AutoGineer
Generated banner for Tao Port Huggingface Model
Generated logo for Tao Port Huggingface Model

Tao Port Huggingface Model

NVIDIAOfficialCreated by NVIDIA Corporation

  • Skill
  • Workflow
  • Software development
  • Apache-2.0
  • Medium risk
Download ZIP

From the author

Integrate a HuggingFace Computer Vision model into the NVIDIA TAO Toolkit ecosystem (tao-core config, tao-pytorch trainer, tao-deploy TensorRT pipeline). Use when the user asks to "integrate a HuggingFace model into TAO", "add an HF model to TAO Toolkit", "wire a HuggingFace ViT/DETR/ SegFormer into tao-pytorch", "build a TAO trainer + deploy pipeline for an HF CV model", or pastes a HuggingFace model URL/ID and wants it turned into a TAO model. Covers the full 7-phase loop: prerequisites check, HuggingFace inspection and validation, codebase exploration, tao-core configuration and native trainer implementation, ONNX export plus TensorRT deploy integration, packaging and L0 testing, container-based end-to-end validation, and (conditional) accuracy/latency tuning. Supports classification, object detection, semantic / instance / panoptic segmentation, zero-shot detection, and depth estimation.
— the description by NVIDIA Corporation

AutoGineer summary

Written by AutoGineer from the skill’s own files, not by the author

This skill provides a structured, seven-phase workflow for porting HuggingFace Computer Vision models into the NVIDIA TAO Toolkit. It guides the user through inspecting the model, configuring the TAO core, implementing the trainer, and setting up the deployment pipeline for TensorRT. The process emphasizes a local-only development loop, requiring the user to manage four independent repository clones and execute all validation, testing, and training tasks within locally prepared TAO Toolkit containers.

What it does

  • Validates HuggingFace model compatibility for computer vision tasks.
  • Configures TAO core and implements native trainers for PyTorch.
  • Integrates ONNX export and TensorRT deployment pipelines.
  • Automates packaging and L0 unit testing for model components.
  • Performs end-to-end validation within containerized environments.
  • Provides optional tuning for accuracy, latency, and model size.

When to use it

  • When you need to bring a custom HuggingFace computer vision model into the TAO Toolkit.
  • When you require a full training-to-deployment pipeline for a new model architecture.
  • When you need to verify model performance and exportability to TensorRT.
  • When you are working in a local environment with NVIDIA GPU support.

Advantages

  • Ensures consistency across training, export, and deployment stages.
  • Provides a rigorous, step-by-step debugging and testing loop.
  • Maintains environment isolation using containerized workflows.
  • Supports a wide range of computer vision tasks including detection and segmentation.

Usage examples

“Integrate a HuggingFace model into TAO.”
The agent initiates the Phase 0 prerequisite check to verify your environment and prepare the necessary TAO Toolkit container images.
“Wire a HuggingFace SegFormer into tao-pytorch.”
The agent begins the codebase exploration phase to map the model to the appropriate TAO reference architecture and prepare the configuration.

Install

  1. Download the skill as a ZIP:

    https://mp-staging.autogineer.com/agent-skills/nvidia/tao-port-huggingface-model/download

  2. Unzip it into the cross-client skills folder of your project:

    mkdir -p .agents/skills && unzip tao-port-huggingface-model.zip -d .agents/skills

Documentation by NVIDIA Corporation

Standalone install? If this session was not initialized by the TAO skill bank plugin, run the tao-setup skill first (host preflight, credentials, cross-skill discovery).

<!-- Copyright (c) 2026, NVIDIA CORPORATION. All rights reserved. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. -->

TAO-HF Integration Skill

Integrate a HuggingFace (HF) Computer Vision model into the NVIDIA TAO Toolkit ecosystem. Work the phases iteratively — not purely linearly — via a build → test → debug → fix → retest loop at every step: when something fails, diagnose and fix before moving on; when it passes, move to the next step.

This SKILL.md is the workflow coordinator. Each phase has a dedicated references/phase-N-*.md with the full step-by-step content, code, docker invocations, and gates. Read the matching reference at the start of each phase — the summaries below are not sufficient.

Local-Only Rule

All work is strictly local. Do NOT push/commit/branch on any remote (GitLab, GitHub, HuggingFace), create merge/pull requests or issues, or upload/publish Docker images to any registry or artifact store. You may only read/clone from remotes — all edits, Docker builds, and test runs stay on the local machine.

