diff --git a/.github/workflows/fast_tests.yml b/.github/workflows/fast_tests.yml index 10ec91199..27a4043e7 100644 --- a/.github/workflows/fast_tests.yml +++ b/.github/workflows/fast_tests.yml @@ -61,6 +61,7 @@ jobs: MUJOCO_GL: egl HF_HOME: /mnt/cache/.cache/huggingface HF_LEROBOT_HOME: /mnt/cache/.cache/huggingface/lerobot + HF_USER_TOKEN: ${{ secrets.LEROBOT_HF_USER }} steps: - uses: actions/checkout@v6 with: @@ -89,5 +90,10 @@ jobs: - name: Install lerobot with test extras run: uv sync --extra "test" + - name: Login to Hugging Face + run: | + uv run hf auth login --token "$HF_USER_TOKEN" --add-to-git-credential + uv run hf auth whoami + - name: Run pytest run: uv run pytest tests -vv --maxfail=10 diff --git a/.github/workflows/full_tests.yml b/.github/workflows/full_tests.yml index fd5e422b3..8dd1fcb1c 100644 --- a/.github/workflows/full_tests.yml +++ b/.github/workflows/full_tests.yml @@ -60,6 +60,7 @@ jobs: MUJOCO_GL: egl HF_HOME: /mnt/cache/.cache/huggingface HF_LEROBOT_HOME: /mnt/cache/.cache/huggingface/lerobot + HF_USER_TOKEN: ${{ secrets.LEROBOT_HF_USER }} steps: - uses: actions/checkout@v6 with: @@ -87,6 +88,11 @@ jobs: - name: Install lerobot with all extras run: uv sync --extra all # TODO(Steven): Make flash-attn optional + - name: Login to Hugging Face + run: | + uv run hf auth login --token "$HF_USER_TOKEN" --add-to-git-credential + uv run hf auth whoami + - name: Run pytest (all extras) run: uv run pytest tests -vv --maxfail=10 @@ -162,6 +168,7 @@ jobs: HF_LEROBOT_HOME: /home/user_lerobot/.cache/huggingface/lerobot TORCH_HOME: /home/user_lerobot/.cache/torch TRITON_CACHE_DIR: /home/user_lerobot/.cache/triton + HF_USER_TOKEN: ${{ secrets.LEROBOT_HF_USER }} container: image: ${{ needs.build-and-push-docker.outputs.image_tag }} # zizmor: ignore[unpinned-images] options: --gpus all --shm-size "16gb" @@ -173,6 +180,12 @@ jobs: shell: bash working-directory: /lerobot steps: + - name: Login to Hugging Face + run: | + hf auth login --token "$HF_USER_TOKEN" --add-to-git-credential + hf auth whoami + - name: Fix ptxas permissions + run: chmod +x /lerobot/.venv/lib/python3.10/site-packages/triton/backends/nvidia/bin/ptxas - name: Run pytest on GPU run: pytest tests -vv --maxfail=10 - name: Run end-to-end tests diff --git a/.github/workflows/nightly.yml b/.github/workflows/nightly.yml index 45bfb9bd5..563b5957d 100644 --- a/.github/workflows/nightly.yml +++ b/.github/workflows/nightly.yml @@ -119,6 +119,7 @@ jobs: HF_LEROBOT_HOME: /home/user_lerobot/.cache/huggingface/lerobot TORCH_HOME: /home/user_lerobot/.cache/torch TRITON_CACHE_DIR: /home/user_lerobot/.cache/triton + HF_USER_TOKEN: ${{ secrets.LEROBOT_HF_USER }} container: image: ${{ needs.build-docker-cpu-nightly.outputs.image_tag }} # zizmor: ignore[unpinned-images] options: --shm-size "16gb" @@ -130,6 +131,10 @@ jobs: shell: bash working-directory: /lerobot steps: + - name: Login to Hugging Face + run: | + hf auth login --token "$HF_USER_TOKEN" --add-to-git-credential + hf auth whoami - name: Run pytest on CPU run: pytest tests -vv --maxfail=10 - name: Run end-to-end tests @@ -146,6 +151,7 @@ jobs: HF_LEROBOT_HOME: /home/user_lerobot/.cache/huggingface/lerobot TORCH_HOME: /home/user_lerobot/.cache/torch TRITON_CACHE_DIR: /home/user_lerobot/.cache/triton + HF_USER_TOKEN: ${{ secrets.LEROBOT_HF_USER }} container: image: ${{ needs.build-docker-gpu-nightly.outputs.image_tag }} # zizmor: ignore[unpinned-images] options: --gpus all --shm-size "16gb" @@ -157,6 +163,10 @@ jobs: shell: bash working-directory: /lerobot steps: + - name: Login to Hugging Face + run: | + hf auth login --token "$HF_USER_TOKEN" --add-to-git-credential + hf auth whoami - name: Run pytest on GPU run: pytest tests -vv --maxfail=10 - name: Run end-to-end tests @@ -174,6 +184,7 @@ jobs: TORCH_HOME: /home/user_lerobot/.cache/torch TRITON_CACHE_DIR: /home/user_lerobot/.cache/triton CUDA_VISIBLE_DEVICES: "0,1,2,3" + HF_USER_TOKEN: ${{ secrets.LEROBOT_HF_USER }} container: image: ${{ needs.build-docker-gpu-nightly.outputs.image_tag }} # zizmor: ignore[unpinned-images] options: --gpus all --shm-size "16gb" @@ -185,6 +196,10 @@ jobs: shell: bash working-directory: /lerobot steps: + - name: Login to Hugging Face + run: | + hf auth login --token "$HF_USER_TOKEN" --add-to-git-credential + hf auth whoami - name: Verify GPU availability run: | nvidia-smi @@ -193,4 +208,3 @@ jobs: - name: Run multi-GPU training tests # TODO(Steven): Investigate why motors tests are failing in multi-GPU setup run: pytest tests -vv --maxfail=10 --ignore=tests/motors/ - timeout-minutes: 10 diff --git a/.github/workflows/unbound_deps_tests.yml b/.github/workflows/unbound_deps_tests.yml index 3f4ea3316..19de38e3b 100644 --- a/.github/workflows/unbound_deps_tests.yml +++ b/.github/workflows/unbound_deps_tests.yml @@ -48,6 +48,7 @@ jobs: MUJOCO_GL: egl HF_HOME: /mnt/cache/.cache/huggingface HF_LEROBOT_HOME: /mnt/cache/.cache/huggingface/lerobot + HF_USER_TOKEN: ${{ secrets.LEROBOT_HF_USER }} steps: - uses: actions/checkout@v6 with: @@ -79,7 +80,10 @@ jobs: - name: Install lerobot with all extras run: uv sync --extra all # TODO(Steven): Make flash-attn optional - + - name: Login to Hugging Face + run: | + uv run hf auth login --token "$HF_USER_TOKEN" --add-to-git-credential + uv run hf auth whoami - name: Run pytest (all extras) run: uv run pytest tests -vv @@ -137,6 +141,7 @@ jobs: HF_LEROBOT_HOME: /home/user_lerobot/.cache/huggingface/lerobot TORCH_HOME: /home/user_lerobot/.cache/torch TRITON_CACHE_DIR: /home/user_lerobot/.cache/triton + HF_USER_TOKEN: ${{ secrets.LEROBOT_HF_USER }} container: image: ${{ needs.build-and-push-docker.outputs.image_tag }} # zizmor: ignore[unpinned-images] options: --gpus all --shm-size "16gb" @@ -148,6 +153,10 @@ jobs: shell: bash working-directory: /lerobot steps: + - name: Login to Hugging Face + run: | + hf auth login --token "$HF_USER_TOKEN" --add-to-git-credential + hf auth whoami - name: Run pytest on GPU run: pytest tests -vv - name: Run end-to-end tests diff --git a/AI_POLICY.md b/AI_POLICY.md new file mode 100644 index 000000000..272ee8c12 --- /dev/null +++ b/AI_POLICY.md @@ -0,0 +1,25 @@ +# AI Usage Policy + +The LeRobot project welcomes contributions from everyone, and we have a few guidelines regarding AI usage to ensure high code quality, clear communication, and a healthy open-source ecosystem: + +- **Please disclose significant AI assistance.** If you used AI tools (e.g., Copilot, Claude, Cursor, ChatGPT) to generate a substantial portion of your code or text, let us know in your PR description. Transparency helps us review your changes more effectively. +- **Own your code (The Human-in-the-Loop).** You must fully understand all the changes you are proposing. If you cannot explain what your AI-assisted code does or how it interacts with LeRobot's broader architecture, please take the time to learn and test it before submitting. +- **Keep issues and discussions focused.** You are welcome to use AI to help draft issues or PR descriptions, but please review and edit them carefully before posting. AI can often be overly verbose; trimming the noise and getting straight to the point helps our maintainers address your needs faster. + +Our core maintainers also use AI tools to aid their workflows, but they do so while bringing deep contextual knowledge of the LeRobot codebase to validate the output. We ask all contributors to apply that same level of rigor. + +## Remember the Human Maintainers + +Please remember that LeRobot is maintained by a dedicated team of humans. + +Every discussion, issue, and pull request is read and reviewed by real people. While AI tools can generate thousands of lines of code in seconds, reviewing that code still takes human time and energy. Submitting unverified or low-effort AI output puts an unfair burden on our maintainers. + +Today, the quality of the AI output still heavily depends on the developer driving the tool. We ask that you respect our maintainers' time by thoroughly vetting, testing, and refining your submissions. + +## AI is Welcome Here + +LeRobot operates at the cutting edge of AI and robotics, and many of our maintainers actively embrace AI coding assistants as valuable productivity tools. We are a pro-AI project! + +Our reason for having an AI policy is not an anti-AI stance. Rather, it exists to ensure that AI is used to enhance human contributions, not replace them with unverified noise. It's about how the tools are used, not the tools themselves. + +We value the unique human insight you bring to the LeRobot community. Let AI empower your workflow, but always let your own judgment take the wheel. diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md index c51a48831..82147d363 100644 --- a/CONTRIBUTING.md +++ b/CONTRIBUTING.md @@ -2,7 +2,7 @@ Everyone is welcome to contribute, and we value everybody's contribution. Code is not the only way to help the community. Answering questions, helping others, reaching out, and improving the documentation are immensely valuable. -Whichever way you choose to contribute, please be mindful to respect our [code of conduct](./CODE_OF_CONDUCT.md). +Whichever way you choose to contribute, please be mindful to respect our [code of conduct](./CODE_OF_CONDUCT.md) and our [AI policy](./AI_POLICY.md). ## Ways to Contribute diff --git a/MANIFEST.in b/MANIFEST.in index c1fb2ea75..c1fce3b5a 100644 --- a/MANIFEST.in +++ b/MANIFEST.in @@ -1,2 +1,3 @@ include src/lerobot/templates/lerobot_modelcard_template.md include src/lerobot/datasets/card_template.md +include src/lerobot/envs/metaworld_config.json diff --git a/docker/Dockerfile.internal b/docker/Dockerfile.internal index c1dfa1dae..ed7d10495 100644 --- a/docker/Dockerfile.internal +++ b/docker/Dockerfile.internal @@ -85,6 +85,8 @@ RUN if [ "$UNBOUND_DEPS" = "true" ]; then \ RUN uv pip install --no-cache ".[all]" +RUN chmod +x /lerobot/.venv/lib/python${PYTHON_VERSION}/site-packages/triton/backends/nvidia/bin/ptxas + # Copy the rest of the application source code # Make sure to have the git-LFS files for testing COPY --chown=user_lerobot:user_lerobot . . diff --git a/docs/source/async.mdx b/docs/source/async.mdx index 3244fc2a3..fcc3f1d1e 100644 --- a/docs/source/async.mdx +++ b/docs/source/async.mdx @@ -48,7 +48,7 @@ python -m lerobot.async_inference.robot_client \ --task="dummy" \ # POLICY: The task to run the policy on (`Fold my t-shirt`). Not necessarily defined for all policies, such as `act` --policy_type=your_policy_type \ # POLICY: the type of policy to run (smolvla, act, etc) --pretrained_name_or_path=user/model \ # POLICY: the model name/path on server to the checkpoint to run (e.g., lerobot/smolvla_base) - --policy_device=mps \ # POLICY: the device to run the policy on, on the server + --policy_device=mps \ # POLICY: the device to run the policy on, on the server (cuda, mps, xpu, cpu) --actions_per_chunk=50 \ # POLICY: the number of actions to output at once --chunk_size_threshold=0.5 \ # CLIENT: the threshold for the chunk size before sending a new observation to the server --aggregate_fn_name=weighted_average \ # CLIENT: the function to aggregate actions on overlapping portions diff --git a/docs/source/earthrover_mini_plus.mdx b/docs/source/earthrover_mini_plus.mdx index cfc3a2eef..37986a7a2 100644 --- a/docs/source/earthrover_mini_plus.mdx +++ b/docs/source/earthrover_mini_plus.mdx @@ -170,13 +170,13 @@ Once you can drive the robot well, you can start recording data to train AI mode We use Hugging Face to store your data online. First, log in with your token from [Hugging Face settings](https://huggingface.co/settings/tokens): ```bash -huggingface-cli login --token ${HUGGINGFACE_TOKEN} --add-to-git-credential +hf auth login --token ${HUGGINGFACE_TOKEN} --add-to-git-credential ``` Store your Hugging Face username: ```bash -HF_USER=$(huggingface-cli whoami | head -n 1) +HF_USER=$(hf auth whoami | awk -F': *' 'NR==1 {print $2}') echo $HF_USER ``` diff --git a/docs/source/envhub.mdx b/docs/source/envhub.mdx index df103d0dd..36c08a8b3 100644 --- a/docs/source/envhub.mdx +++ b/docs/source/envhub.mdx @@ -155,10 +155,10 @@ Upload your repository to Hugging Face: pip install huggingface_hub # Login to Hugging Face -huggingface-cli login +hf auth login # Create a new repository -huggingface-cli repo create my-custom-env --type space --org my-org +hf repo create my-org/my-custom-env # Initialize git and push git init diff --git a/docs/source/il_robots.mdx b/docs/source/il_robots.mdx index bad88f88e..e49132a8e 100644 --- a/docs/source/il_robots.mdx +++ b/docs/source/il_robots.mdx @@ -159,7 +159,7 @@ We use the Hugging Face hub features for uploading your dataset. If you haven't Add your token to the CLI by running this command: ```bash -huggingface-cli login --token ${HUGGINGFACE_TOKEN} --add-to-git-credential +hf auth login --token ${HUGGINGFACE_TOKEN} --add-to-git-credential ``` Then store your Hugging Face repository name in a variable: @@ -327,7 +327,7 @@ You can look for other LeRobot datasets on the hub by searching for `LeRobot` [t You can also push your local dataset to the Hub manually, running: ```bash -huggingface-cli upload ${HF_USER}/record-test ~/.cache/huggingface/lerobot/{repo-id} --repo-type dataset +hf upload ${HF_USER}/record-test ~/.cache/huggingface/lerobot/{repo-id} --repo-type dataset ``` #### Record function @@ -491,7 +491,7 @@ If your local computer doesn't have a powerful GPU you could utilize Google Cola Once training is done, upload the latest checkpoint with: ```bash -huggingface-cli upload ${HF_USER}/act_so101_test \ +hf upload ${HF_USER}/act_so101_test \ outputs/train/act_so101_test/checkpoints/last/pretrained_model ``` @@ -499,7 +499,7 @@ You can also upload intermediate checkpoints with: ```bash CKPT=010000 -huggingface-cli upload ${HF_USER}/act_so101_test${CKPT} \ +hf upload ${HF_USER}/act_so101_test${CKPT} \ outputs/train/act_so101_test/checkpoints/${CKPT}/pretrained_model ``` diff --git a/docs/source/lekiwi.mdx b/docs/source/lekiwi.mdx index b339225d8..7e7c1a680 100644 --- a/docs/source/lekiwi.mdx +++ b/docs/source/lekiwi.mdx @@ -279,13 +279,13 @@ We use the Hugging Face hub features for uploading your dataset. If you haven't Add your token to the CLI by running this command: ```bash -huggingface-cli login --token ${HUGGINGFACE_TOKEN} --add-to-git-credential +hf auth login --token ${HUGGINGFACE_TOKEN} --add-to-git-credential ``` Then store your Hugging Face repository name in a variable: ```bash -HF_USER=$(huggingface-cli whoami | head -n 1) +HF_USER=$(hf auth whoami | awk -F': *' 'NR==1 {print $2}') echo $HF_USER ``` diff --git a/examples/backward_compatibility/replay.py b/examples/backward_compatibility/replay.py index f7c47bec5..13fdfd5f5 100644 --- a/examples/backward_compatibility/replay.py +++ b/examples/backward_compatibility/replay.py @@ -57,7 +57,7 @@ class DatasetReplayConfig: repo_id: str # Episode to replay. episode: int - # Root directory where the dataset will be stored (e.g. 