hdf5
hdf5
¶
Abstract dataset schema for HDF5-based datasets.
Hdf5DatasetSchema
¶
Bases: DatasetSchema
Abstract schema for HDF5-based datasets.
HDF5 datasets store all data (observations, images, actions) in HDF5 files.
Subclasses define the HDF5 structure and extraction logic via extract_episode.
Initialize the HDF5 dataset schema.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
hdf5_paths
|
list[str]
|
List of paths to HDF5 files |
required |
zarr_path
|
str
|
Path to save/load the zarr file |
required |
metadata
|
DatasetMetadata
|
Metadata to use for creating the zarr store from the raw data. |
required |
dataset_type
|
str
|
Type of dataset used by the schema (e.g., 'libero', 'tso', 'metaworld') |
required |
Source code in src/versatil/data/raw/schemas/hdf5.py
get_demo_names
abstractmethod
¶
Get list of demonstration/episode names in the specified HDF5 file.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
hdf5_path
|
str
|
Path to the HDF5 file |
required |
Returns:
| Type | Description |
|---|---|
list[str]
|
List of demo identifiers as strings. |
Source code in src/versatil/data/raw/schemas/hdf5.py
extract_episode
abstractmethod
¶
Extract episodic data from a single HDF5 group.