diff --git a/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml b/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml index 070b1e7..5f613cf 100644 --- a/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml +++ b/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml @@ -41,7 +41,7 @@ spec: #- --enforce-eager - --max-model-len=136876 - --enable-auto-tool-choice - - --kv-cache-dtype=fp8 + - --kv-cache-dtype=bfloat16 - --max-num-batched-tokens=32768 - --max-num-seqs=32 - --block-size=32