'dataset/path'). + # Root directory where the dataset will be stored (e.g. 'dataset/path'). If None, defaults to $HF_LEROBOT_HOME/repo_id. root: str | Path | None = None # Limit the frames per second. By default, uses the policy fps. fps: int = 30 diff --git a/pyproject.toml b/pyproject.toml index 315507135..9e48f53a5 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -25,7 +25,7 @@ discord = "https://discord.gg/s3KuuzsPFb" [project] name = "lerobot" -version = "0.4.4" +version = "0.4.5" description = "🤗 LeRobot: State-of-the-art Machine Learning for Real-World Robotics in Pytorch" dynamic = ["readme"] license = { text = "Apache-2.0" } @@ -96,9 +96,12 @@ dependencies = [ # Common pygame-dep = ["pygame>=2.5.1,<2.7.0"] placo-dep = ["placo>=0.9.6,<0.10.0"] -transformers-dep = ["transformers>=5.1.0,<6.0.0"] +transformers-dep = ["transformers>=5.3.0,<6.0.0"] grpcio-dep = ["grpcio==1.73.1", "protobuf>=6.31.1,<6.32.0"] can-dep = ["python-can>=4.2.0,<5.0.0"] +peft-dep = ["peft>=0.18.0,<1.0.0"] +scipy-dep = ["scipy>=1.14.0,<2.0.0"] +qwen-vl-utils-dep = ["qwen-vl-utils>=0.0.11,<0.1.0"] # Motors feetech = ["feetech-servo-sdk>=1.0.0,<2.0.0"] @@ -130,17 +133,17 @@ phone = ["hebi-py>=2.8.0,<2.12.0", "teleop>=0.1.0,<0.2.0", "fastapi<1.0"] # Policies wallx = [ "lerobot[transformers-dep]", - "peft>=0.18.0,<1.0.0", - "scipy==1.15.3", # TODO: Relax version - "torchdiffeq==0.2.5", # TODO: Relax version - "qwen-vl-utils==0.0.11" # TODO: Relax version + "lerobot[peft]", + "lerobot[scipy-dep]", + "torchdiffeq>=0.2.4,<0.3.0", + "lerobot[qwen-vl-utils-dep]", ] -pi = ["lerobot[transformers-dep]", "scipy==1.15.3"] # TODO: Relax scipy version +pi = ["lerobot[transformers-dep]", "lerobot[scipy-dep]"] smolvla = ["lerobot[transformers-dep]", "num2words>=0.5.14,<0.6.0", "accelerate>=1.7.0,<2.0.0", "safetensors>=0.4.3,<1.0.0"] multi_task_dit = ["lerobot[transformers-dep]"] groot = [ "lerobot[transformers-dep]", - "peft>=0.13.0,<1.0.0", + "lerobot[peft]", "dm-tree>=0.1.8,<1.0.0", "timm>=1.0.0,<1.1.0", "safetensors>=0.4.3,<1.0.0", @@ -149,13 +152,13 @@ groot = [ "ninja>=1.11.1,<2.0.0", "flash-attn>=2.5.9,<3.0.0 ; sys_platform != 'darwin'" ] -sarm = ["lerobot[transformers-dep]", "faker>=33.0.0,<35.0.0", "matplotlib>=3.10.3,<4.0.0", "qwen-vl-utils>=0.0.11,<0.1.0"] +sarm = ["lerobot[transformers-dep]", "faker>=33.0.0,<35.0.0", "matplotlib>=3.10.3,<4.0.0", "lerobot[qwen-vl-utils-dep]"] xvla = ["lerobot[transformers-dep]"] hilserl = ["lerobot[transformers-dep]", "gym-hil>=0.1.13,<0.2.0", "lerobot[grpcio-dep]", "lerobot[placo-dep]"] # Features async = ["lerobot[grpcio-dep]", "matplotlib>=3.10.3,<4.0.0"] -peft = ["lerobot[transformers-dep]", "peft>=0.18.0,<1.0.0"] +peft = ["lerobot[transformers-dep]", "lerobot[peft-dep]"] # Development dev = ["pre-commit>=3.7.0,<5.0.0", "debugpy>=1.8.1,<1.9.0", "lerobot[grpcio-dep]", "grpcio-tools==1.73.1", "mypy>=1.19.1"] @@ -215,6 +218,9 @@ lerobot-edit-dataset="lerobot.scripts.lerobot_edit_dataset:main" lerobot-setup-can="lerobot.scripts.lerobot_setup_can:main" # ---------------- Tool Configurations ---------------- +[tool.setuptools.package-data] +lerobot = ["envs/*.json"] + [tool.setuptools.packages.find] where = ["src"] diff --git a/src/lerobot/async_inference/robot_client.py b/src/lerobot/async_inference/robot_client.py index e4d21652a..da576eb48 100644 --- a/src/lerobot/async_inference/robot_client.py +++ b/src/lerobot/async_inference/robot_client.py @@ -49,23 +49,18 @@ import torch from lerobot.cameras.opencv.configuration_opencv import OpenCVCameraConfig # noqa: F401 from lerobot.cameras.realsense.configuration_realsense import RealSenseCameraConfig # noqa: F401 -from lerobot.robots import ( # noqa: F401 - Robot, - RobotConfig, - bi_so_follower, - koch_follower, +from lerobot.robots import ( + RobotConfig, # noqa: F401 make_robot_from_config, - omx_follower, - so_follower, ) from lerobot.transport import ( services_pb2, # type: ignore services_pb2_grpc, # type: ignore ) from lerobot.transport.utils import grpc_channel_options, send_bytes_in_chunks +from lerobot.utils.import_utils import register_third_party_plugins from .configs import RobotClientConfig -from .constants import SUPPORTED_ROBOTS from .helpers import ( Action, FPSTracker, @@ -485,8 +480,9 @@ class RobotClient: def async_client(cfg: RobotClientConfig): logging.info(pformat(asdict(cfg))) - if cfg.robot.type not in SUPPORTED_ROBOTS: - raise ValueError(f"Robot {cfg.robot.type} not yet supported!") + # TODO: Assert if checking robot support is still needed with the plugin system + # if cfg.robot.type not in SUPPORTED_ROBOTS: + # raise ValueError(f"Robot {cfg.robot.type} not yet supported!") client = RobotClient(cfg) @@ -512,4 +508,5 @@ def async_client(cfg: RobotClientConfig): if __name__ == "__main__": + register_third_party_plugins() async_client() # run the client diff --git a/src/lerobot/configs/default.py b/src/lerobot/configs/default.py index f613b5251..dcb0cbd54 100644 --- a/src/lerobot/configs/default.py +++ b/src/lerobot/configs/default.py @@ -27,7 +27,7 @@ class DatasetConfig: # "dataset_index" into the returned item. The index mapping is made according to the order in which the # datasets are provided. repo_id: str - # Root directory where the dataset will be stored (e.g. 'dataset/path'). + # Root directory where the dataset will be stored (e.g. 'dataset/path'). If None, defaults to $HF_LEROBOT_HOME/repo_id. root: str | None = None episodes: list[int] | None = None image_transforms: ImageTransformsConfig = field(default_factory=ImageTransformsConfig) diff --git a/src/lerobot/datasets/aggregate.py b/src/lerobot/datasets/aggregate.py index 7020545d2..b32116233 100644 --- a/src/lerobot/datasets/aggregate.py +++ b/src/lerobot/datasets/aggregate.py @@ -289,7 +289,9 @@ def aggregate_datasets( logging.info("Find all tasks") unique_tasks = pd.concat([m.tasks for m in all_metadata]).index.unique() - dst_meta.tasks = pd.DataFrame({"task_index": range(len(unique_tasks))}, index=unique_tasks) + dst_meta.tasks = pd.DataFrame( + {"task_index": range(len(unique_tasks))}, index=pd.Index(unique_tasks, name="task") + ) meta_idx = {"chunk": 0, "file": 0} data_idx = {"chunk": 0, "file": 0} diff --git a/src/lerobot/datasets/card_template.md b/src/lerobot/datasets/card_template.md index ee26a78f5..1eced9f4c 100644 --- a/src/lerobot/datasets/card_template.md +++ b/src/lerobot/datasets/card_template.md @@ -7,6 +7,13 @@ This dataset was created using [LeRobot](https://github.com/huggingface/lerobot). +{% if repo_id is defined and repo_id %} + + + + +{% endif %} + ## Dataset Description {{ dataset_description | default("", true) }} diff --git a/src/lerobot/datasets/dataset_tools.py b/src/lerobot/datasets/dataset_tools.py index 123d455c6..546b3d67f 100644 --- a/src/lerobot/datasets/dataset_tools.py +++ b/src/lerobot/datasets/dataset_tools.py @@ -89,8 +89,8 @@ def delete_episodes( Args: dataset: The source LeRobotDataset. episode_indices: List of episode indices to delete. - output_dir: Directory to save the new dataset. If None, uses default location. - repo_id: Repository ID for the new dataset. If None, appends "_modified" to original. + output_dir: Root directory where the edited dataset will be stored. If not specified, defaults to $HF_LEROBOT_HOME/repo_id. Equivalent to new_root in EditDatasetConfig. + repo_id: Edited dataset identifier. Equivalent to new_repo_id in EditDatasetConfig. """ if not episode_indices: raise ValueError("No episodes to delete") @@ -152,7 +152,7 @@ def split_dataset( dataset: The source LeRobotDataset to split. splits: Either a dict mapping split names to episode indices, or a dict mapping split names to fractions (must sum to <= 1.0). - output_dir: Base directory for output datasets. If None, uses default location. + output_dir: Root directory where the split datasets will be stored. If not specified, defaults to $HF_LEROBOT_HOME/repo_id. Examples: Split by specific episodes @@ -243,8 +243,8 @@ def merge_datasets( Args: datasets: List of LeRobotDatasets to merge. - output_repo_id: Repository ID for the merged dataset. - output_dir: Directory to save the merged dataset. If None, uses default location. + output_repo_id: Merged dataset identifier. + output_dir: Root directory where the merged dataset will be stored. If not specified, defaults to $HF_LEROBOT_HOME/output_repo_id. """ if not datasets: raise ValueError("No datasets to merge") @@ -288,8 +288,8 @@ def modify_features( dataset: The source LeRobotDataset. add_features: Optional dict mapping feature names to (feature_values, feature_info) tuples. remove_features: Optional feature name(s) to remove. Can be a single string or list. - output_dir: Directory to save the new dataset. If None, uses default location. - repo_id: Repository ID for the new dataset. If None, appends "_modified" to original. + output_dir: Root directory where the edited dataset will be stored. If not specified, defaults to $HF_LEROBOT_HOME/repo_id. Equivalent to new_root in EditDatasetConfig. + repo_id: Edited dataset identifier. Equivalent to new_repo_id in EditDatasetConfig. Returns: New dataset with features modified. @@ -390,8 +390,8 @@ def add_features( Args: dataset: The source LeRobotDataset. features: Dictionary mapping feature names to (feature_values, feature_info) tuples. - output_dir: Directory to save the new dataset. If None, uses default location. - repo_id: Repository ID for the new dataset. If None, appends "_modified" to original. + output_dir: Root directory where the edited dataset will be stored. If not specified, defaults to $HF_LEROBOT_HOME/repo_id. Equivalent to new_root in EditDatasetConfig. + repo_id: Edited dataset identifier. Equivalent to new_repo_id in EditDatasetConfig. Returns: New dataset with all features added. @@ -427,8 +427,8 @@ def remove_feature( Args: dataset: The source LeRobotDataset. feature_names: Name(s) of features to remove. Can be a single string or list. - output_dir: Directory to save the new dataset. If None, uses default location. - repo_id: Repository ID for the new dataset. If None, appends "_modified" to original. + output_dir: Root directory where the edited dataset will be stored. If not specified, defaults to $HF_LEROBOT_HOME/repo_id. Equivalent to new_root in EditDatasetConfig. + repo_id: Edited dataset identifier. Equivalent to new_repo_id in EditDatasetConfig. Returns: New dataset with features removed. @@ -567,20 +567,22 @@ def _copy_and_reindex_data( def _keep_episodes_from_video_with_av( input_path: Path, output_path: Path, - episodes_to_keep: list[tuple[float, float]], + episodes_to_keep: list[tuple[int, int]], fps: float, vcodec: str = "libsvtav1", pix_fmt: str = "yuv420p", ) -> None: """Keep only specified episodes from a video file using PyAV. - This function decodes frames from specified time ranges and re-encodes them with + This function decodes frames from specified frame ranges and re-encodes them with properly reset timestamps to ensure monotonic progression. Args: input_path: Source video file path. output_path: Destination video file path. - episodes_to_keep: List of (start_time, end_time) tuples for episodes to keep. + episodes_to_keep: List of (start_frame, end_frame) tuples for episodes to keep. + Ranges are half-open intervals: [start_frame, end_frame), where start_frame + is inclusive and end_frame is exclusive. fps: Frame rate of the video. vcodec: Video codec to use for encoding. pix_fmt: Pixel format for output video. @@ -622,9 +624,10 @@ def _keep_episodes_from_video_with_av( # Create set of (start, end) ranges for fast lookup. # Convert to a sorted list for efficient checking. - time_ranges = sorted(episodes_to_keep) + frame_ranges = sorted(episodes_to_keep) # Track frame index for setting PTS and current range being processed. + src_frame_count = 0 frame_count = 0 range_idx = 0 @@ -634,21 +637,20 @@ def _keep_episodes_from_video_with_av( if frame is None: continue - # Get frame timestamp. - frame_time = float(frame.pts * frame.time_base) if frame.pts is not None else 0.0 - - # Check if frame is in any of our