Volumes
Input Volumes
Input volumes are used to specify the data that will be made available to the job during its execution. These can include files, datasets, or other resources that the job needs to access. The following input volume types are supported:
- compute_horde_sdk.v1.InputVolume
alias of
InlineInputVolume
|HuggingfaceInputVolume
|HTTPInputVolume
- class compute_horde_sdk.v1.HuggingfaceInputVolume
Volume for downloading resources from Huggingface.
By default, it downloads the entire repository and copies its structure. To narrow it down, use the
allow_patterns
field. If a file is inside a subfolder, it will be placed under the same path in the volume.- repo_id: str
Huggingface repository ID, in the format “namespace/name”.
- repo_type: str | None
Set to “dataset” or “space” for a dataset or space, None or “model” for a model.
- revision: str | None
Git revision ID: branch name / tag / commit hash.
- allow_patterns: str | list[str] | None
If provided, only files matching at least one pattern are downloaded.
Output Volumes
Output volumes are used to specify where the outputs of the job will be stored. These can include files generated by the job. The following output volume types are supported:
- compute_horde_sdk.v1.OutputVolume
alias of
HTTPOutputVolume