EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

Revert "[feature] Rework Ascend NPU graph support" (#9385)

Browse Source
This commit is contained in:
Even Zhou
2025-08-20 15:35:10 +08:00
committed by GitHub
parent 1ec9769753
commit de2dd73831
18 changed files with 81 additions and 546 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
benchmark/kernels/fused_moe_triton/benchmark_torch_compile_fused_moe.py
View File

@@ -9,7 +9,7 @@ from transformers import AutoConfig
from sglang.srt.layers.moe.fused_moe_triton.fused_moe import (
fused_moe as fused_moe_triton,
)
from sglang.srt.model_executor.graph_runner import set_torch_compile_config
from sglang.srt.model_executor.cuda_graph_runner import set_torch_compile_config
def get_model_config(model_name: str, tp_size: int):