desired time ranges. + # Check if frame is in any of our desired frame ranges. # Skip ranges that have already passed. - while range_idx < len(time_ranges) and frame_time >= time_ranges[range_idx][1]: + while range_idx < len(frame_ranges) and src_frame_count >= frame_ranges[range_idx][1]: range_idx += 1 # If we've passed all ranges, stop processing. - if range_idx >= len(time_ranges): + if range_idx >= len(frame_ranges): break # Check if frame is in current range. - start_ts, end_ts = time_ranges[range_idx] - if frame_time < start_ts: + start_frame = frame_ranges[range_idx][0] + + if src_frame_count < start_frame: + src_frame_count += 1 continue # Frame is in range - create a new frame with reset timestamps. @@ -661,6 +663,7 @@ def _keep_episodes_from_video_with_av( for pkt in v_out.encode(new_frame): out.mux(pkt) + src_frame_count += 1 frame_count += 1 # Flush encoder. @@ -749,15 +752,17 @@ def _copy_and_reindex_videos( f"videos/{video_key}/to_timestamp" ] else: - # Build list of time ranges to keep, in sorted order. + # Build list of frame ranges to keep, in sorted order. sorted_keep_episodes = sorted(episodes_in_file, key=lambda x: episode_mapping[x]) - episodes_to_keep_ranges: list[tuple[float, float]] = [] - + episodes_to_keep_ranges: list[tuple[int, int]] = [] for old_idx in sorted_keep_episodes: src_ep = src_dataset.meta.episodes[old_idx] - from_ts = src_ep[f"videos/{video_key}/from_timestamp"] - to_ts = src_ep[f"videos/{video_key}/to_timestamp"] - episodes_to_keep_ranges.append((from_ts, to_ts)) + from_frame = round(src_ep[f"videos/{video_key}/from_timestamp"] * src_dataset.meta.fps) + to_frame = round(src_ep[f"videos/{video_key}/to_timestamp"] * src_dataset.meta.fps) + assert src_ep["length"] == to_frame - from_frame, ( + f"Episode length mismatch: {src_ep['length']} vs {to_frame - from_frame}" + ) + episodes_to_keep_ranges.append((from_frame, to_frame)) # Use PyAV filters to efficiently re-encode only the desired segments. assert src_dataset.meta.video_path is not None @@ -1470,7 +1475,9 @@ def modify_tasks( # Collect all unique tasks and create new task mapping unique_tasks = sorted(set(episode_to_task.values())) - new_task_df = pd.DataFrame({"task_index": list(range(len(unique_tasks)))}, index=unique_tasks) + new_task_df = pd.DataFrame( + {"task_index": list(range(len(unique_tasks)))}, index=pd.Index(unique_tasks, name="task") + ) task_to_index = {task: idx for idx, task in enumerate(unique_tasks)} logging.info(f"Modifying tasks in {dataset.repo_id}") @@ -1524,7 +1531,7 @@ def modify_tasks( def convert_image_to_video_dataset( dataset: LeRobotDataset, - output_dir: Path, + output_dir: Path | None = None, repo_id: str | None = None, vcodec: str = "libsvtav1", pix_fmt: str = "yuv420p", @@ -1543,8 +1550,8 @@ def convert_image_to_video_dataset( Args: dataset: The source LeRobot dataset with images - output_dir: Directory to save the new video dataset - repo_id: Repository ID for the new dataset (default: original_id + "_video") + output_dir: Root directory where the edited dataset will be stored. If not specified, defaults to $HF_LEROBOT_HOME/repo_id. Equivalent to new_root in EditDatasetConfig. + repo_id: Edited dataset identifier. Equivalent to new_repo_id in EditDatasetConfig. vcodec: Video codec (default: libsvtav1) pix_fmt: Pixel format (default: yuv420p) g: Group of pictures size (default: 2) @@ -1595,6 +1602,7 @@ def convert_image_to_video_dataset( # Video info will be updated after episodes are encoded # Create new metadata for video dataset + output_dir = Path(output_dir) if output_dir is not None else HF_LEROBOT_HOME / repo_id new_meta = LeRobotDatasetMetadata.create( repo_id=repo_id, fps=dataset.meta.fps, diff --git a/src/lerobot/datasets/lerobot_dataset.py b/src/lerobot/datasets/lerobot_dataset.py index 83d452a44..26f0c769c 100644 --- a/src/lerobot/datasets/lerobot_dataset.py +++ b/src/lerobot/datasets/lerobot_dataset.py @@ -314,7 +314,7 @@ class LeRobotDatasetMetadata: if self.tasks is None: new_tasks = tasks task_indices = range(len(tasks)) - self.tasks = pd.DataFrame({"task_index": task_indices}, index=tasks) + self.tasks = pd.DataFrame({"task_index": task_indices}, index=pd.Index(tasks, name="task")) else: new_tasks = [task for task in tasks if task not in self.tasks.index] new_task_indices = range(len(self.tasks), len(self.tasks) + len(new_tasks)) @@ -664,11 +664,11 @@ class LeRobotDataset(torch.utils.data.Dataset): for the README). Args: - repo_id (str): This is the repo id that will be used to fetch the dataset. Locally, the dataset - will be stored under root/repo_id. - root (Path | None, optional): Local directory to use for downloading/writing files. You can also - set the HF_LEROBOT_HOME environment variable to point to a different location. Defaults to - '~/.cache/huggingface/lerobot'. + repo_id (str): This is the repo id that will be used to fetch the dataset. + root (Path | None, optional): Local directory where the dataset will be downloaded and + stored. If set, all dataset files will be stored directly under this path. If not set, the + dataset files will be stored under $HF_LEROBOT_HOME/repo_id (configurable via the + HF_LEROBOT_HOME environment variable). episodes (list[int] | None, optional): If specified, this will only load episodes specified by their episode_index in this list. Defaults to None. image_transforms (Callable | None, optional): You can pass standard v2 image transforms from @@ -747,7 +747,7 @@ class LeRobotDataset(torch.utils.data.Dataset): # Check if cached dataset contains all requested episodes if not self._check_cached_episodes_sufficient(): raise FileNotFoundError("Cached dataset doesn't contain all requested episodes") - except (AssertionError, FileNotFoundError, NotADirectoryError): + except (FileNotFoundError, NotADirectoryError): if is_valid_version(self.revision): self.revision = get_safe_version(self.repo_id, self.revision) self.download(download_videos) @@ -839,7 +839,7 @@ class LeRobotDataset(torch.utils.data.Dataset): hub_api.upload_folder(**upload_kwargs) card = create_lerobot_dataset_card( - tags=tags, dataset_info=self.meta.info, license=license, **card_kwargs + tags=tags, dataset_info=self.meta.info, license=license, repo_id=self.repo_id, **card_kwargs ) card.push_to_hub(repo_id=self.repo_id, repo_type="dataset", revision=branch) @@ -1771,11 +1771,12 @@ class MultiLeRobotDataset(torch.utils.data.Dataset): ) for repo_id, ds in zip(self.repo_ids, self._datasets, strict=True): extra_keys = set(ds.features).difference(intersection_features) - logging.warning( - f"keys {extra_keys} of {repo_id} were disabled as they are not contained in all the " - "other datasets." - ) - self.disabled_features.update(extra_keys) + if extra_keys: + logging.warning( + f"keys {extra_keys} of {repo_id} were disabled as they are not contained in all the " + "other datasets." + ) + self.disabled_features.update(extra_keys) self.image_transforms = image_transforms self.delta_timestamps = delta_timestamps diff --git a/src/lerobot/datasets/utils.py b/src/lerobot/datasets/utils.py index da186bf30..a56740191 100644 --- a/src/lerobot/datasets/utils.py +++ b/src/lerobot/datasets/utils.py @@ -341,6 +341,7 @@ def write_tasks(tasks: pandas.DataFrame, local_dir: Path) -> None: def load_tasks(local_dir: Path) -> pandas.DataFrame: tasks = pd.read_parquet(local_dir / DEFAULT_TASKS_PATH) + tasks.index.name = "task" return tasks diff --git a/src/lerobot/datasets/v30/convert_dataset_v21_to_v30.py b/src/lerobot/datasets/v30/convert_dataset_v21_to_v30.py index 7be37a1b1..81de05686 100644 --- a/src/lerobot/datasets/v30/convert_dataset_v21_to_v30.py +++ b/src/lerobot/datasets/v30/convert_dataset_v21_to_v30.py @@ -36,8 +36,11 @@ Convert a local dataset (works in place): ```bash python src/lerobot/datasets/v30/convert_dataset_v21_to_v30.py \ --repo-id=lerobot/pusht \ - --root=/path/to/local/dataset/directory + --root=/path/to/local/dataset/directory \ --push-to-hub=false + +N.B. Path semantics (v2): --root is the exact dataset folder containing +meta/, data/, videos/. When omitted, defaults to $HF_LEROBOT_HOME/{repo_id}. ``` """ @@ -105,7 +108,7 @@ episodes.jsonl {"episode_index": 1, "tasks": ["Put the blue block in the green bowl"], "length": 266} NEW -meta/episodes/chunk-000/episodes_000.parquet +meta/episodes/chunk-000/file_000.parquet episode_index | video_chunk_index | video_file_index | data_chunk_index | data_file_index | tasks | length ------------------------- OLD @@ -113,15 +116,16 @@ tasks.jsonl {"task_index": 1, "task": "Put the blue block in the green bowl"} NEW -meta/tasks/chunk-000/file_000.parquet +meta/tasks.parquet task_index | task ------------------------- OLD episodes_stats.jsonl +{"episode_index": 1, "stats": {"feature_name": {"min": ..., "max": ..., "mean": ..., "std": ..., "count": ...}}} NEW -meta/episodes_stats/chunk-000/file_000.parquet -episode_index | mean | std | min | max +meta/episodes/chunk-000/file_000.parquet +episode_index | feature_name/min | feature_name/max | feature_name/mean | feature_name/std | feature_name/count ------------------------- UPDATE meta/info.json @@ -170,7 +174,7 @@ def convert_tasks(root, new_root): tasks, _ = legacy_load_tasks(root) task_indices = tasks.keys() task_strings = tasks.values() - df_tasks = pd.DataFrame({"task_index": task_indices}, index=task_strings) + df_tasks = pd.DataFrame({"task_index": task_indices}, index=pd.Index(task_strings, name="task")) write_tasks(df_tasks, new_root) @@ -201,7 +205,6 @@ def convert_data(root: Path, new_root: Path, data_file_size_in_mb: int): image_keys = get_image_keys(root) - ep_idx = 0 chunk_idx = 0 file_idx = 0 size_in_mb = 0 @@ -211,9 +214,23 @@ def convert_data(root: Path, new_root: Path, data_file_size_in_mb: int): logging.info(f"Converting data files from {len(ep_paths)} episodes") - for ep_path in tqdm.tqdm(ep_paths, desc="convert data files"): + for ep_idx, ep_path in enumerate(tqdm.tqdm(ep_paths, desc="convert data files")): ep_size_in_mb = get_parquet_file_size_in_mb(ep_path) ep_num_frames = get_parquet_num_frames(ep_path) + + # Check if we need to start a new file BEFORE creating metadata + if size_in_mb + ep_size_in_mb >= data_file_size_in_mb and len(paths_to_cat) > 0: + # Write the accumulated data files + concat_data_files(paths_to_cat, new_root, chunk_idx, file_idx, image_keys) + + # Move to next file + chunk_idx, file_idx = update_chunk_file_indices(chunk_idx, file_idx, DEFAULT_CHUNK_SIZE) + + # Reset for the next file + size_in_mb = 0 + paths_to_cat = [] + + # Now create metadata with correct chunk/file indices ep_metadata = { "episode_index": ep_idx, "data/chunk_index": chunk_idx, @@ -224,20 +241,7 @@ def convert_data(root: Path, new_root: Path, data_file_size_in_mb: int): size_in_mb += ep_size_in_mb num_frames += ep_num_frames episodes_metadata.append(ep_metadata) - ep_idx += 1 - - if size_in_mb < data_file_size_in_mb: - paths_to_cat.append(ep_path) - continue - - if paths_to_cat: - concat_data_files(paths_to_cat, new_root, chunk_idx, file_idx, image_keys) - - # Reset for the next file - size_in_mb = ep_size_in_mb - paths_to_cat = [ep_path] - - chunk_idx, file_idx = update_chunk_file_indices(chunk_idx, file_idx, DEFAULT_CHUNK_SIZE) + paths_to_cat.append(ep_path) # Write remaining data if any if paths_to_cat: @@ -469,7 +473,7 @@ def convert_dataset( # Set root based on whether local dataset path is provided use_local_dataset = False - root = HF_LEROBOT_HOME / repo_id if root is None else Path(root) / repo_id + root = HF_LEROBOT_HOME / repo_id if root is None else Path(root) if root.exists(): validate_local_dataset_version(root) use_local_dataset = True @@ -553,7 +557,7 @@ if __name__ == "__main__": "--root", type=str, default=None, - help="Local directory to use for downloading/writing the dataset.", + help="Local directory to use for downloading/writing the dataset. Defaults to $HF_LEROBOT_HOME/repo_id.", ) parser.add_argument( "--push-to-hub", diff --git a/src/lerobot/datasets/video_utils.py b/src/lerobot/datasets/video_utils.py index acc24a9e0..8c8494b87 100644 --- a/src/lerobot/datasets/video_utils.py +++ b/src/lerobot/datasets/video_utils.py @@ -227,16 +227,17 @@ def decode_video_frames_torchvision( min_, argmin_ = dist.min(1) is_within_tol = min_ < tolerance_s - assert is_within_tol.all(), ( - f"One or several query timestamps unexpectedly violate the tolerance ({min_[~is_within_tol]} > {tolerance_s=})." - "It means that the closest frame that can be loaded from the video is too far away in time." - "This might be due to synchronization issues with timestamps during data collection." - "To be safe, we advise to ignore this item during training." - f"\nqueried timestamps: {query_ts}" - f"\nloaded timestamps: {loaded_ts}" - f"\nvideo: {video_path}" - f"\nbackend: {backend}" - ) + if not is_within_tol.all(): + raise FrameTimestampError( + f"One or several query timestamps unexpectedly violate the tolerance ({min_[~is_within_tol]} > {tolerance_s=})." + " It means that the closest frame that can be loaded from the video is too far away in time." + " This might be due to synchronization issues with timestamps during data collection." + " To be safe, we advise to ignore this item during training." + f"\nqueried timestamps: {query_ts}" + f"\nloaded timestamps: {loaded_ts}" + f"\nvideo: {video_path}" + f"\nbackend: {backend}" + ) # get closest frames to the query timestamps closest_frames = torch.stack([loaded_frames[idx] for idx in argmin_]) @@ -248,7 +249,11 @@ def decode_video_frames_torchvision( # convert