Submodule Override Strategy

The user clones the four TAO repos (tao-core, tao-pytorch, tao-deploy, tao-dataservices) independently into one working directory. The tao-core/ submodule nested inside each repo points to the original unmodified commit; modifications only exist in the top-level tao-core/. Always install from the top-level tao-core/, never <repo>/tao-core/ — the nested submodule silently ignores all modifications. Override rules: (1) mount the working directory -v $(pwd):/workspace; (2) pip install /workspace/tao-core FIRST, before tao-pytorch/tao-deploy; (3) PYTHONPATH top-level tao-core first, e.g. -e PYTHONPATH=/workspace/tao-core:/workspace/tao-pytorch. See references/cross-cutting.md for the directory tree.

Execution platform

Every test, smoke run, and end-to-end validation executes inside a locally prepared TAO Toolkit container (tao-pytorch-base:latest, tao-deploy-base:latest, optionally tao-dataservices-base:latest — all from Phase 0). The platform skills own how to run them; this skill specifies what. Default platform: local-docker. Phase 0 delegates the driver / CUDA / NCT preflight to tao-setup-nvidia-gpu-host. See references/cross-cutting.md for the authoritative-skill table, bind-mount rationale, and canonical docker-run flag set.

Phase Map

PhaseGoalReference
0Prerequisites + TAO Toolkit images + local image tagsphase-0-prereqs.md
1Inputs, HF-inspection container, validate model + datasetphase-1-inspection.md, hf-inspection.md
2Closest existing TAO reference modelphase-2-codebase.md, task-type-guide.md
3tao-core config + tao-pytorch trainer / eval / inferencephase-3-implementation.md, tao-patterns.md, repo-structure.md
4ONNX export + tao-deploy TRT engine / inference / evalphase-4-deploy.md
5Packaging (console_scripts) + L0 testsphase-5-packaging.md
6Container testing + end-to-end validationphase-6-container-tests.md, docker-patterns.md
7(conditional) Accuracy / latency / size tuningphase-7-optimization.md

Cross-cutting refs: workflow-consistency.md (CLI flow, config field paths, cross-phase dependencies); cross-cutting.md (platform, isolation, module pitfalls, debugging).

IMPORTANT — Continuous Execution Through Phase 6: do NOT stop after Phases 3–5 to wait for the user to run tests. Phase 6 is mandatory — not complete until tests pass inside the containers and the end-to-end pipeline is validated.

Development Loop

At every step: write code → test immediately (import check, unit test, or dry-run) → if it fails, read traceback → diagnose → fix → retest; if it passes, move on. Do NOT accumulate untested code — testing only at the end compounds bugs.

Debugging Playbook

When something fails, consult the symptom → likely-cause → fix table in references/cross-cutting.md before trying random fixes — it covers ModuleNotFoundError, BACKBONE_REGISTRY KeyError, shape mismatch, NaN loss, ONNX/TRT build failures, TRT-vs-PyTorch accuracy gaps, OOM, DDP hangs, checkpoint load failures, and stale-submodule config issues.

Environment Isolation Strategy

All Python work runs inside Docker containers — no host venvs, no pip installs into host Python (the host needs only Docker, from tao-setup-nvidia-gpu-host). Three contexts: A (Phase 1 HF inspection in tao-hf-inspect, python:3.12-slim fallback), B (Phase 3/4/6 smoke/L0/e2e in the prepared container, source via pip install /workspace/tao-core && python setup.py develop), C (host-bind-mount scratch). See references/cross-cutting.md for the contexts in full and the four numbered rules verbatim (--check-only host packages; Phase 1 --user $(id -u):$(id -g) vs. root; HOME=/workspace/PIP_USER=1 fallback; distro package-manager list; root:root trade-off; tao-hf-inspect cleanup).

Phase 0 — Prerequisites Check

Goal: verify Python 3.10+ and git; delegate the driver / CUDA / Docker / NVIDIA Container Toolkit host check to tao-setup-nvidia-gpu-host; verify NGC docker login for nvcr.io. Then ask the user for the TAO Toolkit image references (tao-pytorch, tao-deploy, optionally tao-dataservices), pull, and prepare local tags tao-pytorch-base:latest, tao-deploy-base:latest, tao-dataservices-base:latest for later phases — preparation removes the pre-installed released TAO packages so the user's /workspace/... clones install/load via pip install /workspace/tao-core && python setup.py develop. Hard stop on any failed check. Required user inputs: the image references + credentials (NGC login, HF_TOKEN). Full commands, prompt wording, and per-image Dockerfile snippets: the Phase 0 reference.

