Skip to content

Fix Gemma model detection and Gemma-2 2B attn_types config #1227

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
jlarson4 merged 1 commit into from
Mar 31, 2026
Merged

Fix Gemma model detection and Gemma-2 2B attn_types config #1227

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
4 changes: 2 additions & 2 deletions transformer_lens/loading_from_pretrained.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -853,7 +853,7 @@ def convert_hf_model_config(model_name: str, **kwargs: Any):
else:
# 4B, 12B, 27B and medgemma are multimodal
architecture = "Gemma3ForConditionalGeneration"
elif "gemma-2" in official_model_name.lower():
elif "gemma-2-" in official_model_name.lower():
architecture = "Gemma2ForCausalLM"
elif "gemma" in official_model_name.lower():
architecture = "GemmaForCausalLM"
Expand Down Expand Up @@ -2009,7 +2009,7 @@ def convert_hf_model_config(model_name: str, **kwargs: Any):
"n_key_value_heads": 4,
"window_size": 4096,
"use_local_attn": True,
"attn_types": ["global", "local"] * 21, # Alternate global and local attn
"attn_types": ["global", "local"] * 13, # Alternate global and local attn
"attn_scores_soft_cap": 50.0,
"output_logits_soft_cap": 30.0,
"gated_mlp": True,
Expand Down
Loading