to the pytorch format which is float32 in [0,1] range (and channel first) closest_frames = closest_frames.type(torch.float32) / 255 - assert len(timestamps) == len(closest_frames) + if len(timestamps) != len(closest_frames): + raise FrameTimestampError( + f"Number of retrieved frames ({len(closest_frames)}) does not match " + f"number of queried timestamps ({len(timestamps)})" + ) return closest_frames @@ -353,15 +358,16 @@ def decode_video_frames_torchcodec( min_, argmin_ = dist.min(1) is_within_tol = min_ < tolerance_s - assert is_within_tol.all(), ( - f"One or several query timestamps unexpectedly violate the tolerance ({min_[~is_within_tol]} > {tolerance_s=})." - "It means that the closest frame that can be loaded from the video is too far away in time." - "This might be due to synchronization issues with timestamps during data collection." - "To be safe, we advise to ignore this item during training." - f"\nqueried timestamps: {query_ts}" - f"\nloaded timestamps: {loaded_ts}" - f"\nvideo: {video_path}" - ) + if not is_within_tol.all(): + raise FrameTimestampError( + f"One or several query timestamps unexpectedly violate the tolerance ({min_[~is_within_tol]} > {tolerance_s=})." + " It means that the closest frame that can be loaded from the video is too far away in time." + " This might be due to synchronization issues with timestamps during data collection." + " To be safe, we advise to ignore this item during training." + f"\nqueried timestamps: {query_ts}" + f"\nloaded timestamps: {loaded_ts}" + f"\nvideo: {video_path}" + ) # get closest frames to the query timestamps closest_frames = torch.stack([loaded_frames[idx] for idx in argmin_]) diff --git a/src/lerobot/policies/diffusion/configuration_diffusion.py b/src/lerobot/policies/diffusion/configuration_diffusion.py index 3d30e0941..91b3df214 100644 --- a/src/lerobot/policies/diffusion/configuration_diffusion.py +++ b/src/lerobot/policies/diffusion/configuration_diffusion.py @@ -55,10 +55,16 @@ class DiffusionConfig(PreTrainedConfig): normalization_mapping: A dictionary that maps from a str value of FeatureType (e.g., "STATE", "VISUAL") to a corresponding NormalizationMode (e.g., NormalizationMode.MIN_MAX) vision_backbone: Name of the torchvision resnet backbone to use for encoding images. - crop_shape: (H, W) shape to crop images to as a preprocessing step for the vision backbone. Must fit - within the image size. If None, no cropping is done. - crop_is_random: Whether the crop should be random at training time (it's always a center crop in eval - mode). + resize_shape: (H, W) shape to resize images to as a preprocessing step for the vision + backbone. If None, no resizing is done and the original image resolution is used. + crop_ratio: Ratio in (0, 1] used to derive the crop size from resize_shape + (crop_h = int(resize_shape[0] * crop_ratio), likewise for width). + Set to 1.0 to disable cropping. Only takes effect when resize_shape is not None. + crop_shape: (H, W) shape to crop images to. When resize_shape is set and crop_ratio < 1.0, + this is computed automatically. Can also be set directly for legacy configs that use + crop-only (without resize). If None and no derivation applies, no cropping is done. + crop_is_random: Whether the crop should be random at training time (it's always a center + crop in eval mode). pretrained_backbone_weights: Pretrained weights from torchvision to initialize the backbone. `None` means no pretrained weights. use_group_norm: Whether to replace batch normalization with group normalization in the backbone. @@ -114,7 +120,9 @@ class DiffusionConfig(PreTrainedConfig): # Architecture / modeling. # Vision backbone. vision_backbone: str = "resnet18" - crop_shape: tuple[int, int] | None = (84, 84) + resize_shape: tuple[int, int] | None = None + crop_ratio: float = 1.0 + crop_shape: tuple[int, int] | None = None crop_is_random: bool = True pretrained_backbone_weights: str | None = None use_group_norm: bool = True @@ -175,6 +183,25 @@ class DiffusionConfig(PreTrainedConfig): f"Got {self.noise_scheduler_type}." ) + if self.resize_shape is not None and ( + len(self.resize_shape) != 2 or any(d <= 0 for d in self.resize_shape) + ): + raise ValueError(f"`resize_shape` must be a pair of positive integers. Got {self.resize_shape}.") + if not (0 < self.crop_ratio <= 1.0): + raise ValueError(f"`crop_ratio` must be in (0, 1]. Got {self.crop_ratio}.") + + if self.resize_shape is not None: + if self.crop_ratio < 1.0: + self.crop_shape = ( + int(self.resize_shape[0] * self.crop_ratio), + int(self.resize_shape[1] * self.crop_ratio), + ) + else: + # Explicitly disable cropping for resize+ratio path when crop_ratio == 1.0. + self.crop_shape = None + if self.crop_shape is not None and (self.crop_shape[0] <= 0 or self.crop_shape[1] <= 0): + raise ValueError(f"`crop_shape` must have positive dimensions. Got {self.crop_shape}.") + # Check that the horizon size and U-Net downsampling is compatible. # U-Net downsamples by 2 with each stage. downsampling_factor = 2 ** len(self.down_dims) @@ -202,13 +229,12 @@ class DiffusionConfig(PreTrainedConfig): if len(self.image_features) == 0 and self.env_state_feature is None: raise ValueError("You must provide at least one image or the environment state among the inputs.") - if self.crop_shape is not None: + if self.resize_shape is None and self.crop_shape is not None: for key, image_ft in self.image_features.items(): if self.crop_shape[0] > image_ft.shape[1] or self.crop_shape[1] > image_ft.shape[2]: raise ValueError( - f"`crop_shape` should fit within the images shapes. Got {self.crop_shape} " - f"for `crop_shape` and {image_ft.shape} for " - f"`{key}`." + f"`crop_shape` should fit within the image shapes. Got {self.crop_shape} " + f"for `crop_shape` and {image_ft.shape} for `{key}`." ) # Check that all input images have the same shape. diff --git a/src/lerobot/policies/diffusion/modeling_diffusion.py b/src/lerobot/policies/diffusion/modeling_diffusion.py index 7525c9252..aa8d5dd14 100644 --- a/src/lerobot/policies/diffusion/modeling_diffusion.py +++ b/src/lerobot/policies/diffusion/modeling_diffusion.py @@ -142,6 +142,9 @@ class DiffusionPolicy(PreTrainedPolicy): """Run the batch through the model and compute the loss for training or validation.""" if self.config.image_features: batch = dict(batch) # shallow copy so that adding a key doesn't modify the original + for key in self.config.image_features: + if self.config.n_obs_steps == 1 and batch[key].ndim == 4: + batch[key] = batch[key].unsqueeze(1) batch[OBS_IMAGES] = torch.stack([batch[key] for key in self.config.image_features], dim=-4) loss = self.diffusion.compute_loss(batch) # no output_dict so returning None @@ -451,12 +454,18 @@ class DiffusionRgbEncoder(nn.Module): def __init__(self, config: DiffusionConfig): super().__init__() # Set up optional preprocessing. - if config.crop_shape is not None: + if config.resize_shape is not None: + self.resize = torchvision.transforms.Resize(config.resize_shape) + else: + self.resize = None + + crop_shape = config.crop_shape + if crop_shape is not None: self.do_crop = True # Always use center crop for eval - self.center_crop = torchvision.transforms.CenterCrop(config.crop_shape) + self.center_crop = torchvision.transforms.CenterCrop(crop_shape) if config.crop_is_random: - self.maybe_random_crop = torchvision.transforms.RandomCrop(config.crop_shape) + self.maybe_random_crop = torchvision.transforms.RandomCrop(crop_shape) else: self.maybe_random_crop = self.center_crop else: @@ -482,13 +491,16 @@ class DiffusionRgbEncoder(nn.Module): # Set up pooling and final layers. # Use a dry run to get the feature map shape. - # The dummy input should take the number of image channels from `config.image_features` and it should - # use the height and width from `config.crop_shape` if it is provided, otherwise it should use the - # height and width from `config.image_features`. + # The dummy shape mirrors the runtime preprocessing order: resize -> crop. # Note: we have a check in the config class to make sure all images have the same shape. images_shape = next(iter(config.image_features.values())).shape - dummy_shape_h_w = config.crop_shape if config.crop_shape is not None else images_shape[1:] + if config.crop_shape is not None: + dummy_shape_h_w = config.crop_shape + elif config.resize_shape is not None: + dummy_shape_h_w = config.resize_shape + else: + dummy_shape_h_w = images_shape[1:] dummy_shape = (1, images_shape[0], *dummy_shape_h_w) feature_map_shape = get_output_shape(self.backbone, dummy_shape)[1:] @@ -504,7 +516,10 @@ class DiffusionRgbEncoder(nn.Module): Returns: (B, D) image feature. """ - # Preprocess: maybe crop (if it was set up in the __init__). + # Preprocess: resize if configured, then crop if configured. + + if self.resize is not None: + x = self.resize(x) if self.do_crop: if self.training: # noqa: SIM108 x = self.maybe_random_crop(x) diff --git a/src/lerobot/policies/pi05/modeling_pi05.py b/src/lerobot/policies/pi05/modeling_pi05.py index 4a74250a0..dc5eb20ec 100644 --- a/src/lerobot/policies/pi05/modeling_pi05.py +++ b/src/lerobot/policies/pi05/modeling_pi05.py @@ -99,10 +99,11 @@ def create_sinusoidal_pos_embedding( # see openpi `create_sinusoidal_pos_embedd def sample_beta(alpha, beta, bsize, device): # see openpi `sample_beta` (exact copy) - alpha_t = torch.as_tensor(alpha, dtype=torch.float32, device=device) - beta_t = torch.as_tensor(beta, dtype=torch.float32, device=device) + # Beta sampling uses _sample_dirichlet which isn't implemented for MPS, so sample on CPU + alpha_t = torch.tensor(alpha, dtype=torch.float32) + beta_t = torch.tensor(beta, dtype=torch.float32) dist = torch.distributions.Beta(alpha_t, beta_t) - return dist.sample((bsize,)) + return dist.sample((bsize,)).to(device) def make_att_2d_masks(pad_masks, att_masks): # see openpi `make_att_2d_masks` (exact copy) diff --git a/src/lerobot/policies/pi_gemma.py b/src/lerobot/policies/pi_gemma.py index 35a6ae0d2..05f031d08 100644 --- a/src/lerobot/policies/pi_gemma.py +++ b/src/lerobot/policies/pi_gemma.py @@ -260,7 +260,7 @@ class PiGemmaModel(GemmaModel): # type: ignore[misc] causal_mask = create_causal_mask( config=self.config, - input_embeds=inputs_embeds, + inputs_embeds=inputs_embeds, attention_mask=attention_mask, cache_position=cache_position, past_key_values=past_key_values, diff --git a/src/lerobot/policies/sarm/processor_sarm.py b/src/lerobot/policies/sarm/processor_sarm.py index 5c617282a..8f2bc23db 100644 --- a/src/lerobot/policies/sarm/processor_sarm.py +++ b/src/lerobot/policies/sarm/processor_sarm.py @@ -277,9 +277,7 @@ class SARMEncodingProcessorStep(ProcessorStep): # When language is perturbed, targets are zero so perturbed samples don't contribute to progress loss if self.dataset_meta is not None: - episodes_df = None - if self.sparse_subtask_names != ["task"]: - episodes_df = self.dataset_meta.episodes.to_pandas() + episodes_df = self.dataset_meta.episodes.to_pandas() # Generate sparse targets if self.sparse_temporal_proportions is not None: diff --git a/src/lerobot/policies/smolvla/configuration_smolvla.py b/src/lerobot/policies/smolvla/configuration_smolvla.py index c696265f2..b861b856b 100644 --- a/src/lerobot/policies/smolvla/configuration_smolvla.py +++ b/src/lerobot/policies/smolvla/configuration_smolvla.py @@ -106,6 +106,9 @@ class SmolVLAConfig(PreTrainedConfig): # Real-Time Chunking (RTC) configuration rtc_config: RTCConfig | None = None + compile_model: bool = False # Whether to use torch.compile for model optimization + compile_mode: str = "max-autotune" # Torch compile mode + def __post_init__(self): super().__post_init__() diff --git a/src/lerobot/policies/smolvla/modeling_smolvla.py b/src/lerobot/policies/smolvla/modeling_smolvla.py index 10544a949..e49226d26 100644 --- a/src/lerobot/policies/smolvla/modeling_smolvla.py +++ b/src/lerobot/policies/smolvla/modeling_smolvla.py @@ -593,6 +593,12 @@ class VLAFlowMatching(nn.Module): self.prefix_length = self.config.prefix_length self.rtc_processor = rtc_processor + # Compile model if requested + if config.compile_model: + torch.set_float32_matmul_precision("high") + self.sample_actions = torch.compile(self.sample_actions, mode=config.compile_mode) + self.forward = torch.compile(self.forward, mode=config.compile_mode) + def _rtc_enabled(self): return self.config.rtc_config is not None and self.config.rtc_config.enabled diff --git a/src/lerobot/policies/smolvla/smolvlm_with_expert.py b/src/lerobot/policies/smolvla/smolvlm_with_expert.py index 555c40773..caca41dab 100644 --- a/src/lerobot/policies/smolvla/smolvlm_with_expert.py +++ b/src/lerobot/policies/smolvla/smolvlm_with_expert.py @@ -77,7 +77,6 @@ class SmolVLMWithExpertModel(nn.Module): print(f"Loading {model_id} weights ...") self.vlm = AutoModelForImageTextToText.from_pretrained( model_id, - device_map=device, torch_dtype="bfloat16", low_cpu_mem_usage=True, ) diff --git a/src/lerobot/policies/wall_x/configuration_wall_x.py b/src/lerobot/policies/wall_x/configuration_wall_x.py index 3962b56f6..5269c4e10 100644 --- a/src/lerobot/policies/wall_x/configuration_wall_x.py +++ b/src/lerobot/policies/wall_x/configuration_wall_x.py @@ -55,7 +55,7 @@ class WallXConfig(PreTrainedConfig): pretrained_name_or_path: str = "x-square-robot/wall-oss-flow" # Tokenizer settings - action_tokenizer_path: str | None = "physical-intelligence/fast" + action_tokenizer_path: str | None = "lerobot/fast-action-tokenizer" # Action prediction mode: "diffusion" or "fast" prediction_mode: str = "diffusion" diff --git a/src/lerobot/policies/wall_x/modeling_wall_x.py b/src/lerobot/policies/wall_x/modeling_wall_x.py index 36f896998..84ee05743 100644 --- a/src/lerobot/policies/wall_x/modeling_wall_x.py +++ b/src/lerobot/policies/wall_x/modeling_wall_x.py @@ -261,10 +261,15 @@ class Qwen2_5_VLMoEForAction(Qwen2_5_VLForConditionalGeneration): and optional LoRA fine-tuning support. """ - _tied_weights_keys = ["lm_head.weight"] + _tied_weights_keys = {"lm_head.weight": "model.embed_tokens.weight"} config_class = Qwen2_5_VLConfig _no_split_modules = ["Qwen2_5_VLDecoderLayer_with_MoE", "Qwen2_5_VLVisionBlock"] + def init_weights(self): + if getattr(self.model, "language_model", None) is not None: + return + super().init_weights() + @classmethod def from_pretrained( cls, @@ -312,6 +317,11 @@ class Qwen2_5_VLMoEForAction(Qwen2_5_VLForConditionalGeneration): processor.action_processor = action_tokenizer else: action_tokenizer = None + + # add pad_token_id to config + config.pad_token_id = processor.tokenizer.pad_token_id + config.text_config.pad_token_id = processor.tokenizer.pad_token_id + # Initialize model with configuration and processor model = cls(config, processor=processor, action_tokenizer=action_tokenizer, **kwargs) diff --git a/src/lerobot/policies/wall_x/qwen_model/configuration_qwen2_5_vl.py b/src/lerobot/policies/wall_x/qwen_model/configuration_qwen2_5_vl.py index 731ef3b3e..19874b6ff 100644 --- a/src/lerobot/policies/wall_x/qwen_model/configuration_qwen2_5_vl.py +++ b/src/lerobot/policies/wall_x/qwen_model/configuration_qwen2_5_vl.py @@ -21,6 +21,7 @@ class Qwen2_5_VLVisionConfig(PretrainedConfig): window_size=112, out_hidden_size=3584, fullatt_block_indexes=[7, 15, 23, 31], + initializer_range=0.02, **kwargs, ): super().