Gate: all prerequisite checks pass; the user supplied the required image references; tao-pytorch-base:latest and tao-deploy-base:latest exist locally; tao-dataservices-base:latest exists if dataservices work is anticipated.

Phase 1 — Information Gathering & Validation

Goal: decide whether to proceed at all. Gather credentials, locate/clone the four TAO repos, create a consistent working branch, launch the tao-hf-inspect container (Context A), validate the HF model is CV with a supported pipeline_tag, extract config + state-dict schema, sanity-check ONNX export, clean up. Full steps: Phase 1 references.

Reject if: pipeline_tag is NLP / audio / LLM (non-CV); AutoConfig raises; or ONNX export fundamentally cannot work (no rewrite path).

Gate: all 4 TAO repos located/cloned with a consistent branch; pipeline_tag confirmed CV; model_type, image_size, hidden_size, num_labels extracted; state-dict keys documented + HF→TAO remapping plan drafted; ONNX export sanity check passed (or failure understood); user confirmed model_short_name + task type. (Full checklist: phase-1-inspection.md.) Present findings and get user confirmation first.

Phase 2 — Codebase Exploration

Goal: find the closest existing TAO reference model for the detected pipeline_tag, read its implementation across tao-core / tao-pytorch / tao-deploy, and decide whether the backbone exists in backbone_v2/ or is new.

The HF pipeline_tag → TAO reference model mapping (classification → classification_pyt, detection → dino/rtdetr, segmentation → segformer, instance → mask2former, panoptic → oneformer, zero-shot → grounding_dino, depth → mono_depth) drives everything downstream (config, architecture, loss, ONNX shape, TRT builder, deploy classes, metrics, dataset format). See the Phase 2 references for the full reference list (12 files per model), the backbone_v2/ and tao-dataservices coverage checks, and per-task architecture.

If a new backbone is needed, decide the strategy (timm wrap > re-implement > HF black-box wrap) before Phase 3 — it changes weight loading, ONNX export, deploy. Never dual-inherit from transformers.PreTrainedModel and BackboneBase (metaclass conflict — compose instead).

Gate: reference TAO model identified + all 12 reference locations read; task-type implications understood (architecture, loss, ONNX outputs, deploy classes, metrics, dataset); backbone coverage decided (reuse / wrap timm / new); dataservices coverage checked. (Full checklist: phase-2-codebase.md.)

Phase 3 — TAO Core Configuration & Native Implementation

Goal: write the tao-core config schema + the tao-pytorch trainer / native inference / evaluation, smoke-testing between steps. (<model_name> = snake_case short-name; <ModelName> = PascalCase.)

Steps 1–7 (each builds on the previous, smoke-test between): tao-core config (1), tao-pytorch trainer (2), multi-GPU/multi-node (3), native inference → result.csv (4), native evaluation → results.json (5), MLOps for training and eval/infer → status.json (6–7). The ExperimentConfig(CommonExperimentConfig) must contain model, dataset, train, evaluate, inference, export, gen_trt_engine, quantize. All ??? fields are MISSING (user supplies via YAML/CLI); the augmentation.mean/std, model.head.in_channels, checkpoint-name, and onnx_file matches are in the checklist below.

Full per-step bodies, code, the canonical experiment_spec.yaml, and smoke-test commands: the Phase 3 references.

Gates: Step 1 — ExperimentConfig imports cleanly in-container; Step 2 — build_model(cfg) runs + PLModel instantiates in-container; Phase 3 — all 7 steps complete, smoke tests pass, no missing __init__.py.

Phase 4 — Export, Deployment & TensorRT Integration

Goal: ONNX export from tao-pytorch, then TRT engine builder + inference + evaluation in tao-deploy reusing the tao-core ExperimentConfig.

