Models and Datasets#
We provide recorded HDF5, generated HDF5, LeRobot datasets, and trained models for G1 and H2. Choose the input for the stage you want to run, then follow that stage’s setup instructions. The table lists earlier stages you can skip when starting with each published input. Complete the simulation and model setup required by the stage you select.
Published Input |
Earlier Stages to Skip |
Start Here |
|---|---|---|
Recorded HDF5 (teleoperated recording; conversion-only) |
Recording and generation |
Dataset conversion |
Generated HDF5 (AutoData-amplified; conversion-only) |
Recording and generation |
Dataset conversion |
LeRobot dataset |
Recording, generation and conversion |
Policy training |
Model |
Recording through training |
Policy evaluation |
AutoData generation requires a recording whose saved software versions and simulation settings match this checkout. The published HDF5 files use different settings and are provided for conversion.
Complete Installation first. For H2 conversion,
prepare the GR00T runtime
and keep ISAAC_GR00T_ROOT set in this terminal. H2 simulation and replay
also need Asset Preparation.
Access and Storage#
Use the Hugging Face links below to find the workflow models and datasets. If a download requires authentication, sign in from the installed workflow environment:
uv run --frozen --no-sync hf auth login
uv run --frozen --no-sync hf auth whoami
For training and inference, follow the GR00T N1.7 and Cosmos-Reason2-2B setup instructions on the stage pages. If a download returns 401/403, check the active account and follow the instructions on the repository page.
These commands use immutable revisions recorded in runtime.toml under
artifacts.
HDF5 and LeRobot downloads use the dataset_revision pin. Models use
model_revision. Keep the metadata embedded in each HDF5 file unchanged. It
records software versions, simulation settings, and the source recording for
generated data. Conversion preserves this information.
Download only the dataset or model you need. Choose a new output directory for training so downloaded files and previous runs remain intact.
Published G1 Recordings#
The published 25-demonstration HDF5 recording (779 MB) is not compatible with the current converter’s environment contract. For training, download the already-converted LeRobot dataset instead. To convert your own recordings, follow G1 dataset conversion.
G1 Generated Data for Conversion#
Download 2,500 generated G1 demonstrations (79.2 GB).
Demonstration Dataset for G1 Policy Training#
We provide a training-ready LeRobot dataset of 2,500 generated G1 demonstrations. It is already converted: skip recording, data generation, and conversion and continue with training setup after downloading it.
The same Hugging Face repository also contains source HDF5 files. Those are
separate from the LeRobot dataset used here. The command below downloads only
the converted e2e_ra_g1_apple_to_plate_2500_generated_demos_lerobot directory.
Run it from the workflow repository root.
If you converted your own dataset in the previous module, skip this download and continue to G1 policy training setup below.
Before downloading, complete Hugging Face access setup.
uv run --frozen --no-sync hf download nvidia/GR00T-E2E-Reference-Workflow-SIM-G1-N1.7-AppletoPlate-Dataset \
--revision 36234f4b48dad27f218d8e29ec7710ad5a90b48c \
--repo-type dataset \
--include 'e2e_ra_g1_apple_to_plate_2500_generated_demos_lerobot/**' \
--local-dir outputs/g1/downloads
The dataset is saved in
outputs/g1/downloads/e2e_ra_g1_apple_to_plate_2500_generated_demos_lerobot,
which contains data/, videos/, and meta/. Select Provided dataset in
Training Inputs.
Continue with G1 policy training setup,
starting with GR00T preparation and model access. The demonstrations use G1 in Isaac Lab-Arena and include navigation and
base-height commands; demonstrations recorded with a fixed-base G1 omit these commands.
Trained Policy for G1#
We provide a trained policy for G1 so you can skip recording, data generation, conversion, and training, or check your work against our reference. Run the command below from the workflow repository root to download it, then follow G1 policy evaluation to run it in simulation.
If you want to train your own model, return to G1 policy training.
Before downloading, complete Hugging Face access setup.
uv run --frozen --no-sync hf download nvidia/GR00T-E2E-Reference-Workflow-SIM-G1-N1.7-AppletoPlate-Model \
--revision d3075997eb71095238324746f0507f5d707cd895 \
--exclude optimizer.pt rng_state.pth scheduler.pt trainer_state.json training_args.bin \
--local-dir outputs/g1/published-checkpoint
The model is saved in outputs/g1/published-checkpoint. Keep the
.cache/huggingface/ directory created inside the download directory; the
launcher uses it to check that the model matches the pinned download.
Continue with G1 policy training or evaluation.
Recorded H2 Demonstrations for Conversion#
Download 10 recorded H2 demonstrations (1.18 GB).
H2 Generated Data for Conversion#
Download 17 generated H2 demonstrations (1.89 GB).
Demonstration Dataset for H2 Policy Training#
We provide a dataset of H2 demonstrations so you can skip recording, data generation, and conversion, or check your work against our reference. Run the command below from the workflow repository root to download it.
To use your own converted dataset, skip this download and continue with policy training inputs.
Before downloading, complete Hugging Face access setup.
uv run --frozen --no-sync hf download nvidia/GR00T-E2E-Reference-Workflow-SIM-H2-N1.7-AppleHandoverPlate-Dataset \
--revision 72ad851a5d0b845d368f031c744876613fb0b3c7 \
--repo-type dataset \
--include 'generated/lerobot/**' \
--local-dir outputs/h2/downloads
The dataset is saved in outputs/h2/downloads/generated/lerobot. Continue to
Set the Training Inputs
and select Provided dataset. Use a new H2_RUN_DIR for training outputs.
This dataset contains 1,000 generated episodes; the 10 human seeds are not extra training episodes. It is
self-contained and does not need the raw HDF5 files. To run the
training-input validation,
set H2_TRAINING_DEMOS=1000 and keep the downloaded H2_LEROBOT path.
Trained Policy for H2#
We provide a trained policy for H2 so you can skip recording, data generation, conversion, and training, or check your work against our reference. Run the command below from the workflow repository root to download it, then follow H2 policy evaluation to run it in simulation.
If you want to train your own model, proceed to H2 policy training.
Before downloading, complete Hugging Face access setup.
uv run --frozen --no-sync hf download nvidia/GR00T-E2E-Reference-Workflow-SIM-H2-N1.7-AppleHandoverPlate-Model \
--revision 13397daebd2c34986fc18820374a4a2d33861482 \
--exclude optimizer.pt rng_state.pth scheduler.pt trainer_state.json training_args.bin \
--local-dir outputs/h2/published-checkpoint
The model is saved in outputs/h2/published-checkpoint. Select Provided model
when setting up policy evaluation for H2;
that choice sets the checkpoint path and source for both preflight and server startup.
Keep the .cache/huggingface/ directory created inside the download directory;
the launcher uses it to check that the model matches the pinned download.
The inference download does not need optimizer or trainer-resume state. Model files, weight shards, processor, normalization statistics, embodiment, and action-horizon checks still apply. Complete the simulator and GR00T setup on the evaluation page; no headset is needed.
Continue with H2 policy training or evaluation.