fix: fast tokenizer conversion should happen offline #106
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Outdated suggestions cannot be applied.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.
Motivation
The server is launched with
HF_HUB_OFFLINE=1
and is meant to treat model files as read-only; however, the fast tokenizer conversion happening in thelauncher
does not follow this (if arevision
is not passed). This can cause problems if a model in HF Hub is updated and the tokenizer conversion downloads the tokenizer files for the new commit of the model but then the server doesn't download the new model files... the server fails to load because it can't find the model files.Modifications
local_files_only=True
with and without the revision arg when doing the fast tokenizer conversionHF_HUB_OFFLINE=1
in the env as well for good measureResult
Fast tokenizer conversion in the launcher should never download new files.
Related Issues
local_files_only
ifrevision
is passed: 🐛 disallow downloads for fast tokenizer conversion #63