Steps 8–11: ONNX exporter (8 — task-specific input/output names, batch_size=-1 ⇒ dynamic batch); TRT engine builder (9 — subclass EngineBuilder or reuse ClassificationEngineBuilder; write specs/{gen_trt_engine,inference,evaluate}.yaml, same ExperimentConfig schema, augmentation.mean/std MUST match training); TRT inference → result.csv (10); TRT eval → results.json (11). See the Phase 4 reference for full code and the Phase 3+4 gate (3 in-container checks: imports, model build + forward, ONNX round-trip).

Module pitfalls: tao-pytorch and tao-deploy have separate hydra_runner and monitor_status — use the deploy versions in deploy scripts. ExperimentConfig comes from nvidia_tao_core in both (same schema/field paths).

Phase 3+4 gate: all three in-container checks pass (tao-pytorch imports + model + ONNX export; tao-deploy imports).

Phase 5 — Packaging & L0 Testing

Goal: register the model as a console_script in both repos and add unit tests.

Steps 12–15: register '<model_name>=...:main' in console_scripts of tao-pytorch/setup.py (12) and tao-deploy/setup.py (13, creating the deploy entrypoint/<model_name>.py via entrypoint_hydra); deploy L0 tests (14); trainer L0 tests — Trainer(..., fast_dev_run=True) + @pytest.mark.cv_unit @pytest.mark.<model_name> (15). See the Phase 5 reference for exact entry-point strings, code, and L0 test file lists.

Gate: entrypoints registered; pytest files exist and follow the marker convention. Do NOT stop — go directly to Phase 6.

Cross-Phase Data Flow & Consistency Verification

Before Docker testing, verify the chain train → export → gen_trt_engine → inference / evaluate (the *_model_latest.pth → .onnx → .engine artifact flow + the config fields each stage reads/writes — full diagram in references/cross-cutting.md).

Consistency checklist (verify before proceeding): self.checkpoint_filename → the *_latest.pth name evaluate.checkpoint / export.checkpoint reference; augmentation.mean/std identical across training spec, inference.yaml, evaluate.yaml, engine-builder preprocess_mode; ONNX input_names=['input'] / output_names=['output'] (detection/instance-seg use task-specific names); export.input_width/input_height match dataset.img_size; model.head.in_channels matches model_params_mapping.py; classes.txt at dataset.root_dir readable by both repos; all __init__.py exist (incl. scripts/__init__.py for get_subtasks() via pkgutil). Full paths: workflow-consistency.md.

Phase 6 — Container Testing & End-to-End Validation

Mandatory — start immediately after Phase 5. All TAO models ship as Docker images; code that only works outside a container is incomplete. Testing runs directly inside the TAO Toolkit container — no image build in the loop: mount → install source (setup.py develop) → run pytest / pylint / pydocstyle / flake8 directly. Use vanilla commands, NOT the ci/run_functional_tests.py / ci/run_static_tests.py wrappers (internal-mirror-only; public github.com/NVIDIA-TAO/ mirrors have no ci/ dir).

Steps 16–25: verify local image tags exist (16); unit tests for tao-core / tao-pytorch (-m cv_unit, --shm-size=16G) / tao-deploy (17–19); lint (20); wheels (21); end-to-end — train dry-run + export in one tao-pytorch session, then gen_trt_engine + inference + evaluate in one tao-deploy session (same session critical — --rm discards installs) (22); cross-check native vs TRT (23); debug shells (24); optional release images (25).

Full commands (every docker run, per-container env-vars, exact pytest / lint invocations + full-suite variants, the train/export/gen_trt_engine/inference/evaluate one-liner with all CLI overrides, the ci/ note, the fix-and-retest loop) and build scripts / runner patterns: see the Phase 6 references.

Phase 6 gate (Done criteria): tao-core / tao-pytorch / tao-deploy unit tests pass in their containers; static tests pass (or only legacy lint warnings); wheels build; end-to-end <model_name>_model_latest.pth → model.onnx → model.engine → non-empty result.csv + results.json; native vs TRT agree within tolerance.

Phase 7 — Optimization & Tuning (conditional)

Enter only if Phase 6 passes but accuracy / latency / size needs improvement. Ask the user for target metrics first.

Diagnostic categories: accuracy too low; TRT-vs-native gap; training too slow; inference too slow. Techniques: Step 27 — hyperparameter tuning; Step 28 — INT8 quantization (PTQ via torchao / modelopt, TRT INT8 + calibration); Step 29 — channel pruning + retrain; Step 30 — knowledge distillation; Step 31 — resolution tuning (TAO interpolates ViT positional embeddings automatically). See the Phase 7 reference for each category's checks, config blocks, YAML overrides, decision tree, and rationale.

