added query-key norm to accomodate OLMo2 #1894
Open
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
| k = k.reshape(B, T, -1) # (B, T, nh_k * hs) | ||
| k = self.norm_k(k) | ||
| k = k.view(B, T, self.config.n_query_groups, self.config.head_size) | ||
|
|
There was a problem hiding this comment.
Is there a reason why you do this normalization here and not right after the qkv.split?
If you move it there, you will not have to do .reshape and .view again.
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Outdated suggestions cannot be applied.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.
https://github.com/huggingface/transformers/blob/main/src/transformers/models/olmo2/modeling_olmo2.py
OLMo2 applies RMSNorm to the q and k matrices in its attention layer, something that is not yet supported by litgpt's architecture.
To support the addition of OLMo2, this PR adds an option to norm the q and k matrices via the
config.norm_qkoption which defaults to False.Currently, the method for qk norm is assumed to follow the overall norm class.