spatial_recall_2d),
+[`spatial_recall_3d/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/spatial_recall_3d),
 and the newer
-[`spatial_recall_v2/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/spatial_recall_v2)
+[`spatial_recall_v2/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/spatial_recall_v2)
 are synthetic recall benchmarks measuring how well each mixer (Hyena,
 attention, Mamba, CKConv) routes information across long-range
 spatial/sequence positions.  See
-[`spatial_recall_v2/TRACKER.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/examples/spatial_recall_v2/TRACKER.md)
+[`spatial_recall_v2/TRACKER.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/examples/spatial_recall_v2/TRACKER.md)
 for the v2 task suite.
 
 ## Benchmarks
 
-[`examples/vit_b_benchmark/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/vit_b_benchmark)
+[`examples/vit_b_benchmark/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/vit_b_benchmark)
 holds the throughput-comparison configs used to produce the numbers in
 {doc}`../benchmarks`.
 
 ## Scientific
 
 ### The Well
 
-[`examples/well/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/examples/well)
+[`examples/well/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/examples/well)
 covers The Well PDE benchmark — see its
-[README](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/examples/well/README.md)
+[README](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/examples/well/README.md)
 for sub-datasets and baselines.

docs/getting_started.md

@@ -2,7 +2,7 @@
 
 This page walks a new user from a fresh checkout to a working Hyena
 forward pass.  For the full installation matrix (dev container, Docker,
-Apptainer, conda, venv) see the project [README](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/README.md).
+Apptainer, conda, venv) see the project [README](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/README.md).
 
 ## Requirements
 
@@ -39,7 +39,7 @@ pip install --no-build-isolation -e .
 ```
 
 Docker, Apptainer, enroot/SLURM, and dev-container instructions live in
-the [project README](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/README.md#installation).
+the [project README](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/README.md#installation).
 
 ## Hello, Hyena
 

docs/index.rst

@@ -56,9 +56,9 @@ Where to go next
 Contributor docs
 ----------------
 
-- `CONVENTIONS.md <https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/CONVENTIONS.md>`_ —
+- `CONVENTIONS.md <https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/CONVENTIONS.md>`_ —
   Google-style docstring guide and PR checklist (lives at the repo root).
-- `docs-tracker.md <https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/docs-tracker.md>`_ —
+- `docs-tracker.md <https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/docs-tracker.md>`_ —
   documentation coverage status per file.
 
 Related projects

docs/reports.md

@@ -2,29 +2,29 @@
 
 Long-form technical reports backed by code, figures, and reproducible
 scripts.  Each topic lives in its own subfolder under
-[`reports/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/tree/main/reports)
+[`reports/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/tree/main/reports)
 with one `REPORT.md` plus the supporting `.py` and `.png` files it
 cites.  See the
-[source README](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/reports/README.md)
+[source README](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/reports/README.md)
 for the regeneration conventions (snake_case topic names, scripts take
 `--output-dir`, runs from repo root with `PYTHONPATH=.`).
 
 ## Current topics
 
-| Topic                                                                                                                                                            | Summary                                                                                                                                                           |
-| ---------------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- |
-| [`ckconv_block_diagonal_kernel/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/reports/ckconv_block_diagonal_kernel/REPORT.md)             | Block-diagonal multi-ω₀ SIREN kernel + block-aligned Gaussian mask for ViT-5 hybrid Hyena.  Resolution scaling rule (`ω₀ ← m·ω₀`) verified across 1×/2×/4× grids. |
-| [`siren_omega0_dimensional_scaling/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/reports/siren_omega0_dimensional_scaling/REPORT.md)     | SIREN ω₀ dimensional scaling rule and supporting figures.                                                                                                         |
-| [`spatial_recall/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/reports/spatial_recall/REPORT.md)                                         | Qualitative target-vs-prediction snapshots for the 1D/2D/3D EMNIST spatial-recall task suite (simple copy, mask selection, color selection, color conditioning).  |
-| [`vit5_imagenet_dataloader_profiling/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/reports/vit5_imagenet_dataloader_profiling/REPORT.md) | Feb-2026 investigation that diagnosed the CPU-decode bottleneck on ViT-5-Small ImageNet and motivated the move to the DALI-fused dataloader.                      |
+| Topic                                                                                                                                                    | Summary                                                                                                                                                           |
+| -------------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- |
+| [`ckconv_block_diagonal_kernel/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/reports/ckconv_block_diagonal_kernel/REPORT.md)             | Block-diagonal multi-ω₀ SIREN kernel + block-aligned Gaussian mask for ViT-5 hybrid Hyena.  Resolution scaling rule (`ω₀ ← m·ω₀`) verified across 1×/2×/4× grids. |
+| [`siren_omega0_dimensional_scaling/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/reports/siren_omega0_dimensional_scaling/REPORT.md)     | SIREN ω₀ dimensional scaling rule and supporting figures.                                                                                                         |
+| [`spatial_recall/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/reports/spatial_recall/REPORT.md)                                         | Qualitative target-vs-prediction snapshots for the 1D/2D/3D EMNIST spatial-recall task suite (simple copy, mask selection, color selection, color conditioning).  |
+| [`vit5_imagenet_dataloader_profiling/`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/reports/vit5_imagenet_dataloader_profiling/REPORT.md) | Feb-2026 investigation that diagnosed the CPU-decode bottleneck on ViT-5-Small ImageNet and motivated the move to the DALI-fused dataloader.                      |
 
 ## Adding a new report
 
 1. `mkdir reports/<topic>/` (descriptive snake_case name).
 1. Drop a `REPORT.md` plus any scripts and figures inside; keep
    image links relative to the topic folder.
 1. Add a row to the index table in
-   [`reports/README.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/reports/README.md)
+   [`reports/README.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/reports/README.md)
    and mirror it here.
 1. Re-run every script once before committing so the figures match
    the reported numbers.

docs/repository_overview.md

@@ -9,7 +9,7 @@ tree (`benchmarks/`), and the supporting infrastructure (`scripts/`,
 ## Layout
 