__init__(**kwargs) @@ -38,6 +39,7 @@ class Qwen2_5_VLVisionConfig(PretrainedConfig): self.window_size = window_size self.fullatt_block_indexes = fullatt_block_indexes self.out_hidden_size = out_hidden_size + self.initializer_range = initializer_range class Qwen2_5_VLConfig(PretrainedConfig): diff --git a/src/lerobot/policies/wall_x/qwen_model/qwen2_5_vl_moe.py b/src/lerobot/policies/wall_x/qwen_model/qwen2_5_vl_moe.py index a1309ea9a..ecf3eb371 100644 --- a/src/lerobot/policies/wall_x/qwen_model/qwen2_5_vl_moe.py +++ b/src/lerobot/policies/wall_x/qwen_model/qwen2_5_vl_moe.py @@ -602,19 +602,40 @@ class Qwen2_5_VisionTransformerPretrainedModel(Qwen2_5_VLPreTrainedModel): return hidden_states +def _compute_default_rope_parameters_qwen2_5_vl(config, device=None): + """ + compute default rope parameters for Qwen2_5_VL + """ + base = config.text_config.rope_parameters["rope_theta"] + dim = config.hidden_size // config.num_attention_heads + inv_freq = 1.0 / ( + base ** (torch.arange(0, dim, 2, dtype=torch.int64).to(device=device, dtype=torch.float) / dim) + ) + return inv_freq, 1.0 + + class Qwen2_5_VLRotaryEmbedding(nn.Module): def __init__(self, config: Qwen2_5_VLConfig, device=None): super().__init__() # BC: "rope_type" was originally "type" if hasattr(config, "rope_scaling") and config.rope_scaling is not None: self.rope_type = config.rope_scaling.get("rope_type", config.rope_scaling.get("type")) + elif hasattr(config, "rope_parameters") and config.rope_parameters is not None: + self.rope_type = config.rope_parameters.get("rope_type", "default") else: self.rope_type = "default" self.max_seq_len_cached = config.max_position_embeddings self.original_max_seq_len = config.max_position_embeddings self.config = config - self.rope_init_fn = ROPE_INIT_FUNCTIONS[self.rope_type] + + if self.rope_type == "default": + self.rope_init_fn = _compute_default_rope_parameters_qwen2_5_vl + self.rope_kwargs = {} + else: + rope_type_key = "linear" if self.rope_type == "linear" else self.rope_type + self.rope_init_fn = ROPE_INIT_FUNCTIONS[rope_type_key] + self.rope_kwargs = {} inv_freq, self.attention_scaling = self.rope_init_fn(self.config, device) self.register_buffer("inv_freq", inv_freq, persistent=False) @@ -1575,7 +1596,7 @@ QWEN2_5_VL_INPUTS_DOCSTRING = r""" class Qwen2_5_VLForConditionalGeneration(Qwen2_5_VLPreTrainedModel, GenerationMixin): - _tied_weights_keys = ["lm_head.weight"] + _tied_weights_keys = {"lm_head.weight": "model.embed_tokens.weight"} config_class = Qwen2_5_VLConfig _no_split_modules = ["Qwen2_5_VLDecoderLayer", "Qwen2_5_VLVisionBlock"] diff --git a/src/lerobot/policies/wall_x/utils.py b/src/lerobot/policies/wall_x/utils.py index 2ea40b377..e08ef69d5 100644 --- a/src/lerobot/policies/wall_x/utils.py +++ b/src/lerobot/policies/wall_x/utils.py @@ -144,7 +144,7 @@ def preprocesser_call( """ # Process image inputs if images is not None and len(images) > 0: - image_inputs = processor.image_processor(images=images, videos=None, return_tensors=return_tensors) + image_inputs = processor.image_processor(images=images, return_tensors=return_tensors) image_grid_thw = image_inputs["image_grid_thw"] else: image_inputs = {} @@ -152,7 +152,7 @@ def preprocesser_call( # Process video inputs if videos is not None: - videos_inputs = processor.image_processor(images=None, videos=videos, return_tensors=return_tensors) + videos_inputs = processor.image_processor(videos=videos, return_tensors=return_tensors) video_grid_thw = videos_inputs["video_grid_thw"] else: videos_inputs = {} diff --git a/src/lerobot/scripts/lerobot_calibrate.py b/src/lerobot/scripts/lerobot_calibrate.py index 1b30021dd..242067978 100644 --- a/src/lerobot/scripts/lerobot_calibrate.py +++ b/src/lerobot/scripts/lerobot_calibrate.py @@ -56,6 +56,7 @@ from lerobot.teleoperators import ( # noqa: F401 make_teleoperator_from_config, omx_leader, openarm_leader, + openarm_mini, so_leader, unitree_g1, ) diff --git a/src/lerobot/scripts/lerobot_dataset_viz.py b/src/lerobot/scripts/lerobot_dataset_viz.py index 29d64554f..c4b676c67 100644 --- a/src/lerobot/scripts/lerobot_dataset_viz.py +++ b/src/lerobot/scripts/lerobot_dataset_viz.py @@ -132,10 +132,13 @@ def visualize_dataset( logging.info("Logging to Rerun") + first_index = None for batch in tqdm.tqdm(dataloader, total=len(dataloader)): + if first_index is None: + first_index = batch["index"][0].item() # iterate over the batch for i in range(len(batch["index"])): - rr.set_time("frame_index", sequence=batch["frame_index"][i].item()) + rr.set_time("frame_index", sequence=batch["index"][i].item() - first_index) rr.set_time("timestamp", timestamp=batch["timestamp"][i].item()) # display each camera image diff --git a/src/lerobot/scripts/lerobot_edit_dataset.py b/src/lerobot/scripts/lerobot_edit_dataset.py index afdc95efd..49825317d 100644 --- a/src/lerobot/scripts/lerobot_edit_dataset.py +++ b/src/lerobot/scripts/lerobot_edit_dataset.py @@ -21,6 +21,9 @@ This script allows you to delete episodes, split datasets, merge datasets, remove features, modify tasks, and convert image datasets to video format. When new_repo_id is specified, creates a new dataset. +Path semantics (v2): --root and --new_root are exact dataset folders containing +meta/, data/, videos/. When omitted, defaults to $HF_LEROBOT_HOME/{repo_id}. + Usage Examples: Delete episodes 0, 2, and 5 from a dataset: @@ -29,16 +32,31 @@ Delete episodes 0, 2, and 5 from a dataset: --operation.type delete_episodes \ --operation.episode_indices "[0, 2, 5]" -Delete episodes and save to a new dataset: +Delete episodes from a local dataset at a specific path: lerobot-edit-dataset \ --repo_id lerobot/pusht \ - --new_repo_id lerobot/pusht_filtered \ + --root /path/to/pusht \ --operation.type delete_episodes \ --operation.episode_indices "[0, 2, 5]" -Split dataset by fractions: +Delete episodes and save to a new dataset at a specific path and with a new repo_id: lerobot-edit-dataset \ --repo_id lerobot/pusht \ + --new_repo_id lerobot/pusht_filtered \ + --new_root /path/to/pusht_filtered \ + --operation.type delete_episodes \ + --operation.episode_indices "[0, 2, 5]" + +Split dataset by fractions (pusht_train, pusht_val): + lerobot-edit-dataset \ + --repo_id lerobot/pusht \ + --operation.type split \ + --operation.splits '{"train": 0.8, "val": 0.2}' + +Split dataset by fractions and save split datasets to a specific folder (base_folder/train, base_folder/val): + lerobot-edit-dataset \ + --repo_id lerobot/pusht \ + --new_root /path/to/base_folder \ --operation.type split \ --operation.splits '{"train": 0.8, "val": 0.2}' @@ -56,15 +74,29 @@ Split into more than two splits: Merge multiple datasets: lerobot-edit-dataset \ - --repo_id lerobot/pusht_merged \ + --new_repo_id lerobot/pusht_merged \ --operation.type merge \ --operation.repo_ids "['lerobot/pusht_train', 'lerobot/pusht_val']" +Merge multiple datasets to a specific output path: + lerobot-edit-dataset \ + --new_repo_id lerobot/pusht_merged \ + --new_root /path/to/pusht_merged \ + --operation.type merge \ + --operation.repo_ids "['lerobot/pusht_train', 'lerobot/pusht_val']" + +Merge multiple datasets from a list of local dataset paths: + lerobot-edit-dataset \ + --new_repo_id lerobot/pusht_merged \ + --operation.type merge \ + --operation.repo_ids "['pusht_train', 'pusht_val']" \ + --operation.roots "['/path/to/pusht_train', '/path/to/pusht_val']" + Remove camera feature: lerobot-edit-dataset \ --repo_id lerobot/pusht \ --operation.type remove_feature \ - --operation.feature_names "['observation.images.top']" + --operation.feature_names "['observation.image']" Modify tasks - set a single task for all episodes (WARNING: modifies in-place): lerobot-edit-dataset \ @@ -88,8 +120,8 @@ Modify tasks - set default task with overrides for specific episodes (WARNING: m Convert image dataset to video format and save locally: lerobot-edit-dataset \ --repo_id lerobot/pusht_image \ - --operation.type convert_image_to_video \ - --operation.output_dir /path/to/output/pusht_video + --new_root /path/to/output/pusht_video \ + --operation.type convert_image_to_video Convert image dataset to video format and save with new repo_id: lerobot-edit-dataset \ @@ -167,6 +199,7 @@ class SplitConfig(OperationConfig): @dataclass class MergeConfig(OperationConfig): repo_ids: list[str] | None = None + roots: list[str] | None = None @OperationConfig.register_subclass("remove_feature") @@ -200,36 +233,46 @@ class ConvertImageToVideoConfig(OperationConfig): @OperationConfig.register_subclass("info") @dataclass class InfoConfig(OperationConfig): - type: str = "info" show_features: bool = False @dataclass class EditDatasetConfig: - repo_id: str + # Operation configuration. operation: OperationConfig + # Input dataset identifier. Always required unless for Merge operation. + repo_id: str | None = None + # Root directory where the input dataset is stored. If not specified, defaults to $HF_LEROBOT_HOME/repo_id. root: str | None = None + # Edited dataset identifier. When both new_repo_id (resp. new_root) and repo_id (resp. root) are identical, modifications are applied in-place and a backup of the original dataset is created. Required for Merge operation. new_repo_id: str | None = None + # Root directory where the edited dataset will be stored. If not specified, defaults to $HF_LEROBOT_HOME/new_repo_id. For Split operation, this is the base directory for the split datasets. + new_root: str | None = None + # Upload dataset to Hugging Face hub. push_to_hub: bool = False -def get_output_path(repo_id: str, new_repo_id: str | None, root: Path | None) -> tuple[str, Path]: - if new_repo_id: - output_repo_id = new_repo_id - output_dir = root / new_repo_id if root else HF_LEROBOT_HOME / new_repo_id - else: - output_repo_id = repo_id - dataset_path = root / repo_id if root else HF_LEROBOT_HOME / repo_id - old_path = Path(str(dataset_path) + "_old") +def get_output_path( + repo_id: str, + new_repo_id: str | None, + root: Path | str | None, + new_root: Path | str | None, +) -> tuple[str, Path]: + input_path = Path(root) if root else HF_LEROBOT_HOME / repo_id - if dataset_path.exists(): - if old_path.exists(): - shutil.rmtree(old_path) - shutil.move(str(dataset_path), str(old_path)) + output_repo_id = new_repo_id if new_repo_id else repo_id + output_path = Path(new_root) if new_root else HF_LEROBOT_HOME / output_repo_id - output_dir = dataset_path + # In case of in-place modification, create a backup of the original dataset (if it exists) + if output_path == input_path: + backup_path = input_path.with_name(input_path.name + "_old") - return output_repo_id, output_dir + if input_path.exists(): + if backup_path.exists(): + shutil.rmtree(backup_path) + shutil.move(input_path, backup_path) + + return output_repo_id, output_path def handle_delete_episodes(cfg: EditDatasetConfig) -> None: @@ -241,11 +284,15 @@ def handle_delete_episodes(cfg: EditDatasetConfig) -> None: dataset = LeRobotDataset(cfg.repo_id, root=cfg.root) output_repo_id, output_dir = get_output_path( - cfg.repo_id, cfg.new_repo_id, Path(cfg.root) if cfg.root else None + cfg.repo_id, + new_repo_id=cfg.new_repo_id, + root=cfg.root, + new_root=cfg.new_root, ) - if cfg.new_repo_id is None: - dataset.root = Path(str(dataset.root) + "_old") + # In case of in-place modification, make the dataset point to the backup directory + if output_dir == dataset.root: + dataset.root = dataset.root.with_name(dataset.root.name + "_old") logging.info(f"Deleting episodes {cfg.operation.episode_indices} from {cfg.repo_id}") new_dataset = delete_episodes( @@ -272,19 +319,27 @@ def handle_split(cfg: EditDatasetConfig) -> None: "splits dict must be specified with split names as keys and fractions/episode lists as