chart/env/prod.yaml

image:
  repository: huggingface
  name: chat-ui

nodeSelector:
  role-huggingchat: "true"

tolerations:
  - key: "huggingface.co/huggingchat"
    operator: "Equal"
    value: "true"
    effect: "NoSchedule"

serviceAccount:
  enabled: true
  create: true
  name: huggingchat-prod

ingress:
  path: "/chat"
  annotations:
    alb.ingress.kubernetes.io/healthcheck-path: "/healthcheck"
    alb.ingress.kubernetes.io/listen-ports: "[{\"HTTP\": 80}, {\"HTTPS\": 443}]"
    alb.ingress.kubernetes.io/group.name: "hub-prod"
    alb.ingress.kubernetes.io/scheme: "internet-facing"
    alb.ingress.kubernetes.io/ssl-redirect: "443"
    alb.ingress.kubernetes.io/tags: "Env=prod,Project=hub,Terraform=true"
    alb.ingress.kubernetes.io/target-node-labels: "role-hub-utils=true"
    kubernetes.io/ingress.class: "alb"

ingressInternal:
  enabled: true
  path: "/chat"
  annotations:
    alb.ingress.kubernetes.io/group.name: hub-prod-internal-public
    alb.ingress.kubernetes.io/healthcheck-path: "/healthcheck"
    alb.ingress.kubernetes.io/listen-ports: "[{\"HTTP\": 80}, {\"HTTPS\": 443}]"
    alb.ingress.kubernetes.io/load-balancer-name: hub-prod-internal-public
    alb.ingress.kubernetes.io/target-group-attributes: deregistration_delay.timeout_seconds=30
    alb.ingress.kubernetes.io/target-node-labels: role-hub-lb=true
    alb.ingress.kubernetes.io/target-type: ip
    kubernetes.io/ingress.class: "alb"

envVars:
  ADMIN_CLI_LOGIN: "false"
  ALTERNATIVE_REDIRECT_URLS: '["huggingchat://login/callback"]'
  APP_BASE: "/chat"
  ALLOW_IFRAME: "false"
  COMMUNITY_TOOLS: "true"
  COOKIE_SAMESITE: "lax"
  COOKIE_SECURE: "true"
  ENABLE_ASSISTANTS: "true"
  ENABLE_CONFIG_MANAGER: "false"
  LOG_LEVEL: "debug"
  NODE_ENV: "prod"
  NODE_LOG_STRUCTURED_DATA: true
  OLD_MODELS: >
    [
      { "name": "bigcode/starcoder" },
      { "name": "OpenAssistant/oasst-sft-6-llama-30b-xor" },
      { "name": "HuggingFaceH4/zephyr-7b-alpha" },
      { "name": "openchat/openchat_3.5" },
      { "name": "openchat/openchat-3.5-1210" },
      { "name": "tiiuae/falcon-180B-chat" },
      { "name": "codellama/CodeLlama-34b-Instruct-hf" },
      { "name": "google/gemma-7b-it" },
      { "name": "meta-llama/Llama-2-70b-chat-hf" },
      { "name": "codellama/CodeLlama-70b-Instruct-hf" },
      { "name": "openchat/openchat-3.5-0106" },
      { "name": "meta-llama/Meta-Llama-3-70B-Instruct" },
      { "name": "meta-llama/Meta-Llama-3.1-405B-Instruct-FP8" },
      {
        "name": "CohereForAI/c4ai-command-r-plus-08-2024",
        "transferTo": "CohereLabs/c4ai-command-r-plus-08-2024"
      },
      {
        "name": "CohereForAI/c4ai-command-r-plus",
        "transferTo": "CohereLabs/c4ai-command-r-plus-08-2024"
      },
      {
        "name": "01-ai/Yi-1.5-34B-Chat",
        "transferTo": "CohereLabs/c4ai-command-r-plus-08-2024"
      },
      {
        "name": "mistralai/Mixtral-8x7B-Instruct-v0.1",
        "transferTo": "mistralai/Mistral-Small-3.1-24B-Instruct-2503"
      },
      {
        "name": "NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO",
        "transferTo": "NousResearch/Hermes-3-Llama-3.1-8B"
      },
      {
        "name": "mistralai/Mistral-7B-Instruct-v0.3",
        "transferTo": "mistralai/Mistral-Small-3.1-24B-Instruct-2503"
      },
      {
        "name": "microsoft/Phi-3-mini-4k-instruct",
        "transferTo": "microsoft/Phi-4"
      },
      {
        "name": "meta-llama/Meta-Llama-3.1-70B-Instruct",
        "transferTo": "meta-llama/Llama-3.3-70B-Instruct"
      },
      {
        "name": "Qwen/QwQ-32B-Preview",
        "transferTo": "Qwen/QwQ-32B"
      },
      {
        "name": "mistralai/Mistral-Nemo-Instruct-2407",
        "transferTo": "mistralai/Mistral-Small-3.1-24B-Instruct-2503"
      },
      {
        "name": "microsoft/Phi-3.5-mini-instruct",
        "transferTo": "microsoft/Phi-4"
      },
      {
        "name": "Qwen/Qwen2.5-Coder-32B-Instruct",
        "transferTo": "Qwen/QwQ-32B"
      },
      {
        "name": "meta-llama/Llama-3.2-11B-Vision-Instruct",
        "transferTo" : "Qwen/Qwen2.5-VL-32B-Instruct"
      }
    ]
  PUBLIC_ORIGIN: "https://huggingface.co"
  PUBLIC_SHARE_PREFIX: "https://hf.co/chat"
  PUBLIC_ANNOUNCEMENT_BANNERS: >
    [
      {
        "title": "Qwen 3 235B is available!",
        "linkTitle": "Try it out!",
        "linkHref": "https://huggingface.co/chat/models/Qwen/Qwen3-235B-A22B"
      }
    ]
  PUBLIC_APP_NAME: "HuggingChat"
  PUBLIC_APP_ASSETS: "huggingchat"
  PUBLIC_APP_COLOR: "yellow"
  PUBLIC_APP_DESCRIPTION: "Making the community's best AI chat models available to everyone."
  PUBLIC_APP_GUEST_MESSAGE: "Sign in with a free Hugging Face account to continue using HuggingChat."
  PUBLIC_APP_DATA_SHARING: 0
  PUBLIC_PLAUSIBLE_SCRIPT_URL: "/js/script.js"
  TASK_MODEL: >
        {
        "name": "NousResearch/Hermes-3-Llama-3.1-8B",
        "unlisted": true,
        "endpoints": [{"type" : "inference-client"}],
        "parameters": {
          "temperature": 0.1,
          "max_new_tokens": 256
        }
        }


  HF_ORG_ADMIN: '644171cfbd0c97265298aa99'
  HF_ORG_EARLY_ACCESS: '5e67bd5b1009063689407478'
infisical:
  enabled: true
  env: "prod-us-east-1"

autoscaling:
  enabled: true
  minReplicas: 2
  maxReplicas: 30
  targetMemoryUtilizationPercentage: "50"
  targetCPUUtilizationPercentage: "50"

resources:
  requests:
    cpu: 2
    memory: 4Gi
  limits:
    cpu: 4
    memory: 8Gi