ENH: Add default target layers for gemma2 architecture #2078
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
Google's gemma 2 models have a slightly different architecture than gemma 1 and thus a different model_type attribute. This PR adds default target_layer for gemma 2 that correspond to the default target_layer of gemma 1.
|
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. |
SunMarc
approved these changes
Sep 19, 2024
src/peft/utils/constants.py
Outdated
Comment on lines
74
to
75
| "gemma": ["input_layernorm", "post_attention_layernorm", "norm"], | ||
| "gemma2": ["input_layernorm", "norm"], |
There was a problem hiding this comment.
Is it normal that they are not the same since you said that it was the same layers as gemma1 in the description of the PR.
There was a problem hiding this comment.
Not sure what I did there, I checked again and gemma2 has in fact more layer norm layers than gemma1, not fewer. I updated the targets accordingly.
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Outdated suggestions cannot be applied.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.
Google's gemma 2 models have a slightly different architecture than gemma 1 and thus a different
model_typeattribute. This PR adds defaulttarget_layers for gemma 2 that correspond to the defaulttarget_layers of gemma 1.