EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

Cleaning codes for speculative attention mode (#10149)

Browse Source
This commit is contained in:
Baizhou Zhang
2025-09-08 17:38:06 -07:00
committed by GitHub
parent 148022fc36
commit 8ad700f735
7 changed files with 14 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/sglang/srt/models/deepseek_v2.py
View File

@@ -1050,7 +1050,7 @@ class DeepseekV2AttentionMLA(nn.Module):
or forward_batch.forward_mode.is_draft_extend()
):
# Use the specified backend for speculative operations (both verify and draft extend)
if global_server_args_dict["speculative_attention_backend"] == "decode":
if global_server_args_dict["speculative_attention_mode"] == "decode":
attention_backend = global_server_args_dict["decode_attention_backend"]
else: # default to prefill
attention_backend = global_server_args_dict["prefill_attention_backend"]