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

try new vllm config

Browse source
This commit is contained in:
sirius0xdev 2026-05-04 21:53:08 +00:00
parent a066eea5b3
commit 680a16e4a8
2 changed files with 7 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
infrastructure/gpus/base/vllm-servers/rtx6000-vllm.yaml
View file

@ -38,14 +38,14 @@ 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=0.98
- --max-model-len=262144 - --max-model-len=262144
- --enable-auto-tool-choice - --enable-auto-tool-choice
- --enable-chunked-prefill - --disable-chunked-prefill
- --max-num-batched-tokens=4096 - --max-num-batched-tokens=65536
- --max-num-seqs=8 - --max-num-seqs=8
- --trust-remote-code - --trust-remote-code
- --dtype=auto - --dtype=fp8
- --enable-prefix-caching - --enable-prefix-caching
- --tool-call-parser=qwen3_xml - --tool-call-parser=qwen3_xml
- --reasoning-parser=qwen3 - --reasoning-parser=qwen3
@ -59,8 +59,8 @@ spec:
requests: requests:
nvidia.com/gpu: 1 nvidia.com/gpu: 1
memory: "120Gi" memory: "250Gi"
cpu: "8" cpu: "40"
startupProbe: startupProbe:
httpGet: httpGet:

2
infrastructure/gpus/staging/kustomization.yaml
View file

@ -1,5 +1,5 @@
apiVersion: kustomize.config.k8s.io/v1beta1 apiVersion: kustomize.config.k8s.io/v1beta1
kind: Kustomization kind: Kustomization
resources: resources:
#- ../base/vllm-servers/ - ../base/vllm-servers/
# - ../base/keda-gpu-scaling/ # - ../base/keda-gpu-scaling/