EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

[bugfix] fix fp32 trans nz (#5068)

Browse Source 
### What this PR does / why we need it?
fix fp32 trans nz error, disable fp32 dtype trans nz.

Signed-off-by: zzzzwwjj <1183291235@qq.com>
This commit is contained in:
zzzzwwjj
2025-12-16 15:04:31 +08:00
committed by GitHub
parent 11e6d6c291
commit 8c41770f1f
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm_ascend/utils.py
View File

@@ -85,7 +85,7 @@ def is_enable_nz(dtype: Optional[torch.dtype] = torch.int8,
and getattr(vllm_config.speculative_config, 'method',
None) in ("eagle", "eagle3"))
if dtype in [torch.float16, torch.bfloat16]:
if dtype in [torch.float16, torch.bfloat16, torch.float32]:
return _ENABLE_NZ if _IS_EAGLE_MODE else False
return _ENABLE_NZ