# SPDX-FileCopyrightText: Copyright 2026 SAP SE or an SAP affiliate company and cobaltcore-dev contributors
# SPDX-License-Identifier: Apache-2.0

apiVersion: thalamus.cloud/v1alpha1
kind: Model
metadata:
  name: smollm2-135m
  namespace: thalamus
spec:
  backend: native
  serving:
    engine:
      image: docker.io/vllm/vllm-openai-cpu:v0.22.1
      args:
        - --max-model-len=512
        - --max-num-seqs=1
        - --enforce-eager
      env:
        - name: VLLM_CPU_KVCACHE_SPACE
          value: "1"
        - name: VLLM_CPU_OMP_THREADS_BIND
          value: nobind
      resources:
        requests:
          cpu: "2"
          memory: 6Gi
        limits:
          cpu: "4"
          memory: 6Gi
    epp:
      image: ghcr.io/llm-d/llm-d-router-endpoint-picker:v0.9.0
      resources:
        requests:
          cpu: 100m
          memory: 256Mi
        limits:
          cpu: 200m
          memory: 512Mi
  weights:
    type: hf
    hf:
      repoId: HuggingFaceTB/SmolLM2-135M-Instruct
      tokenSecret:
        name: hf-token
        key: HF_TOKEN