xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-04-23 13:17:04 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update fused_marlin_moe_fake

Browse Source 
Signed-off-by: mgoin <mgoin64@gmail.com>
This commit is contained in:
mgoin 2025-09-18 11:43:58 -07:00
parent d643d6a418
commit e8c78d992c
1 changed files with 4 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/model_executor/layers/fused_moe/fused_marlin_moe.py
View File

@ -212,6 +212,8 @@ def fused_marlin_moe(hidden_states: torch.Tensor,
def fused_marlin_moe_fake(hidden_states: torch.Tensor,
w1: torch.Tensor,
w2: torch.Tensor,
bias1: Optional[torch.Tensor],
bias2: Optional[torch.Tensor],
w1_scale: torch.Tensor,
w2_scale: torch.Tensor,
gating_output: torch.Tensor,
@ -220,9 +222,10 @@ def fused_marlin_moe_fake(hidden_states: torch.Tensor,
quant_type_id: int,
apply_router_weight_on_input: bool = False,
global_num_experts: int = -1,
activation: Optional[str] = "silu",
expert_map: Optional[torch.Tensor] = None,
global_scale1: Optional[torch.Tensor] = None,
global_scale2: Optional[torch.Tensor] = None,
expert_map: Optional[torch.Tensor] = None,
g_idx1: Optional[torch.Tensor] = None,
g_idx2: Optional[torch.Tensor] = None,
sort_indices1: Optional[torch.Tensor] = None,