Implement custom kernel for LLaMA rotary embedding (#14)

This commit is contained in:
Woosuk Kwon
2023-03-30 11:04:21 -07:00
committed by GitHub
parent 80a2f812f1
commit 88c0268a18
10 changed files with 318 additions and 69 deletions
+1 -1
View File
@@ -12,7 +12,7 @@ from cacheflow.parallel_utils.tensor_parallel import gather_from_tensor_model_pa
class Sampler(nn.Module):
def __init__(self) -> None:
super(Sampler, self).__init__()
super().__init__()
def forward(
self,