vllm_no_ui_deployment.yml

---
version: "2.0"
services:
 vllm:
   image: vllm/vllm-openai:v0.4.0.post1
   expose:
     - port: 8000
       as: 8000
       to:
          - global: true

   command:
     - bash
     - "-c"
   args:
     - >-
       huggingface-cli download MaziyarPanahi/WizardLM-2-7B-AWQ && /usr/bin/python3 -m vllm.entrypoints.openai.api_server --model MaziyarPanahi/WizardLM-2-7B-AWQ
   env:
     - HUGGING_FACE_HUB_TOKEN=hf_********************
     - VLLM_API_KEY=MYPASSWORD

profiles:
 compute:
   vllm:
     resources:
       cpu:
         units: 16
       memory:
         size: 100Gi
       storage:
         - size: 80Gi
         - name: shm
           size: 10Gi
           attributes:
             class: ram
       gpu:
         units: 1
         attributes:
           vendor:
             nvidia:

 placement:
   dcloud:
     pricing:
       vllm:
         denom: uakt
         amount: 1000000
     
deployment:
 vllm:
   dcloud:
     profile: vllm
     count: 1