xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-01-11 04:24:28 +08:00
Code Issues Packages Projects Releases Wiki Activity

[BugFix] Fix interleaved sliding window not set for Gemma3n (#21863)

Browse Source 
Signed-off-by: Yong Hoon Shin <yhshin@meta.com>
This commit is contained in:
Yong Hoon Shin 2025-07-29 16:34:19 -07:00 committed by GitHub
parent 176bbce1db
commit 9266d98048
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 14 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
vllm/config.py
View File

@ -723,11 +723,16 @@ class ModelConfig:
)
# Workaround for Gemma 2 which uses interleaved sliding window
# attention, but it's not specified in its config. TODO: remove this
# when Gemma 2 is fixed in Transformers.
# attention, but it's not specified in its config.
# TODO: remove this when Gemma 2 config updated in HuggingFace.
if self.hf_text_config.model_type == "gemma2":
self.hf_text_config.sliding_window_pattern = 2
# TODO: remove this when Gemma 3n config updated in HuggingFace.
if self.hf_text_config.model_type == "gemma3n_text":
# 4 sliding window attention followed by 1 full attention
self.hf_text_config.sliding_window_pattern = "LLLLG"
sliding_window = getattr(self.hf_text_config, "sliding_window", None)
sliding_window_pattern = getattr(self.hf_text_config,
"sliding_window_pattern", None)

9
vllm/model_executor/models/gemma3n.py
View File

@ -297,8 +297,13 @@ class Gemma3nAttention(nn.Module):
has_weight=False)
layer_idx = extract_layer_index(prefix)
if config.layer_types[layer_idx] == "sliding_attention":
self.sliding_window = config.sliding_window
is_sliding_window = (
getattr(config, "interleaved_sliding_window", None) is not None
and config.layer_types[layer_idx] == "sliding_attention")
if is_sliding_window:
self.sliding_window = config.interleaved_sliding_window
rope_theta = config.rope_local_base_freq
rope_scaling = {"rope_type": "default"}
else: