vrr/kvcache-ai-ktransformers
2
0
Fork 0
mirror of https://github.com/kvcache-ai/ktransformers.git synced 2025-09-08 13:39:48 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Update gate.py

Browse source
This commit is contained in:
Atream 2025-03-19 12:16:48 +08:00 • committed by GitHub
parent aa8f985f85
commit 44599229cd
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
ktransformers/operators/gate.py
View file

@ -125,7 +125,7 @@ class KMoEGate(BaseInjectedModule, KMoEGateBase):
# adapted from https://github.com/vllm-project/vllm/blob/c77620d22d43daa7e0440e6267cbdd83f849ac64/vllm/model_executor/layers/fused_moe/fused_moe.py#L1071 # adapted from https://github.com/vllm-project/vllm/blob/c77620d22d43daa7e0440e6267cbdd83f849ac64/vllm/model_executor/layers/fused_moe/fused_moe.py#L1071
# This is used by the Deepseek-V2 and Deepseek-V3 model # This is used by the Deepseek-V2 and Deepseek-V3 model
@torch.compile(dynamic=True) #@torch.compile(dynamic=True)
def grouped_topk(hidden_states: torch.Tensor, def grouped_topk(hidden_states: torch.Tensor,
gating_output: torch.Tensor, gating_output: torch.Tensor,
topk: int, topk: int,