diff --git a/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml b/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml index 0c4989e..b993e20 100644 --- a/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml +++ b/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml @@ -37,7 +37,7 @@ spec: - --port=8000 - --tensor-parallel-size=1 - --tokenizer-mode=hf - - --gpu-memory-utilization=0.98 + - --gpu-memory-utilization=0.95 - --max-model-len=131072 - --enable-auto-tool-choice - --kv-cache-dtype=fp8