Argument

$ARGUMENTS

If provided, interpret $ARGUMENTS as the HuggingFace model ID or URL to start Phase 1. If credentials or model short-name are not included, ask the user for them before proceeding.

Files

references/phase-6-container-tests.md

<!-- Copyright (c) 2026, NVIDIA CORPORATION. All rights reserved. Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. -->

Full Phase 6 walkthrough — TAO Toolkit container inventory, container-based tao-core / tao-pytorch / tao-deploy testing, static (lint) tests, wheel builds, end-to-end pipeline validation (train → export → TRT build → TRT inference / evaluate), native vs TRT cross-check, interactive debug shells, and (optional) release Docker image build.

Phase 6 — Container Testing & End-to-End Validation

Use this reference only when the parent SKILL.md points here for the current task. If this file conflicts with current SKILL.md, skill_info.yaml, schemas, or platform/model skills, the current authoritative source wins.

Contents

  • Phase 6 — Container Testing & End-to-End Validation

    • Local TAO Toolkit Image Tags

    • Step 16 — Verify the Local Image Tags are Ready

    • Step 17 — Test tao-core

    • Step 18 — Test tao-pytorch

    • Step 19 — Test tao-deploy

    • Step 20 — Run Static Tests (Linting)

    • Step 21 — Build Wheels

    • Step 22 — End-to-End Pipeline Validation

    • Step 23 — Cross-Check: Compare Native vs TRT Results

    • Step 24 — Interactive Container for Debugging

    • Step 25 — Build Release Docker Images (Optional)

Mandatory — proceed immediately after Phase 5. Do not wait for user instruction to start this phase. All TAO models ship as Docker images — code that only works outside a container is incomplete. Mount the source into the prepared TAO Toolkit containers, install it, run the tests, and validate the end-to-end pipeline now.

How TAO testing works: TAO testing does NOT build Docker images. Instead, it:

  1. Runs tests directly inside the TAO Toolkit container (started from the local image tag prepared in Phase 0)

  2. Installs tao-core + the target repo at run time via pip install /workspace/tao-core and setup.py develop

  3. Invokes pytest directly on the relevant tests/ subdirectory

  4. Builds wheels inside the same container

  5. Builds release Docker images separately (using release/docker/Dockerfile) only for distribution — NOT for testing

The local flow is therefore: mount the source, pip install /workspace/tao-core, python setup.py develop, run pytest. See docker-patterns.md (sibling reference) for Docker build scripts, runner commands, and related patterns.

Public repos and the ci/ directory: NVIDIA's internal TAO mirrors carry helper scripts under ci/ (e.g. ci/run_functional_tests.py, ci/run_static_tests.py, ci/utils.py) that wrap pytest with testmon, pylint with module discovery, and Docker prefix generation. These scripts are NOT present in the public github mirrors at github.com/NVIDIA-TAO/ — do not invoke them. Use the vanilla pytest + lint commands shown below instead; they produce equivalent output and work on either mirror.

Local TAO Toolkit Image Tags

Phase 0 prepared these local image tags from the TAO Toolkit container references the user supplied. Every docker run command in this phase references the local tag — never the underlying registry image directly — so this table never needs editing per release.

RepoLocal tag (prepared in Phase 0)Underlying TAO Toolkit image (user-supplied)Packages removed during prep
tao-coretao-pytorch-base:latest (or nvcr.io/nvidia/pytorch:24.03-py3)n/a — uses public NGC PyTorch image directly, or piggybacks on the prepared tao-pytorch imagen/a
tao-pytorchtao-pytorch-base:latesttao-pytorch image (e.g. nvcr.io/<org>/tao-toolkit:<version>-pyt)nvidia_tao_pytorch, nvidia_tao_core
tao-deploytao-deploy-base:latesttao-deploy image (e.g. nvcr.io/<org>/tao-toolkit:<version>-deploy)nvidia_tao_deploy, nvidia_tao_core
tao-dataservicestao-dataservices-base:latest (optional)tao-dataservices image (e.g. nvcr.io/<org>/tao-toolkit:<version>-data-services)nvidia_tao_ds, nvidia_tao_pytorch, nvidia_tao_core

Detect host architecture with uname -m (x86_64 → x86, aarch64 → ARM64). The TAO Toolkit images are typically multi-arch manifests, so a single image reference works on both x86 and ARM64 hosts — Docker auto-selects the matching layer.

