Keep hffs cache in workers when streaming #7820
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
| if name == "token" and getattr(self, "storage_options", None) is not None: | ||
| if "hf" not in self.storage_options: | ||
| self.storage_options["hf"] = {"token": value, "endpoint": config.HF_ENDPOINT} | ||
| self.storage_options["hf"] = {"endpoint": config.HF_ENDPOINT, "token": value} |
There was a problem hiding this comment.
Did you need to make this change? seems weird since dicts aren't ordered in python
There was a problem hiding this comment.
because fsspec cache that maps a filesystem argument to the cached instance is sensitive to the order ^^'
with those changes every instance of HfFileSystem in datasets uses the same order
| storage_options = { | ||
| "token": token, | ||
| "endpoint": config.HF_ENDPOINT, | ||
| "token": token, |
There was a problem hiding this comment.
same, weird that you need this :S
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
(and also reorder the hffs args to improve caching)
When using
DataLoader(iterable_dataset, num_workers=...)the dataset is pickled and passed to the worker. However previously the resulting dataset would be in a process with an empty hffs cache. By keeping the cache attached toIterableDataset, the cached hffs instances are pickled with the dataset and re-populates the cache in the DataLoader workersthis requires huggingface/huggingface_hub#3443 to work effectively though, otherwise the unpickled hffs cache would start empty
cc @andimarafioti @LTMeyer