values" ) + if cfg.new_repo_id is not None: + logging.warning( + "split uses the original dataset identifier --repo_id to generate split names. The --new_repo_id parameter is ignored." + ) + dataset = LeRobotDataset(cfg.repo_id, root=cfg.root) logging.info(f"Splitting dataset {cfg.repo_id} with splits: {cfg.operation.splits}") - split_datasets = split_dataset(dataset, splits=cfg.operation.splits) + split_datasets = split_dataset( + dataset, + splits=cfg.operation.splits, + output_dir=cfg.new_root, + ) for split_name, split_ds in split_datasets.items(): - split_repo_id = f"{cfg.repo_id}_{split_name}" logging.info( f"{split_name}: {split_ds.meta.total_episodes} episodes, {split_ds.meta.total_frames} frames" ) if cfg.push_to_hub: - logging.info(f"Pushing {split_name} split to hub as {split_repo_id}") + logging.info(f"Pushing {split_name} split to hub as {split_ds.repo_id}") LeRobotDataset(split_ds.repo_id, root=split_ds.root).push_to_hub() @@ -295,18 +350,29 @@ def handle_merge(cfg: EditDatasetConfig) -> None: if not cfg.operation.repo_ids: raise ValueError("repo_ids must be specified for merge operation") - if not cfg.repo_id: - raise ValueError("repo_id must be specified as the output repository for merged dataset") + if cfg.repo_id is not None or cfg.root is not None: + logging.warning( + "merge uses --new_repo_id and --new_root for the merged dataset. The --repo_id and --root parameters are ignored." + ) - logging.info(f"Loading {len(cfg.operation.repo_ids)} datasets to merge") - datasets = [LeRobotDataset(repo_id, root=cfg.root) for repo_id in cfg.operation.repo_ids] + if cfg.operation.roots: + if len(cfg.operation.roots) != len(cfg.operation.repo_ids): + raise ValueError("repo_ids and roots must have the same length for merge operation") + logging.info(f"Loading {len(cfg.operation.roots)} datasets to merge") + datasets = [ + LeRobotDataset(repo_id=repo_id, root=root) + for repo_id, root in zip(cfg.operation.repo_ids, cfg.operation.roots, strict=True) + ] + else: + logging.info(f"Loading {len(cfg.operation.repo_ids)} datasets to merge") + datasets = [LeRobotDataset(repo_id) for repo_id in cfg.operation.repo_ids] - output_dir = Path(cfg.root) / cfg.repo_id if cfg.root else HF_LEROBOT_HOME / cfg.repo_id + output_dir = Path(cfg.new_root) if cfg.new_root else HF_LEROBOT_HOME / cfg.new_repo_id - logging.info(f"Merging datasets into {cfg.repo_id}") + logging.info(f"Merging datasets into {cfg.new_repo_id}") merged_dataset = merge_datasets( datasets, - output_repo_id=cfg.repo_id, + output_repo_id=cfg.new_repo_id, output_dir=output_dir, ) @@ -316,7 +382,7 @@ def handle_merge(cfg: EditDatasetConfig) -> None: ) if cfg.push_to_hub: - logging.info(f"Pushing to hub as {cfg.repo_id}") + logging.info(f"Pushing to hub as {cfg.new_repo_id}") LeRobotDataset(merged_dataset.repo_id, root=output_dir).push_to_hub() @@ -329,11 +395,15 @@ def handle_remove_feature(cfg: EditDatasetConfig) -> None: dataset = LeRobotDataset(cfg.repo_id, root=cfg.root) output_repo_id, output_dir = get_output_path( - cfg.repo_id, cfg.new_repo_id, Path(cfg.root) if cfg.root else None + cfg.repo_id, + new_repo_id=cfg.new_repo_id, + root=cfg.root, + new_root=cfg.new_root, ) - if cfg.new_repo_id is None: - dataset.root = Path(str(dataset.root) + "_old") + # In case of in-place modification, make the dataset point to the backup directory + if output_dir == dataset.root: + dataset.root = dataset.root.with_name(dataset.root.name + "_old") logging.info(f"Removing features {cfg.operation.feature_names} from {cfg.repo_id}") new_dataset = remove_feature( @@ -361,9 +431,10 @@ def handle_modify_tasks(cfg: EditDatasetConfig) -> None: if new_task is None and episode_tasks_raw is None: raise ValueError("Must specify at least one of new_task or episode_tasks for modify_tasks operation") - # Warn about in-place modification behavior - if cfg.new_repo_id is not None: - logging.warning("modify_tasks modifies datasets in-place. The --new_repo_id parameter is ignored.") + if cfg.new_repo_id is not None or cfg.new_root is not None: + logging.warning( + "modify_tasks modifies datasets in-place. The --new_repo_id and --new_root parameters are ignored." + ) dataset = LeRobotDataset(cfg.repo_id, root=cfg.root) logging.warning(f"Modifying dataset in-place at {dataset.root}. Original data will be overwritten.") @@ -399,32 +470,30 @@ def handle_convert_image_to_video(cfg: EditDatasetConfig) -> None: dataset = LeRobotDataset(cfg.repo_id, root=cfg.root) # Determine output directory and repo_id - # Priority: 1) new_repo_id, 2) operation.output_dir, 3) auto-generated name + # Priority: 1) new_root, 2) new_repo_id, 3) operation.output_dir, 4) auto-generated name output_dir_config = getattr(cfg.operation, "output_dir", None) + if output_dir_config: + logging.warning( + "--operation.output_dir is deprecated and will be removed in future versions. " + "Please use --new_root instead." + ) - if cfg.new_repo_id: - # Use new_repo_id for both local storage and hub push + if cfg.new_root: + output_dir = Path(cfg.new_root) + output_repo_id = cfg.new_repo_id or f"{cfg.repo_id}_video" + logging.info(f"Saving to new_root: {output_dir} as {output_repo_id}") + elif cfg.new_repo_id: output_repo_id = cfg.new_repo_id - # Place new dataset as a sibling to the original dataset - # Get the parent of the actual dataset root (not cfg.root which might be the lerobot cache dir) - # Extract just the dataset name (after last slash) for the local directory - local_dir_name = cfg.new_repo_id.split("/")[-1] - output_dir = dataset.root.parent / local_dir_name + output_dir = HF_LEROBOT_HOME / cfg.new_repo_id logging.info(f"Saving to new dataset: {cfg.new_repo_id} at {output_dir}") elif output_dir_config: - # Use custom output directory for local-only storage output_dir = Path(output_dir_config) - # Extract repo name from output_dir for the dataset output_repo_id = output_dir.name - logging.info(f"Saving to local directory: {output_dir}") + logging.info(f"Saving to local directory: {output_dir} as {output_repo_id}") else: - # Auto-generate name: append "_video" to original repo_id output_repo_id = f"{cfg.repo_id}_video" - # Place new dataset as a sibling to the original dataset - # Extract just the dataset name (after last slash) for the local directory - local_dir_name = output_repo_id.split("/")[-1] - output_dir = dataset.root.parent / local_dir_name - logging.info(f"Saving to auto-generated location: {output_dir}") + output_dir = HF_LEROBOT_HOME / output_repo_id + logging.info(f"Saving to auto-generated location: {output_dir} as {output_repo_id}") logging.info(f"Converting dataset {cfg.repo_id} to video format") @@ -499,8 +568,20 @@ def handle_info(cfg: EditDatasetConfig): sys.stdout.write(f"{feature_dump_str}\n") +def _validate_config(cfg: EditDatasetConfig) -> None: + if isinstance(cfg.operation, MergeConfig): + if not cfg.new_repo_id: + raise ValueError("--new_repo_id is required for merge operation (the merged dataset identifier)") + else: + if not cfg.repo_id: + raise ValueError( + f"--repo_id is required for {cfg.operation.type} operation (the input dataset identifier)" + ) + + @parser.wrap() def edit_dataset(cfg: EditDatasetConfig) -> None: + _validate_config(cfg) operation_type = cfg.operation.type if operation_type == "delete_episodes": diff --git a/src/lerobot/scripts/lerobot_find_joint_limits.py b/src/lerobot/scripts/lerobot_find_joint_limits.py index 082d11803..bcb93ba12 100644 --- a/src/lerobot/scripts/lerobot_find_joint_limits.py +++ b/src/lerobot/scripts/lerobot_find_joint_limits.py @@ -61,6 +61,7 @@ from lerobot.teleoperators import ( # noqa: F401 make_teleoperator_from_config, omx_leader, openarm_leader, + openarm_mini, so_leader, ) from lerobot.utils.robot_utils import precise_sleep diff --git a/src/lerobot/scripts/lerobot_record.py b/src/lerobot/scripts/lerobot_record.py index ec04975d4..72708ba23 100644 --- a/src/lerobot/scripts/lerobot_record.py +++ b/src/lerobot/scripts/lerobot_record.py @@ -125,6 +125,7 @@ from lerobot.teleoperators import ( # noqa: F401 make_teleoperator_from_config, omx_leader, openarm_leader, + openarm_mini, reachy2_teleoperator, so_leader, unitree_g1, @@ -154,7 +155,7 @@ class DatasetRecordConfig: repo_id: str # A short but accurate description of the task performed during the recording (e.g. "Pick the Lego block and drop it in the box on the right.") single_task: str - # Root directory where the dataset will be stored (e.g. 'dataset/path'). + # Root directory where the dataset will be stored (e.g. 'dataset/path'). If None, defaults to $HF_LEROBOT_HOME/repo_id. root: str | Path | None = None # Limit the frames per second. fps: int = 30 @@ -333,6 +334,7 @@ def record_loop( preprocessor.reset() postprocessor.reset() + no_action_count = 0 timestamp = 0 start_episode_t = time.perf_counter() while timestamp < control_time_s: @@ -380,11 +382,13 @@ def record_loop( act = {**arm_action, **base_action} if len(base_action) > 0 else arm_action act_processed_teleop = teleop_action_processor((act, obs)) else: - logging.info( - "No policy or teleoperator provided, skipping action generation." - "This is likely to happen when resetting the environment without a teleop device." - "The robot won't be at its rest position at the start of the next episode." - ) + no_action_count += 1 + if no_action_count == 1 or no_action_count % 10 == 0: + logging.warning( + "No policy or teleoperator provided, skipping action generation. " + "This is likely to happen when resetting the environment without a teleop device. " + "The robot won't be at its rest position at the start of the next episode." + ) continue # Applies a pipeline to the action, default is IdentityProcessor diff --git a/src/lerobot/scripts/lerobot_replay.py b/src/lerobot/scripts/lerobot_replay.py index 8e2a394b9..7c0b5b96b 100644 --- a/src/lerobot/scripts/lerobot_replay.py +++ b/src/lerobot/scripts/lerobot_replay.py @@ -80,7 +80,7 @@ class DatasetReplayConfig: repo_id: str # Episode to replay. episode: int - # Root directory where the dataset will be stored (e.g. 'dataset/path'). + # Root directory where the dataset will be stored (e.g. 'dataset/path'). If None, defaults to $HF_LEROBOT_HOME/repo_id. root: str | Path | None = None # Limit the frames per second. By default, uses the policy fps. fps: int = 30 diff --git a/src/lerobot/scripts/lerobot_setup_motors.py b/src/lerobot/scripts/lerobot_setup_motors.py index 01af95b61..2c962a6e2 100644 --- a/src/lerobot/scripts/lerobot_setup_motors.py +++ b/src/lerobot/scripts/lerobot_setup_motors.py @@ -43,6 +43,7 @@ from lerobot.teleoperators import ( # noqa: F401 koch_leader, make_teleoperator_from_config, omx_leader, + openarm_mini, so_leader, ) @@ -51,6 +52,7 @@ COMPATIBLE_DEVICES = [ "koch_leader", "omx_follower", "omx_leader", + "openarm_mini", "so100_follower", "so100_leader", "so101_follower", diff --git a/src/lerobot/scripts/lerobot_teleoperate.py b/src/lerobot/scripts/lerobot_teleoperate.py index b6aa4a750..dad479b2e 100644 --- a/src/lerobot/scripts/lerobot_teleoperate.py +++ b/src/lerobot/scripts/lerobot_teleoperate.py @@ -94,6 +94,7 @@ from lerobot.teleoperators import ( # noqa: F401 make_teleoperator_from_config, omx_leader, openarm_leader, + openarm_mini, reachy2_teleoperator, so_leader, unitree_g1, diff --git a/src/lerobot/scripts/lerobot_train.py b/src/lerobot/scripts/lerobot_train.py index 93b99e245..04d43d91e 100644 --- a/src/lerobot/scripts/lerobot_train.py +++ b/src/lerobot/scripts/lerobot_train.py @@ -24,6 +24,7 @@ import torch from accelerate import Accelerator from termcolor import colored from torch.optim import Optimizer +from tqdm import tqdm from lerobot.configs import parser from lerobot.configs.train import TrainPipelineConfig @@ -51,6 +52,7 @@ from lerobot.utils.utils import ( format_big_number, has_method, init_logging, + inside_slurm, ) @@ -378,10 +380,10 @@ def train(cfg: TrainPipelineConfig, accelerator: Accelerator | None = None): "dataloading_s": AverageMeter("data_s", ":.3f"), } - # Use effective batch size for proper epoch calculation in distributed training + # Keep global batch size for logging; MetricsTracker handles world size internally. effective_batch_size = cfg.batch_size * accelerator.num_processes train_tracker = MetricsTracker( - effective_batch_size, + cfg.batch_size, dataset.num_frames, dataset.num_episodes, train_metrics, @@ -390,6 +392,14 @@ def train(cfg: TrainPipelineConfig, accelerator: Accelerator | None = None): ) if is_main_process: + progbar = tqdm( + total=cfg.steps - step, + desc="Training", + unit="step", + disable=inside_slurm(), + position=0, + leave=True, + ) logging.info( f"Start offline training on a fixed dataset, with effective batch size: {effective_batch_size}" ) @@ -414,6 +424,8 @@ def train(cfg: TrainPipelineConfig, accelerator: Accelerator | None = None): # Note: eval and checkpoint happens *after* the `step`th training update has completed, so we # increment `step` here. step += 1 + if is_main_process: + progbar.update(1) train_tracker.step() is_log_step = cfg.log_freq > 0 and step % cfg.log_freq == 0 and is_main_process is_saving_step = step % cfg.save_freq == 0 or step == cfg.steps @@ -507,6 +519,9 @@ def train(cfg: TrainPipelineConfig, accelerator: Accelerator | None = None): accelerator.wait_for_everyone() + if is_main_process: + progbar.close() + if eval_env: close_envs(eval_env) diff --git a/src/lerobot/teleoperators/openarm_mini/__init__.py b/src/lerobot/teleoperators/openarm_mini/__init__.py new file mode 100644 index 000000000..8620af1d7 --- /dev/null +++ b/src/lerobot/teleoperators/openarm_mini/__init__.py @@ -0,0 +1,20 @@ +#!