xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-14 17:26:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

[Qwen] Remove cuda hard-code in qwen3 next (#25243)

Browse Source 
Signed-off-by: Icey <1790571317@qq.com>
This commit is contained in:
Icey 2025-09-19 20:25:12 +08:00 committed by GitHub
parent 5089fd749c
commit 838d7116ba
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/model_executor/models/qwen3_next.py
View File

@ -306,7 +306,7 @@ class Qwen3NextGatedDeltaNet(nn.Module, MambaBase):
eps=self.layer_norm_epsilon, eps=self.layer_norm_epsilon,
group_size=None, group_size=None,
norm_before_gate=True, norm_before_gate=True,
device=torch.cuda.current_device(), device=current_platform.current_device(),
dtype=config.torch_dtype, dtype=config.torch_dtype,
) )