diff --git a/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml b/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml index 5cd1928..2baa422 100644 --- a/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml +++ b/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml @@ -42,7 +42,7 @@ spec: - --enable-auto-tool-choice - --kv-cache-dtype=bfloat16 - --max-num-batched-tokens=65536 - - --max-num-seqs=64 + - --max-num-seqs=32 - --block-size=32 - --language-model-only - --enable-chunked-prefill