Step 16 — Verify the Local Image Tags are Ready

The local image tags should already be prepared from Phase 0. Verify they're available and confirm the preparation succeeded:

docker images | grep -E 'tao-pytorch-base|tao-deploy-base|tao-dataservices-base'

# Preparation sanity check — these should all print "not installed"
docker run --rm tao-pytorch-base:latest \
  bash -c "pip show nvidia_tao_pytorch nvidia_tao_core 2>&1 | grep -E '(Name|not installed)'"
docker run --rm tao-deploy-base:latest \
  bash -c "pip show nvidia_tao_deploy nvidia_tao_core 2>&1 | grep -E '(Name|not installed)'"

If any tag is missing or any pre-installed nvidia_tao_* package still shows up, re-run the pull + preparation commands from Phase 0 (phase-0-prereqs.md, sibling reference). If the user has not yet supplied an image reference for one of the components, ask them now — same prompt wording Phase 0 uses.

Step 17 — Test tao-core

Run tao-core tests inside the prepared tao-pytorch container (matching the CI Jenkinsfile.release pattern):

docker run --rm --gpus all \
  -v $(pwd):/workspace \
  -w /workspace/tao-core \
  tao-pytorch-base:latest \
  bash -c "pip install pytest-cov && \
    pip install . && \
    pytest --cov=nvidia_tao_core -v --color=yes"

This validates that our tao-core modifications (new model configs, model_params_mapping, etc.) are correct and importable.

Step 18 — Test tao-pytorch

Install our top-level tao-core (not the empty nested submodule), put the local tao-pytorch source on the path with setup.py develop, then invoke pytest directly. Run only your new model's tests for fast iteration:

docker run --rm --gpus all \
  --shm-size=16G \
  -v $(pwd):/workspace \
  -w /workspace/tao-pytorch \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-pytorch \
  tao-pytorch-base:latest \
  bash -c "pip install /workspace/tao-core && \
    python setup.py develop && \
    pytest tests/cv_unit_test/<model_name>/ -v --color=yes -m 'cv_unit'"

For the full functional suite (used before merging, much slower — equivalent to what NVIDIA's internal ci/run_functional_tests.py wrapper would invoke under the hood):

docker run --rm --gpus all \
  --shm-size=16G \
  -v $(pwd):/workspace \
  -w /workspace/tao-pytorch \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-pytorch \
  tao-pytorch-base:latest \
  bash -c "pip install /workspace/tao-core && \
    python setup.py develop && \
    pytest tests/ -v --color=yes -m 'not slow'"
Step 19 — Test tao-deploy

Same pattern: install our tao-core, install tao-deploy in dev mode, invoke pytest directly on the new model's tests:

docker run --rm --gpus all \
  -v $(pwd):/workspace \
  -w /workspace/tao-deploy \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-deploy \
  tao-deploy-base:latest \
  bash -c "pip install /workspace/tao-core && \
    pip install -e . && \
    pytest tests/<model_name>/ -v --color=yes"

For the full deploy suite:

docker run --rm --gpus all \
  -v $(pwd):/workspace \
  -w /workspace/tao-deploy \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-deploy \
  tao-deploy-base:latest \
  bash -c "pip install /workspace/tao-core && \
    pip install -e . && \
    pytest tests/ -v --color=yes"
Step 20 — Run Static Tests (Linting)

Run pylint (errors-only is fastest), pydocstyle, and flake8 directly on the new model directories. Use --errors-only for the fast path; drop the flag if you want the full report.

# Fast path: pylint --errors-only across the new code in all three repos
docker run --rm \
  -v $(pwd):/workspace \
  -w /workspace \
  tao-pytorch-base:latest \
  bash -c "pip install pylint pydocstyle flake8 && \
    python -m pylint --errors-only \
      tao-pytorch/nvidia_tao_pytorch/cv/<model_name>/ \
      tao-deploy/nvidia_tao_deploy/cv/<model_name>/ \
      tao-core/nvidia_tao_core/config/<model_name>/"