/usr/bin/env python + +# Copyright 2025 The HuggingFace Inc. team. 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. + +from .config_openarm_mini import OpenArmMiniConfig +from .openarm_mini import OpenArmMini + +__all__ = ["OpenArmMini", "OpenArmMiniConfig"] diff --git a/src/lerobot/teleoperators/openarm_mini/config_openarm_mini.py b/src/lerobot/teleoperators/openarm_mini/config_openarm_mini.py new file mode 100644 index 000000000..7dc3e0212 --- /dev/null +++ b/src/lerobot/teleoperators/openarm_mini/config_openarm_mini.py @@ -0,0 +1,30 @@ +#!/usr/bin/env python + +# Copyright 2025 The HuggingFace Inc. team. 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. + +from dataclasses import dataclass + +from ..config import TeleoperatorConfig + + +@TeleoperatorConfig.register_subclass("openarm_mini") +@dataclass +class OpenArmMiniConfig(TeleoperatorConfig): + """Configuration for OpenArm Mini teleoperator with Feetech motors (dual arms).""" + + port_right: str = "/dev/ttyUSB0" + port_left: str = "/dev/ttyUSB1" + + use_degrees: bool = True diff --git a/src/lerobot/teleoperators/openarm_mini/openarm_mini.py b/src/lerobot/teleoperators/openarm_mini/openarm_mini.py new file mode 100644 index 000000000..3fbcecf24 --- /dev/null +++ b/src/lerobot/teleoperators/openarm_mini/openarm_mini.py @@ -0,0 +1,296 @@ +#!/usr/bin/env python + +# Copyright 2025 The HuggingFace Inc. team. 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. + +import logging +import time +from typing import Any + +from lerobot.motors import Motor, MotorCalibration, MotorNormMode +from lerobot.motors.feetech import ( + FeetechMotorsBus, + OperatingMode, +) +from lerobot.processor import RobotAction +from lerobot.utils.decorators import check_if_already_connected, check_if_not_connected + +from ..teleoperator import Teleoperator +from .config_openarm_mini import OpenArmMiniConfig + +logger = logging.getLogger(__name__) + +# Motors whose direction is inverted during readout +RIGHT_MOTORS_TO_FLIP = ["joint_1", "joint_2", "joint_3", "joint_4", "joint_5"] +LEFT_MOTORS_TO_FLIP = ["joint_1", "joint_3", "joint_4", "joint_5", "joint_6", "joint_7"] + + +class OpenArmMini(Teleoperator): + """ + OpenArm Mini Teleoperator with dual Feetech-based arms (8 motors per arm). + + Each arm has 7 joints plus a gripper, using Feetech STS3215 servos. + """ + + config_class = OpenArmMiniConfig + name = "openarm_mini" + + def __init__(self, config: OpenArmMiniConfig): + super().__init__(config) + self.config = config + + norm_mode_body = MotorNormMode.DEGREES + + motors_right = { + "joint_1": Motor(1, "sts3215", norm_mode_body), + "joint_2": Motor(2, "sts3215", norm_mode_body), + "joint_3": Motor(3, "sts3215", norm_mode_body), + "joint_4": Motor(4, "sts3215", norm_mode_body), + "joint_5": Motor(5, "sts3215", norm_mode_body), + "joint_6": Motor(6, "sts3215", norm_mode_body), + "joint_7": Motor(7, "sts3215", norm_mode_body), + "gripper": Motor(8, "sts3215", MotorNormMode.RANGE_0_100), + } + + motors_left = { + "joint_1": Motor(1, "sts3215", norm_mode_body), + "joint_2": Motor(2, "sts3215", norm_mode_body), + "joint_3": Motor(3, "sts3215", norm_mode_body), + "joint_4": Motor(4, "sts3215", norm_mode_body), + "joint_5": Motor(5, "sts3215", norm_mode_body), + "joint_6": Motor(6, "sts3215", norm_mode_body), + "joint_7": Motor(7, "sts3215", norm_mode_body), + "gripper": Motor(8, "sts3215", MotorNormMode.RANGE_0_100), + } + + cal_right = { + k.replace("right_", ""): v for k, v in (self.calibration or {}).items() if k.startswith("right_") + } + cal_left = { + k.replace("left_", ""): v for k, v in (self.calibration or {}).items() if k.startswith("left_") + } + + self.bus_right = FeetechMotorsBus( + port=self.config.port_right, + motors=motors_right, + calibration=cal_right, + ) + + self.bus_left = FeetechMotorsBus( + port=self.config.port_left, + motors=motors_left, + calibration=cal_left, + ) + + @property + def action_features(self) -> dict[str, type]: + features: dict[str, type] = {} + for motor in self.bus_right.motors: + features[f"right_{motor}.pos"] = float + for motor in self.bus_left.motors: + features[f"left_{motor}.pos"] = float + return features + + @property + def feedback_features(self) -> dict[str, type]: + return {} + + @property + def is_connected(self) -> bool: + return self.bus_right.is_connected and self.bus_left.is_connected + + @check_if_already_connected + def connect(self, calibrate: bool = True) -> None: + logger.info(f"Connecting right arm on {self.config.port_right}...") + self.bus_right.connect() + logger.info(f"Connecting left arm on {self.config.port_left}...") + self.bus_left.connect() + + if calibrate: + self.calibrate() + + self.configure() + logger.info(f"{self} connected.") + + @property + def is_calibrated(self) -> bool: + return self.bus_right.is_calibrated and self.bus_left.is_calibrated + + def calibrate(self) -> None: + """ + Run calibration procedure for OpenArm Mini. + + 1. Disable torque + 2. Ask user to position arms in hanging position with grippers closed + 3. Set this as zero position via half-turn homing + 4. Interactive gripper calibration (open/close positions) + 5. Save calibration + """ + if self.calibration: + user_input = input( + f"Press ENTER to use existing calibration for {self.id}, " + f"or type 'c' and press ENTER to run new calibration: " + ) + if user_input.strip().lower() != "c": + logger.info(f"Using existing calibration for {self.id}") + cal_right = { + k.replace("right_", ""): v for k, v in self.calibration.items() if k.startswith("right_") + } + cal_left = { + k.replace("left_", ""): v for k, v in self.calibration.items() if k.startswith("left_") + } + self.bus_right.write_calibration(cal_right) + self.bus_left.write_calibration(cal_left) + return + + logger.info(f"\nRunning calibration for {self}") + + self._calibrate_arm("right", self.bus_right) + self._calibrate_arm("left", self.bus_left) + + self._save_calibration() + print(f"\nCalibration complete and saved to {self.calibration_fpath}") + + def _calibrate_arm(self, arm_name: str, bus: FeetechMotorsBus) -> None: + """Calibrate a single arm with Feetech motors.""" + logger.info(f"\n=== Calibrating {arm_name.upper()} arm ===") + + bus.disable_torque() + + logger.info(f"Setting Phase to 12 for all motors in {arm_name.upper()} arm...") + for motor in bus.motors: + bus.write("Phase", motor, 12) + + for motor in bus.motors: + bus.write("Operating_Mode", motor, OperatingMode.POSITION.value) + + input( + f"\nCalibration: Zero Position ({arm_name.upper()} arm)\n" + "Position the arm in the following configuration:\n" + " - Arm hanging straight down\n" + " - Gripper closed\n" + "Press ENTER when ready..." + ) + + homing_offsets = bus.set_half_turn_homings() + logger.info(f"{arm_name.capitalize()} arm zero position set.") + + print(f"\nSetting motor ranges for {arm_name.upper()} arm\n") + + if self.calibration is None: + self.calibration = {} + + motor_resolution = bus.model_resolution_table[list(bus.motors.values())[0].model] + max_res = motor_resolution - 1 + + for motor_name, motor in bus.motors.items(): + prefixed_name = f"{arm_name}_{motor_name}" + + if motor_name == "gripper": + input( + f"\nGripper Calibration ({arm_name.upper()} arm)\n" + f"Step 1: CLOSE the gripper fully\n" + f"Press ENTER when gripper is closed..." + ) + closed_pos = bus.read("Present_Position", motor_name, normalize=False) + logger.info(f" Gripper closed position recorded: {closed_pos}") + + input("\nStep 2: OPEN the gripper fully\nPress ENTER when gripper is fully open...") + open_pos = bus.read("Present_Position", motor_name, normalize=False) + logger.info(f" Gripper open position recorded: {open_pos}") + + if closed_pos < open_pos: + range_min = int(closed_pos) + range_max = int(open_pos) + drive_mode = 0 + else: + range_min = int(open_pos) + range_max = int(closed_pos) + drive_mode = 1 + + logger.info( + f" {prefixed_name}: range set to [{range_min}, {range_max}] " + f"(0=closed, 100=open, drive_mode={drive_mode})" + ) + else: + range_min = 0 + range_max = max_res + drive_mode = 0 + logger.info(f" {prefixed_name}: range set to [0, {max_res}] (full motor range)") + + self.calibration[prefixed_name] = MotorCalibration( + id=motor.id, + drive_mode=drive_mode, + homing_offset=homing_offsets[motor_name], + range_min=range_min, + range_max=range_max, + ) + + cal_for_bus = { + k.replace(f"{arm_name}_", ""): v + for k, v in self.calibration.items() + if k.startswith(f"{arm_name}_") + } + bus.write_calibration(cal_for_bus) + + def configure(self) -> None: + self.bus_right.disable_torque() + self.bus_right.configure_motors() + for motor in self.bus_right.motors: + self.bus_right.write("Operating_Mode", motor, OperatingMode.POSITION.value) + + self.bus_left.disable_torque() + self.bus_left.configure_motors() + for motor in self.bus_left.motors: + self.bus_left.write("Operating_Mode", motor, OperatingMode.POSITION.value) + + def setup_motors(self) -> None: + print("\nSetting up RIGHT arm motors...") + for motor in reversed(self.bus_right.motors): + input(f"Connect the controller board to the RIGHT '{motor}' motor only and press enter.") + self.bus_right.setup_motor(motor) + print(f"RIGHT '{motor}' motor id set to {self.bus_right.motors[motor].id}") + + print("\nSetting up LEFT arm motors...") + for motor in reversed(self.bus_left.motors): + input(f"Connect the controller board to the LEFT '{motor}' motor only and press enter.") + self.bus_left.setup_motor(motor) + print(f"LEFT '{motor}' motor id set to {self.bus_left.motors[motor].id}") + + @check_if_not_connected + def get_action(self) -> RobotAction: + """Get current action from both arms (read positions from all motors).""" + start = time.perf_counter() + + right_positions = self.bus_right.sync_read("Present_Position") + left_positions = self.bus_left.sync_read("Present_Position") + + action: dict[str, Any] = {} + for motor, val in right_positions.items(): + action[f"right_{motor}.pos"] = -val if motor in RIGHT_MOTORS_TO_FLIP else val + for motor, val in left_positions.items(): + action[f"left_{motor}.pos"] = -val if motor in LEFT_MOTORS_TO_FLIP else val + + dt_ms = (time.perf_counter() - start) * 1e3 + logger.debug(f"{self} read action: {dt_ms:.1f}ms") + return action + + def send_feedback(self, feedback: dict[str, float]) -> None: + raise NotImplementedError("Feedback is not yet implemented for OpenArm Mini.") + + @check_if_not_connected + def disconnect(self) -> None: + self.bus_right.disconnect() + self.bus_left.disconnect() + logger.info(f"{self} disconnected.") diff --git a/src/lerobot/teleoperators/utils.py b/src/lerobot/teleoperators/utils.py index 16454d5ad..db685f396 100644 --- a/src/lerobot/teleoperators/utils.py +++ b/src/lerobot/teleoperators/utils.py @@ -95,6 +95,10 @@ def make_teleoperator_from_config(config: TeleoperatorConfig) -> "Teleoperator": from .bi_openarm_leader import BiOpenArmLeader return BiOpenArmLeader(config) + elif config.type == "openarm_mini": + from .openarm_mini import OpenArmMini + + return OpenArmMini(config) else: try: return cast("Teleoperator", make_device_from_device_class(config)) diff --git a/src/lerobot/utils/control_utils.py b/src/lerobot/utils/control_utils.py index 7cfe177ef..7c605af17 100644 --- a/src/lerobot/utils/control_utils.py +++ b/src/lerobot/utils/control_utils.py @@ -189,7 +189,7 @@ def sanity_check_dataset_name(repo_id, policy_cfg): # Check if dataset_name starts with "eval_" but policy is missing if dataset_name.startswith("eval_") and policy_cfg is None: raise ValueError( - f"Your dataset name begins with 'eval_' ({dataset_name}), but no policy is provided ({policy_cfg.type})." + f"Your dataset name begins with 'eval_' ({dataset_name}), but no policy is provided." ) # Check if dataset_name does not start with "eval_" but policy is provided diff --git a/src/lerobot/utils/logging_utils.py b/src/lerobot/utils/logging_utils.py index c4c1f42e0..1497c0585 100644 --- a/src/lerobot/utils/logging_utils.py +++ b/src/lerobot/utils/logging_utils.py @@ -104,9 +104,10 @@ class MetricsTracker: self.metrics = metrics self.steps = initial_step + world_size = accelerator.num_processes if accelerator else 1 # A sample is an (observation,action) pair, where observation and action # can be on multiple timestamps. In a batch, we have `batch_size` number of samples. - self.samples = self.steps * self._batch_size + self.samples = self.steps * self._batch_size * world_size self.episodes = self.samples / self._avg_samples_per_ep self.epochs = self.samples / self._num_frames self.accelerator = accelerator @@ -132,7 +133,8 @@ class MetricsTracker: Updates metrics that depend on 'step' for one step. """ self.steps += 1 - self.samples += self._batch_size * (self.accelerator.num_processes if self.accelerator else 1) + world_size = self.accelerator.num_processes if self.accelerator else 1 + self.samples += self._batch_size * world_size self.episodes = self.samples / self._avg_samples_per_ep self.epochs = self.samples / self._num_frames diff --git a/tests/artifacts/policies/pusht_diffusion_/actions.safetensors b/tests/artifacts/policies/pusht_diffusion_/actions.safetensors index ef581727d..70b1411ab 100644 --- a/tests/artifacts/policies/pusht_diffusion_/actions.safetensors +++ b/tests/artifacts/policies/pusht_diffusion_/actions.