Prepare 0.15.1 patch #2459
Merged
Prepare 0.15.1 patch #2459
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
…e#2456) `AuxiliaryTrainingWrapper.adapter_state_dict` now utilizes an external state dict for the computation of the module state dict to avoid problems with DeepSpeed (or FSDP) when dealing with distributed parameters. It is not possible to simply wrap everything in `GatheredParameters` context managers since doing that leads to a deadlock when running on more than one process (reasons unclear). Since transformers, or more specifically, accelerate already handles state dict fetching for the whole model, it is more economical to use that state dict and rewrite the methods that before depended on `state_dict()` calls. (cherry picked from commit 7279a9f)
This release is a patch release to release a fix for huggingface#2450 which might result in loss of `modules_to_save` when trained with deepspeed ZerO stage 3.
|
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. |
c42eb22
into
huggingface:patch-release-0.15.1
14 checks passed
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
This release is a patch release to release a fix for #2450 which might result in loss of
modules_to_savewhen trained with deepspeed ZeRO stage 3.Only the squashed commit from #2456 and the version bump are included in this PR.