Skip to content

Add a subtle fix for gemma 2 conversions #1701

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jul 19, 2024
Merged

Add a subtle fix for gemma 2 conversions #1701

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions keras_nlp/src/models/gemma/gemma_backbone.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -54,9 +54,9 @@ class GemmaBackbone(Backbone):
layer_norm_epsilon: float. The epsilon value user for every layer norm
in the transformer model.
dropout: float. Dropout probability for the Transformer encoder.
query_head_dim_normalize: boolean. Whether to normalize attention with
head dimension or hidden_dim/num_query_heads. Gemma2 uses the
second option. Defaults to True.
query_head_dim_normalize: boolean. If `True` normalize the query before
attention with `head_dim`. If `False`, normalize the query with
`hidden_dim / num_query_heads`. Defaults to True.
use_post_ffw_norm: boolean. Whether to normalize after the feedforward
block. Defaults to False.
use_post_attention_norm: boolean. Whether to normalize after the attention
Expand Down
5 changes: 4 additions & 1 deletion keras_nlp/src/utils/transformers/convert_gemma.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -59,7 +59,10 @@ def load_gemma_backbone(cls, preset, load_weights):
"hidden_dim": transformers_config["hidden_size"],
"intermediate_dim": transformers_config["intermediate_size"] * 2,
"head_dim": transformers_config["head_dim"],
"query_head_dim_normalize": False,
"query_head_dim_normalize": (
transformers_config["head_dim"]
== transformers_config["query_pre_attn_scalar"]
),
"use_post_ffw_norm": True,
"use_post_attention_norm": True,
"final_logit_soft_cap": transformers_config[
Expand Down