 ```text
-nvSubquadratic-private/
+nvSubquadratic/
 ├── nvsubquadratic/          library code — see "library tree" below
 ├── experiments/             training framework (PyTorch Lightning)
 │   ├── run.py                 CLI entry point
@@ -154,7 +154,7 @@ config tree (LazyConfig dataclasses) that fully describes one
 experiment.  Running it is
 `python -m experiments.run --config examples/.../<config>.py`.  The
 live roadmap is at
-[`examples/overview_tracker.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic-private/blob/main/examples/overview_tracker.md).
+[`examples/overview_tracker.md`](https://github.com/NVIDIA-BioNeMo/nvSubquadratic/blob/main/examples/overview_tracker.md).
 
 **`benchmarks/`** — The single home for performance measurement.
 Op-level microbenchmarks (`benchmarks/ops/`), end-to-end model

examples/vit5_imagenet/v5_patch/README.md

@@ -31,7 +31,7 @@ Attention does not use prepend_registers, so T = 1 + 4 + `num_patches` (no paddi
 ### 1. Clone and install nvSubquadratic
 
 ```bash
-git clone <repo-url> && cd nvSubquadratic-private
+git clone <repo-url> && cd nvSubquadratic
 
 # PyTorch with CUDA 12.8
 pip install torch torchvision --index-url https://download.pytorch.org/whl/cu128

examples/well/v2/active_matter/_base.py

@@ -39,7 +39,7 @@
 SPATIAL_RESOLUTION = (256, 256)
 WELL_BASE_PATH = os.environ.get(
     "WELL_DATA_PATH",
-    "/workspaces/nvSubquadratic-private/data/the_well/datasets/",
+    "/workspaces/nvSubquadratic/data/the_well/datasets/",
 )
 WELL_DATASET_NAME = "active_matter"