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:19eaaa85f66ba4aa6388dbb83819ffad6ea4363247208f871a8dc385689f6fc8 +oid sha256:54aecbc1af72a4cd5e9261492f5e7601890517516257aacdf2a0ffb3ce281f1b size 992 diff --git a/tests/artifacts/policies/pusht_diffusion_/grad_stats.safetensors b/tests/artifacts/policies/pusht_diffusion_/grad_stats.safetensors index e00ed3238..bea7d4f19 100644 --- a/tests/artifacts/policies/pusht_diffusion_/grad_stats.safetensors +++ b/tests/artifacts/policies/pusht_diffusion_/grad_stats.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:227296eaeeb54acdc3dae2eb8af3d4d08fb87e245337624447140b1e91cfd002 +oid sha256:88a9c3775a2aa1e90a08850521970070a4fcf0f6b82aab43cd8ccc5cf77e0013 size 47424 diff --git a/tests/artifacts/policies/pusht_diffusion_/output_dict.safetensors b/tests/artifacts/policies/pusht_diffusion_/output_dict.safetensors index f29303992..20cc4f547 100644 --- a/tests/artifacts/policies/pusht_diffusion_/output_dict.safetensors +++ b/tests/artifacts/policies/pusht_diffusion_/output_dict.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:271b00cb2f0cd5fd26b1d53463638e3d1a6e92692ec625fcffb420ca190869e5 +oid sha256:91a2635e05a75fe187a5081504c5f35ce3417378813fa2deaf9ca4e8200e1819 size 68 diff --git a/tests/artifacts/policies/pusht_diffusion_/param_stats.safetensors b/tests/artifacts/policies/pusht_diffusion_/param_stats.safetensors index 614cc754e..365a453dd 100644 --- a/tests/artifacts/policies/pusht_diffusion_/param_stats.safetensors +++ b/tests/artifacts/policies/pusht_diffusion_/param_stats.safetensors @@ -1,3 +1,3 @@ version https://git-lfs.github.com/spec/v1 -oid sha256:778fddbbaa64248cee35cb377c02cc2b6076f7ce5855146de677128900617ddf +oid sha256:645bff922ac7bea63ad018ebf77c303c0e4cd2c1c0dc5ef3192865281bef3dc6 size 47424 diff --git a/tests/fixtures/dataset_factories.py b/tests/fixtures/dataset_factories.py index c33fdcb72..f8dd01fec 100644 --- a/tests/fixtures/dataset_factories.py +++ b/tests/fixtures/dataset_factories.py @@ -222,7 +222,7 @@ def tasks_factory(): def _create_tasks(total_tasks: int = 3) -> pd.DataFrame: ids = list(range(total_tasks)) tasks = [f"Perform action {i}." for i in ids] - df = pd.DataFrame({"task_index": ids}, index=tasks) + df = pd.DataFrame({"task_index": ids}, index=pd.Index(tasks, name="task")) return df return _create_tasks diff --git a/tests/policies/pi0_fast/test_pi0_fast_original_vs_lerobot.py b/tests/policies/pi0_fast/test_pi0_fast_original_vs_lerobot.py index 7b1bbce7d..9de781464 100644 --- a/tests/policies/pi0_fast/test_pi0_fast_original_vs_lerobot.py +++ b/tests/policies/pi0_fast/test_pi0_fast_original_vs_lerobot.py @@ -17,7 +17,6 @@ """Test script to verify PI0Fast policy integration with LeRobot vs the original implementation""" # ruff: noqa: E402 -import os import random from copy import deepcopy from typing import Any @@ -28,10 +27,6 @@ import torch pytest.importorskip("transformers") pytest.importorskip("scipy") -pytestmark = pytest.mark.skipif( - os.environ.get("CI") == "true" or os.environ.get("GITHUB_ACTIONS") == "true", - reason="This test requires accepting the model license", -) from lerobot.policies.pi0_fast.configuration_pi0_fast import PI0FastConfig from lerobot.policies.pi0_fast.modeling_pi0_fast import PI0FastPolicy @@ -53,22 +48,23 @@ DUMMY_STATE_DIM = 20 IMAGE_HEIGHT = 224 IMAGE_WIDTH = 224 NUM_VIEWS = 2 # Number of camera views -DEVICE = "cuda" if torch.cuda.is_available() else "cpu" -MODEL_PATH_LEROBOT = "jadechoghari/pi0fast-base" +DEVICE = "cuda" +MODEL_PATH_LEROBOT = "lerobot/pi0fast-base" # Expected action token shape: (batch_size, max_decoding_steps) EXPECTED_ACTION_TOKENS_SHAPE = (1, 2) # Expected first 5 action tokens (for reproducibility check) -EXPECTED_ACTION_TOKENS_FIRST_5 = torch.tensor([255657, 255425]) +EXPECTED_ACTION_TOKENS_FIRST_5 = torch.tensor([255020, 255589]) # Expected actions after detokenization EXPECTED_ACTIONS_SHAPE = (1, 2, 32) # (batch_size, n_action_steps, action_dim) EXPECTED_ACTIONS_MEAN = 0.046403881162405014 EXPECTED_ACTIONS_STD = 0.2607129216194153 -EXPECTED_ACTIONS_FIRST_5 = torch.tensor([-0.0707, 1.4849, 0.0000, 0.0000, 0.0000]) +EXPECTED_ACTIONS_FIRST_5 = torch.tensor([0.0000, 0.3536, 0.0707, 0.0000, 0.0000]) +@require_cuda def set_seed_all(seed: int): """Set random seed for all RNG sources to ensure reproducibility.""" random.seed(seed) @@ -85,6 +81,7 @@ def set_seed_all(seed: int): torch.use_deterministic_algorithms(True, warn_only=True) +@require_cuda def instantiate_lerobot_pi0_fast( from_pretrained: bool = False, model_path: str = MODEL_PATH_LEROBOT, @@ -127,6 +124,7 @@ def instantiate_lerobot_pi0_fast( return policy, preprocessor, postprocessor +@require_cuda def create_dummy_data(device=DEVICE): """Create dummy data for testing both implementations.""" batch_size = 1 @@ -158,22 +156,25 @@ def create_dummy_data(device=DEVICE): # Pytest fixtures @pytest.fixture(scope="module") +@require_cuda def pi0_fast_components(): """Fixture to instantiate and provide all PI0Fast components for tests.""" print(f"\nTesting with DEVICE='{DEVICE}'") print("\n[Setup] Instantiating LeRobot PI0Fast policy...") policy_obj, preprocessor_obj, postprocessor_obj = instantiate_lerobot_pi0_fast(from_pretrained=True) print("Model loaded successfully") - yield policy_obj, preprocessor_obj, postprocessor_obj + return policy_obj, preprocessor_obj, postprocessor_obj @pytest.fixture(scope="module") +@require_cuda def policy(pi0_fast_components): """Fixture to provide the PI0Fast policy for tests.""" return pi0_fast_components[0] @pytest.fixture(scope="module") +@require_cuda def preprocessor(pi0_fast_components): """Fixture to provide the PI0Fast preprocessor for tests.""" return pi0_fast_components[1] diff --git a/tests/policies/pi0_pi05/test_pi0.py b/tests/policies/pi0_pi05/test_pi0.py index 230e43201..e83abf57d 100644 --- a/tests/policies/pi0_pi05/test_pi0.py +++ b/tests/policies/pi0_pi05/test_pi0.py @@ -16,17 +16,8 @@ """Test script to verify PI0 policy integration with LeRobot, only meant to be run locally!""" -import os - -import pytest import torch -# Skip this entire module in CI -pytestmark = pytest.mark.skipif( - os.environ.get("CI") == "true" or os.environ.get("GITHUB_ACTIONS") == "true", - reason="This test requires accepting the model license", -) - from lerobot.policies.factory import make_policy_config # noqa: E402 from lerobot.policies.pi0 import ( # noqa: E402 PI0Config, diff --git a/tests/policies/pi0_pi05/test_pi05.py b/tests/policies/pi0_pi05/test_pi05.py index acb616960..595191689 100644 --- a/tests/policies/pi0_pi05/test_pi05.py +++ b/tests/policies/pi0_pi05/test_pi05.py @@ -16,25 +16,15 @@ """Test script to verify PI0.5 (pi05) support in PI0 policy, only meant to be run locally!""" -import os - -import pytest import torch -from lerobot.utils.random_utils import set_seed - -# Skip this entire module in CI -pytestmark = pytest.mark.skipif( - os.environ.get("CI") == "true" or os.environ.get("GITHUB_ACTIONS") == "true", - reason="This test requires accepting the model license", -) - from lerobot.policies.factory import make_policy_config # noqa: E402 from lerobot.policies.pi05 import ( # noqa: E402 PI05Config, PI05Policy, make_pi05_pre_post_processors, # noqa: E402 ) +from lerobot.utils.random_utils import set_seed from tests.utils import require_cuda # noqa: E402 diff --git a/tests/policies/pi0_pi05/test_pi05_rtc.py b/tests/policies/pi0_pi05/test_pi05_rtc.py index 3a753031f..0dc240638 100644 --- a/tests/policies/pi0_pi05/test_pi05_rtc.py +++ b/tests/policies/pi0_pi05/test_pi05_rtc.py @@ -24,9 +24,10 @@ import torch # Skip this entire module in CI pytestmark = pytest.mark.skipif( os.environ.get("CI") == "true" or os.environ.get("GITHUB_ACTIONS") == "true", - reason="This test requires local OpenPI installation and is not meant for CI", + reason="TODO: This test seems to hang the CI", ) + from lerobot.configs.types import FeatureType, PolicyFeature, RTCAttentionSchedule # noqa: E402 from lerobot.policies.pi05 import PI05Config, PI05Policy, make_pi05_pre_post_processors # noqa: E402 from lerobot.policies.rtc.configuration_rtc import RTCConfig # noqa: E402 diff --git a/tests/policies/pi0_pi05/test_pi0_rtc.py b/tests/policies/pi0_pi05/test_pi0_rtc.py index 68e94dd94..4105e2068 100644 --- a/tests/policies/pi0_pi05/test_pi0_rtc.py +++ b/tests/policies/pi0_pi05/test_pi0_rtc.py @@ -24,9 +24,10 @@ import torch # Skip this entire module in CI pytestmark = pytest.mark.skipif( os.environ.get("CI") == "true" or os.environ.get("GITHUB_ACTIONS") == "true", - reason="This test requires local OpenPI installation and is not meant for CI", + reason="TODO: This test seems to hang the CI", ) + from lerobot.configs.types import FeatureType, PolicyFeature, RTCAttentionSchedule # noqa: E402 from lerobot.policies.pi0 import PI0Config, PI0Policy, make_pi0_pre_post_processors # noqa: E402 from lerobot.policies.rtc.configuration_rtc import RTCConfig # noqa: E402 @@ -88,6 +89,7 @@ def test_pi0_rtc_initialization_without_rtc_config(): print("✓ PI0 RTC initialization without RTC config: Test passed") +@require_cuda def test_pi0_rtc_inference_with_prev_chunk(): """Test PI0 policy inference with RTC and previous chunk.""" set_seed(42) diff --git a/tests/policies/wall_x/test_wallx.py b/tests/policies/wall_x/test_wallx.py index dcd37b8ef..3514fccd1 100644 --- a/tests/policies/wall_x/test_wallx.py +++ b/tests/policies/wall_x/test_wallx.py @@ -29,8 +29,10 @@ from lerobot.policies.wall_x import WallXConfig # noqa: E402 from lerobot.policies.wall_x.modeling_wall_x import WallXPolicy # noqa: E402 from lerobot.policies.wall_x.processor_wall_x import make_wall_x_pre_post_processors # noqa: E402 from lerobot.utils.random_utils import set_seed # noqa: E402 +from tests.utils import require_cuda # noqa: E402 +@require_cuda def test_policy_instantiation(): # Create config set_seed(42) @@ -115,6 +117,7 @@ def test_policy_instantiation(): raise +@require_cuda def test_config_creation(): """Test policy config creation through factory.""" try: @@ -126,8 +129,3 @@ def test_config_creation(): except Exception as e: print(f"Config creation failed: {e}") raise - - -if __name__ == "__main__": - test_policy_instantiation() - test_config_creation() diff --git a/tests/scripts/test_edit_dataset_parsing.py b/tests/scripts/test_edit_dataset_parsing.py index 8800b92ee..4d758ae35 100644 --- a/tests/scripts/test_edit_dataset_parsing.py +++ b/tests/scripts/test_edit_dataset_parsing.py @@ -27,6 +27,7 @@ from lerobot.scripts.lerobot_edit_dataset import ( OperationConfig, RemoveFeatureConfig, SplitConfig, + _validate_config, ) @@ -51,11 +52,23 @@ class TestOperationTypeParsing: ], ) def test_operation_type_resolves_correct_class(self, type_name, expected_cls): - cfg = parse_cfg(["--repo_id", "test/repo", "--operation.type", type_name]) + cfg = parse_cfg( + ["--repo_id", "test/repo", "--new_repo_id", "test/merged", "--operation.type", type_name] + ) assert isinstance(cfg.operation, expected_cls), ( f"Expected {expected_cls.__name__}, got {type(cfg.operation).__name__}" ) + def test_merge_requires_new_repo_id(self): + cfg = parse_cfg(["--operation.type", "merge"]) + with pytest.raises(ValueError, match="--new_repo_id is required for merge"): + _validate_config(cfg) + + def test_non_merge_requires_repo_id(self): + cfg = parse_cfg(["--operation.type", "delete_episodes"]) + with pytest.raises(ValueError, match="--repo_id is required for delete_episodes"): + _validate_config(cfg) + @pytest.mark.parametrize( "type_name, expected_cls", [ @@ -69,6 +82,8 @@ class TestOperationTypeParsing: ], ) def test_get_choice_name_roundtrips(self, type_name, expected_cls): - cfg = parse_cfg(["--repo_id", "test/repo", "--operation.type", type_name]) + cfg = parse_cfg( + ["--repo_id", "test/repo", "--new_repo_id", "test/merged", "--operation.type", type_name] + ) resolved_name = OperationConfig.get_choice_name(type(cfg.operation)) assert resolved_name == type_name diff --git a/tests/utils/test_logging_utils.py b/tests/utils/test_logging_utils.py index 560ba5701..1207534c0 100644 --- a/tests/utils/test_logging_utils.py +++ b/tests/utils/test_logging_utils.py @@ -24,6 +24,11 @@ def mock_metrics(): return {"loss": AverageMeter("loss", ":.3f"), "accuracy": AverageMeter("accuracy", ":.2f")} +class MockAccelerator: + def __init__(self, num_processes: int): + self.num_processes = num_processes + + def test_average_meter_initialization(): meter = AverageMeter("loss", ":.2f") assert meter.name == "loss" @@ -82,6 +87,37 @@ def test_metrics_tracker_step(mock_metrics): assert tracker.epochs == tracker.samples / 1000 +def test_metrics_tracker_initialization_with_accelerator(mock_metrics): + tracker = MetricsTracker( + batch_size=32, + num_frames=1000, + num_episodes=50, + metrics=mock_metrics, + initial_step=10, + accelerator=MockAccelerator(num_processes=2), + ) + assert tracker.steps == 10 + assert tracker.samples == 10 * 32 * 2 + assert tracker.episodes == tracker.samples / (1000 / 50) + assert tracker.epochs == tracker.samples / 1000 + + +def test_metrics_tracker_step_with_accelerator(mock_metrics): + tracker = MetricsTracker( + batch_size=32, + num_frames=1000, + num_episodes=50, + metrics=mock_metrics, + initial_step=5, + accelerator=MockAccelerator(num_processes=2), + ) + tracker.step() + assert tracker.steps == 6 + assert tracker.samples == (5 * 32 * 2) + (32 * 2) + assert tracker.episodes == tracker.samples / (1000 / 50) + assert tracker.epochs == tracker.samples / 1000 + + def test_metrics_tracker_getattr(mock_metrics): tracker = MetricsTracker(batch_size=32, num_frames=1000, num_episodes=50, metrics=mock_metrics) assert tracker.loss == mock_metrics["loss"]