sirius/gcloud-lab
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

dial in vllm

Browse source
This commit is contained in:
sirius0xdev 2026-05-10 03:02:51 +00:00
parent 4e2b0974f4
commit cfedb963e5
1 changed files with 5 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml
View file

@ -37,13 +37,13 @@ spec:
- --port=8000 - --port=8000
- --tensor-parallel-size=1 - --tensor-parallel-size=1
- --tokenizer-mode=hf - --tokenizer-mode=hf
- --gpu-memory-utilization=0.95 - --gpu-memory-utilization=0.90
- --enforce-eager #- --enforce-eager
- --max-model-len=65536 - --max-model-len=98304
- --enable-auto-tool-choice - --enable-auto-tool-choice
- --kv-cache-dtype=fp8 - --kv-cache-dtype=fp8
- --max-num-batched-tokens=65536 - --max-num-batched-tokens=32768
- --max-num-seqs=128 - --max-num-seqs=32
- --block-size=32 - --block-size=32
- --enable-chunked-prefill - --enable-chunked-prefill
- --trust-remote-code - --trust-remote-code