biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Kernel] Add cuda kernel for gpt_oss activation (#22951)

Browse Source 
Signed-off-by: Jee Jee Li <pandaleefree@gmail.com>
This commit is contained in:
Jee Jee Li
2025-08-17 13:03:24 +08:00
committed by GitHub
parent 87f48623a5
commit 4d4061b6e7
9 changed files with 157 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/model_executor/layers/quantization/utils/mxfp4_utils.py
View File

@@ -61,14 +61,14 @@ def _can_support_mxfp4(use_grouped_topk: bool = False,
e_score_correction_bias: Optional[torch.Tensor] = None,
apply_router_weight_on_input: bool = False,
scoring_func: str = "softmax",
activation: str = "swiglu_oai",
activation: str = "swigluoai",
expert_load_view: Optional[torch.Tensor] = None,
logical_to_physical_map: Optional[torch.Tensor] = None,
logical_replica_count: Optional[torch.Tensor] = None):
return not (use_grouped_topk or topk_group or num_expert_group
or expert_map or custom_routing_function
or e_score_correction_bias or apply_router_weight_on_input
or scoring_func != "softmax" or activation != "swiglu_oai"
or scoring_func != "softmax" or activation != "swigluoai"
or expert_load_view or logical_to_physical_map
or logical_replica_count)