Update rtx6000-vllm.yaml

2026-05-02 08:36:04 -04:00 · 2026-05-02 08:36:04 -04:00 · 9a7904c219
commit 9a7904c219
parent 1aceefbedd
1 changed files with 1 additions and 1 deletions
--- a/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml
+++ b/infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml
@ -39,7 +39,7 @@ spec:
            - --tensor-parallel-size=1
            - --tokenizer-mode=hf 
            - --gpu-memory-utilization=0.95
-            - --kv-cache-dtype=fp8_e5m2
+            - --kv-cache-dtype=fp8
            - --max-model-len=256389
            - --enable-auto-tool-choice 
            - --enable-chunked-prefill