From 812c374f7afa3d67a03931dad4858516703942c2 Mon Sep 17 00:00:00 2001 From: dylanyunlon Date: Fri, 7 Aug 2026 07:22:52 +0000 Subject: [PATCH] =?UTF-8?q?fix(critical):=20sync=20baseline.muh=20max=5Fmo?= =?UTF-8?q?del=5Flen=3D100000=20gpu=5Fmem=3D0.90=20=E2=80=94=20match=20com?= =?UTF-8?q?putility-run.yaml?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Root cause of job 105 scoring 0.0: - baseline.muh had max_model_len=256000 + gpu_memory_utilization=0.95 - computility-run.yaml had the safe values (100000 + 0.90) - Platform scheduler sent baseline.muh values to docker run command - Result: OOM on KV cache allocation → service crash → 881/881 Connection refused Diagnosis from submit日志: - benchmark-agent marked success (model loaded OK) - But service crashed before evaluation started - All 881 replay requests → Connection refused - All 5 opencompass benchmarks → 0.0 (aime, gpqa, hle, simpleqa, longbench) Fix: sync baseline.muh to match computility-run.yaml safe values --- baseline.muh | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/baseline.muh b/baseline.muh index 839af03e..8b63be15 100644 --- a/baseline.muh +++ b/baseline.muh @@ -16,8 +16,8 @@ vllm: model_path: /model served_model_name: llm - max_model_len: 256000 - gpu_memory_utilization: 0.95 + max_model_len: 100000 + gpu_memory_utilization: 0.90 tensor_parallel: 4 max_num_seqs: 2 max_num_batched_tokens: 4096