fix llama lm_head load wrong weight error #4259
Closed
Conversation
This file contains 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
baodii
requested review from
RezaYazdaniAminabadi,
jeffra,
mrwyattii,
awan-10,
cmikeh2 and
arashb
as code owners
|
I think it's kind of duplicated with #4206 . But I don't know which way is better. |
This was referenced Sep 7, 2023
|
same as #4206 |
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Outdated suggestions cannot be applied.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.
Hi,
I have found that when we do AutoTP shard loading, if there are more than one checkpoint file, we will get a wrong lm_head weight in llama model.
Because when we replace module we will use embedding_weight as lm_head.weight in advance. In other word, we use embedding_weight as lm_head.weight before we load lm_head.weight when we have more than one checkpoint file.
So I add is_last_cp flag to avoid this situation.