forked from EngineX-Cambricon/enginex-mlu370-vllm
5 lines
110 B
JSON
5 lines
110 B
JSON
{
|
|
"use_cudagraph": true,
|
|
"non_cudagraph_ops": ["silly.attention"],
|
|
"cudagraph_copy_inputs": true
|
|
} |