[Test] Stabilize MiniMax-H3 reference accuracy compile (#6688)

Signed-off-by: david6666666 <530634352@qq.com>
This commit is contained in:
WeiQing Chen authored and GitHub committed 2026-08-28 14:18:28 +08:00
1 parent ab8d45d3cc
commit ed837e350e
1 file changed
+4 -1
@@ -1,5 +1,5 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
# SPDX-FileCopyrightText: Copyright contributors to the vLLM-Omni project
from __future__ import annotations
@@ -119,6 +119,9 @@ def _model_name(model_env_var: str = MODEL_ENV_VAR, component: str = "FL2VA") ->
def _server_args() -> list[str]:
return [
"--trust-remote-code",
# Both official-reference requests use fixed shapes. Keep regional
# compilation static so cold Inductor codegen is stable across CI nodes.
"--no-diffusion-compile-dynamic",
"--num-gpus",
"4",
"--usp",