For the fuller report scoped to the tao-pytorch new model only (uses each repo's .pylintrc if present):

docker run --rm \
  -v $(pwd):/workspace \
  -w /workspace/tao-pytorch \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-pytorch \
  tao-pytorch-base:latest \
  bash -c "pip install /workspace/tao-core && python setup.py develop && \
    pip install pylint pydocstyle flake8 && \
    pylint nvidia_tao_pytorch/cv/<model_name> $([ -f .pylintrc ] && echo --rcfile=.pylintrc) && \
    pydocstyle nvidia_tao_pytorch/cv/<model_name> && \
    flake8 nvidia_tao_pytorch/cv/<model_name>"
Step 21 — Build Wheels

Build wheels inside the prepared TAO Toolkit containers to match the exact CUDA/TensorRT versions (same as CI's buildWheel stage):

# tao-pytorch wheel
docker run --rm --gpus all \
  -v $(pwd):/workspace \
  -w /workspace/tao-pytorch \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-pytorch \
  tao-pytorch-base:latest \
  bash -c "pip install /workspace/tao-core && \
    python setup.py bdist_wheel && \
    ls -la dist/*.whl"

# tao-deploy wheel
docker run --rm --gpus all \
  -v $(pwd):/workspace \
  -w /workspace/tao-deploy \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-deploy \
  tao-deploy-base:latest \
  bash -c "pip install /workspace/tao-core && \
    make build && \
    ls -la dist/*.whl"

The wheels are written to tao-pytorch/dist/ and tao-deploy/dist/ on the host (since the workspace is volume-mounted).

Step 22 — End-to-End Pipeline Validation

After all tests pass, run the full pipeline end-to-end to verify the entire train → export → TRT build → TRT inference chain works:

Step 22a+22b — Train dry-run + Export to ONNX (single tao-pytorch container):

Both train and export must run in the same container session because --rm destroys the container (and its installed packages) when it exits.

docker run --rm --gpus all \
  --shm-size=16G \
  -v $(pwd):/workspace \
  -w /workspace/tao-pytorch \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-pytorch \
  tao-pytorch-base:latest \
  bash -c "pip install /workspace/tao-core && python setup.py develop && \
    <model_name> train \
      -e nvidia_tao_pytorch/cv/<model_name>/experiment_specs/experiment_spec.yaml \
      results_dir=/workspace/results \
      train.num_epochs=1 \
      train.num_gpus=1 \
      dataset.batch_size=2 && \
    <model_name> export \
      -e nvidia_tao_pytorch/cv/<model_name>/experiment_specs/experiment_spec.yaml \
      results_dir=/workspace/results \
      export.checkpoint=/workspace/results/train/<model_name>_model_latest.pth \
      export.onnx_file=/workspace/results/export/model.onnx"

Verify on host:

  • results/train/<model_name>_model_latest.pth exists

  • results/export/model.onnx exists

Step 22c+22d+22e — TRT engine build + inference + evaluation (single tao-deploy container):

docker run --rm --gpus all \
  -v $(pwd):/workspace \
  -w /workspace/tao-deploy \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-deploy \
  tao-deploy-base:latest \
  bash -c "pip install /workspace/tao-core && pip install -e . && \
    <model_name> gen_trt_engine \
      -e nvidia_tao_deploy/cv/<model_name>/specs/gen_trt_engine.yaml \
      results_dir=/workspace/results \
      gen_trt_engine.onnx_file=/workspace/results/export/model.onnx \
      gen_trt_engine.trt_engine=/workspace/results/trt/model.engine \
      gen_trt_engine.tensorrt.data_type=FP16 && \
    <model_name> inference \
      -e nvidia_tao_deploy/cv/<model_name>/specs/inference.yaml \
      results_dir=/workspace/results \
      inference.trt_engine=/workspace/results/trt/model.engine \
      dataset.test_dataset.images_dir=<test_images_dir> && \
    <model_name> evaluate \
      -e nvidia_tao_deploy/cv/<model_name>/specs/evaluate.yaml \
      results_dir=/workspace/results \
      evaluate.trt_engine=/workspace/results/trt/model.engine \
      dataset.test_dataset.images_dir=<test_images_dir>"

Verify on host:

  • results/trt/model.engine exists and has non-zero size

  • results/trt_infer/result.csv exists with predictions

  • results/trt_eval/results.json exists with metrics

Step 23 — Cross-Check: Compare Native vs TRT Results

Verify that the TRT-optimized model produces results consistent with the native PyTorch model:

  1. Run native PyTorch inference on the same test images (Step 4)

  2. Run TRT engine inference on the same test images (Step 22d)

  3. Compare predictions: they should match within floating-point tolerance (FP32 ≈ exact, FP16 ≈ small delta)

  4. If results diverge significantly, the ONNX export or TRT engine build has an issue — debug the conversion pipeline

Step 24 — Interactive Container for Debugging

If any step fails and you need an interactive debugging session:

tao-pytorch interactive shell:

docker run -it --rm --gpus all \
  --shm-size=16G \
  -v $(pwd):/workspace \
  -w /workspace/tao-pytorch \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-pytorch \
  tao-pytorch-base:latest \
  /bin/bash

tao-deploy interactive shell:

docker run -it --rm --gpus all \
  -v $(pwd):/workspace \
  -w /workspace/tao-deploy \
  -e PYTHONPATH=/workspace/tao-core:/workspace/tao-deploy \
  tao-deploy-base:latest \
  /bin/bash

Inside the container, you can:

  • pip install /workspace/tao-core && python setup.py develop to install in dev mode

  • python3 -c "import nvidia_tao_pytorch.cv.<model_name>" to test imports

  • python3 -c "from nvidia_tao_pytorch.cv.backbone_v2.registry import BACKBONE_REGISTRY; print(BACKBONE_REGISTRY)" to verify backbone registration

  • Run individual scripts manually with full control over arguments

  • Use pdb for interactive debugging

Step 25 — Build Release Docker Images (Optional)

Only needed for full distribution testing. The release Docker images use release/docker/Dockerfile (different from docker/Dockerfile) and package the wheels built in Step 21.

# tao-pytorch release image
cd tao-pytorch
docker build \
  --network=host \
  -t tao-pytorch-release:latest \
  -f release/docker/Dockerfile \
  .

# tao-deploy release image
cd ../tao-deploy
docker build \
  --network=host \
  -t tao-deploy-release:latest \
  -f release/docker/Dockerfile.release \
  .

These release images bake the wheels into the container. They're what end-users actually run but are NOT needed for the testing workflow above.

Fix-and-retest loop: If any test fails:

  1. Read the full traceback — identify the failing module and line

  2. Fix the code on the host filesystem (mounted volume — changes are live immediately)

  3. Re-run the failing test (no need to rebuild anything — volume mounts pick up changes)

  4. Once the specific test passes, re-run the full suite to check for regressions

Requirements and compatibility

Works with: Claude Code, Claude apps, OpenAI Codex, GitHub Copilot, Cursor, Gemini CLI, Kiro, OpenCode, Goose, Amp, Roo Code, JetBrains Junie, Factory Droid, Devin (ex-Windsurf), Mistral Vibe, OpenHands, Letta

“Works with” means the tool reads the open Agent Skills format; a skill written for one tool may use features others lack.

Needs: Declares allowed tools, Needs Python, Needs Node.js

The author says: Requires Python 3.10+, NVIDIA driver, CUDA 13.0+, docker + nvidia-container-toolkit, an NGC API key (`docker login nvcr.io`), an HF_TOKEN, and access to the TAO Toolkit container images on `nvcr.io` for `tao-pytorch`, `tao-deploy`, and (optionally) `tao-dataservices` — Phase 0 asks the user for the exact image references and prepares them locally as `tao-pytorch-base:latest`, `tao-deploy-base:latest`, `tao-dataservices-base:latest`. Local clones of `tao-core`, `tao-pytorch`, `tao-deploy`,...

Security

What our scanner noticed· Medium risk

  • Large base64-encoded blobs
  • Instructions that use the network

A static scan is not a guarantee. Skills are third-party content: review a skill before you let an agent use it.

Attribution and licence

Created by
NVIDIA Corporation
Publisher
NVIDIA (Official)
Source
NVIDIA/skills · skills/tao-port-huggingface-model · commit fce3cca
Licence
Apache License 2.0 (redistribution allowed)
  • LICENSE (added by AutoGineer from LICENSE-APACHE)
  • ATTRIBUTION.md (written by AutoGineer)
Synced

AutoGineer is not affiliated with or endorsed by NVIDIA Corporation or NVIDIA.

Download ZIP

The ZIP holds the skill’s files, its licence files and an ATTRIBUTION.md. Download tao-port-huggingface-model.zip