mirror of
https://github.com/huggingface/lerobot.git
synced 2026-08-08 17:39:44 +00:00
CarolinePascal
8353f84f52
Documents every remaining public class/function across configs/: the types.py feature/normalization enums and PolicyFeature, the PreTrainedConfig and RewardModelConfig base classes (full Args: blocks, abstract property/method contracts, from_pretrained overrides), the TrainPipelineConfig and EvalPipelineConfig top-level pipeline configs, default.py's DatasetConfig/WandBConfig/EvalConfig/PeftConfig/JobConfig, dataset.py's DatasetRecordConfig, video.py's VideoEncoderConfig, the accelerator.py ActivationCheckpointingMode enum, and parser.py's CLI argument-parsing helpers. Converts existing inline `#` field comments on config dataclasses to machine-checked Args: blocks (matching each field against its actual constructor signature), and corrects a couple of stale/misattributed comments found along the way (EvalPipelineConfig's misplaced field comment, PreTrainedConfig's phantom normalization_mapping field). Expands docs/source/api/configs.mdx from 5 documented classes to the module's full public surface. Adds lerobot.configs to check_docstrings.py's MODULES_TO_CHECK ratchet and removes the module's ruff D-ignore. Docstrings and docstring-format changes only; no behavioral changes.
90 lines
2.1 KiB
Plaintext
90 lines
2.1 KiB
Plaintext
# Configuration
|
|
|
|
LeRobot configuration is plain dataclasses parsed by [draccus](https://github.com/dlwh/draccus), so every
|
|
field is settable from the CLI. [`TrainPipelineConfig`] is the top-level object for `lerobot-train`.
|
|
|
|
Polymorphic configs (policies, robots, environments) use `draccus.ChoiceRegistry`: a subclass registers
|
|
itself with `@register_subclass("name")` and is then selectable by that name on the command line.
|
|
|
|
## TrainPipelineConfig
|
|
|
|
[[autodoc]] lerobot.configs.train.TrainPipelineConfig
|
|
|
|
## EvalPipelineConfig
|
|
|
|
[[autodoc]] lerobot.configs.eval.EvalPipelineConfig
|
|
|
|
## PreTrainedConfig
|
|
|
|
[[autodoc]] lerobot.configs.PreTrainedConfig
|
|
|
|
## RewardModelConfig
|
|
|
|
[[autodoc]] lerobot.configs.rewards.RewardModelConfig
|
|
|
|
## DatasetConfig
|
|
|
|
[[autodoc]] lerobot.configs.DatasetConfig
|
|
|
|
## DatasetRecordConfig
|
|
|
|
[[autodoc]] lerobot.configs.DatasetRecordConfig
|
|
|
|
## EvalConfig
|
|
|
|
[[autodoc]] lerobot.configs.EvalConfig
|
|
|
|
## WandBConfig
|
|
|
|
[[autodoc]] lerobot.configs.WandBConfig
|
|
|
|
## PeftConfig
|
|
|
|
[[autodoc]] lerobot.configs.PeftConfig
|
|
|
|
## JobConfig
|
|
|
|
[[autodoc]] lerobot.configs.JobConfig
|
|
|
|
## Feature types
|
|
|
|
[[autodoc]] lerobot.configs.FeatureType
|
|
|
|
[[autodoc]] lerobot.configs.PipelineFeatureType
|
|
|
|
[[autodoc]] lerobot.configs.NormalizationMode
|
|
|
|
[[autodoc]] lerobot.configs.PolicyFeature
|
|
|
|
[[autodoc]] lerobot.configs.RTCAttentionSchedule
|
|
|
|
## Video encoding
|
|
|
|
[[autodoc]] lerobot.configs.VideoEncoderConfig
|
|
|
|
[[autodoc]] lerobot.configs.RGBEncoderConfig
|
|
|
|
[[autodoc]] lerobot.configs.DepthEncoderConfig
|
|
|
|
[[autodoc]] lerobot.configs.encoder_config_from_video_info
|
|
|
|
## Distributed training
|
|
|
|
[[autodoc]] lerobot.configs.parallelism.ParallelismConfig
|
|
|
|
[[autodoc]] lerobot.configs.parallelism.ContextParallelConfig
|
|
|
|
[[autodoc]] lerobot.configs.accelerator.AcceleratorConfig
|
|
|
|
[[autodoc]] lerobot.configs.accelerator.FSDPConfig
|
|
|
|
[[autodoc]] lerobot.configs.accelerator.DDPConfig
|
|
|
|
[[autodoc]] lerobot.configs.accelerator.GradientAccumulationConfig
|
|
|
|
[[autodoc]] lerobot.configs.accelerator.CompileConfig
|
|
|
|
[[autodoc]] lerobot.configs.accelerator.ActivationCheckpointingConfig
|
|
|
|
[[autodoc]] lerobot.configs.accelerator.ActivationCheckpointingMode
|