Files
kserve/docs/samples/llmisvc/opt-125m-cpu/llm-inference-service-facebook-opt-125m-cpu-no-scheduler.yaml
2026-05-21 21:44:35 +00:00

35 lines
827 B
YAML

apiVersion: serving.kserve.io/v1alpha1
kind: LLMInferenceService
metadata:
name: facebook-opt-125m-single
spec:
model:
uri: hf://facebook/opt-125m
name: facebook/opt-125m
replicas: 1
router:
route: { }
template:
containers:
- name: main
image: public.ecr.aws/q9t5s3a7/vllm-cpu-release-repo:v0.19.0
securityContext:
runAsNonRoot: false
env:
- name: VLLM_LOGGING_LEVEL
value: DEBUG
- name: VLLM_CPU_KVCACHE_SPACE
value: "1"
resources:
limits:
cpu: '1'
memory: 10Gi
requests:
cpu: '100m'
memory: 8Gi
livenessProbe:
initialDelaySeconds: 30
periodSeconds: 30
timeoutSeconds: 30
failureThreshold: 5