[CI] Stabilize GLM-5.2 PCP evaluation (#51015)
Signed-off-by: khluu <khluu000@gmail.com> Co-authored-by: OpenAI Codex <codex@openai.com>
This commit is contained in:
co-authored by
OpenAI Codex
parent
05b7876f50
commit
a5149b2fee
@@ -13,5 +13,6 @@ server_args: >-
|
||||
--enable-expert-parallel
|
||||
--kv-cache-dtype fp8
|
||||
env:
|
||||
PYTORCH_CUDA_ALLOC_CONF: "expandable_segments:True"
|
||||
VLLM_LOGGING_LEVEL: "DEBUG"
|
||||
VLLM_USE_V2_MODEL_RUNNER: "1"
|
||||
|
||||
Reference in New Issue
Block a user