fix: fix bug on kimi2.5 with dp2 and tp4 (#18604)
Co-authored-by: hanhaowen <hanhaowen@baidu.com>
This commit is contained in:
@@ -1198,7 +1198,7 @@ class TRTLLMMLABackend(FlashInferMLAAttnBackend):
|
||||
out=out,
|
||||
)
|
||||
else:
|
||||
out = torch.empty(
|
||||
out = torch.zeros(
|
||||
q.shape[0],
|
||||
q.shape[1],
|
||||
v.shape[2],
|
||||
|
||||
Reference in New Issue
Block a user