[Feature request] Allow specifying `save_dir` in `snapshot_download`

This will make it easier to load without relying on cache_dir. IMO it also matches the function name snapshot_download better. (The current functionality of snapshot_download feels more like cached_download.)

snapshot_download('bert-base-uncased', save_dir='local_dir')
AutoModel.from_pretrained('local_dir')

Currently it’s like this: https://github.com/huggingface/huggingface_hub/blob/eed85ca66fd29618870d630a067ec641e0e98085/src/huggingface_hub/snapshot_download.py#L47-L49

[osanseviero]

Hi @cccntu,

snapshot_download already returns the path to where the repo was saved, so you can do this

local_repo_path = snapshot_download('bert-base-uncased')
model = AutoModel.from_pretrained(local_repo_path)

No need of building the path yourself 😃 Let us know if this works for you

[cccntu]

For context, my use case is first downloading a snapshot, then COPY to docker image at build time. So when the docker images starts, it won’t need to download the model, or even call the api, it can immediately load the local copy.

But I guess I don’t really care about the storage used by cache, copying (instead of moving) the snapshot sounds better. And copying doesn’t feel like something that should be supported. Closing the issue. Thanks for the comments! @LysandreJik @osanseviero