[Question]:

[mburnwal88]

What is your question?

I have a question regarding the usage of the ModelSpec within the librechat.yaml config file for version 0.7.6. While we are not using ModelSpec, we are encountering issue "Something went wrong. Here's the specific error message we encountered: Encountered an error while attempting to respond: Cannot read properties of null (reading 'call')" for dalle image creation model. Currently, our endpoint configuration consists of azureOpenAI and custom, and by default, the DALL·E plugin is available in the dropdown without any explicit configuration. However, when we try to use modelSpecs there is mandatory field list where I need to listdown all models so in UI for specific endpoint there is no specific dropdown to select model. In UI all the models are appearing in same list but we are expecting we should have option first to select endpoints like Azure Open AI or custom then we can choose model according to our choice. below I have attached the screen shot as well as librechat.yaml file configuration.

More Details

endpoints:
  custom:
    - name: "OpenSources & Custom Models"
      apiKey: "fake"
      baseURL: "https://custom_cloud.com"
      models:
        default: [
          "llava-llama3:latest",
          "codegemma:instruct",
        ]
      titleConvo: true
      titleModel: "current_model"
      summarize: false
      summaryModel: "current_model"
      forcePrompt: false
      modelDisplayLabel: "EXPERIMENTAL-Ollama"
      addParams:
        temperature: 0

  azureOpenAI:
    plugins: true
    groups:
      - group: "general"
        apiKey: "fake_api_key"
        instanceName: "genaillmrepo"
        assistants: false
        version: "2024-02-15-preview"
        baseURL: "https://azureopenai_bas_url.com"
        models:
          gpt-3.5-turbo:
            deploymentName: "gpt-35-turbo-16k"
      - group: "vision2"
        apiKey: "fake_api_key"
        instanceName: "gpt4-vision-4"
        assistants: false
        version: "2024-02-15-preview"
        baseURL: "https://azureopenai_bas_url.com"
        models:
          gpt-4o:
           deploymentName: "gpt-4o"

modelSpecs:
  list:
    - name: 'Azure GPT-3.5 Turbo'
      label: 'Azure OpenAI'
      preset:
        endpoint: 'azureOpenAI'
        models: 'gpt-3.5-turbo'
      config:
        maxTokens: 4096
        temperature: 0.2
    - name: 'Azure GPT-4o'
      label: 'Azure OpenAI'
      preset:
        endpoint: 'azureOpenAI'
        models: 'gpt-4o'
      config:
        maxTokens: 4096
        temperature: 0.2
    - name: 'OpenSources LLava-LLaMA3'
      label: 'OpenSources'
      preset:
        endpoint: 'custom'
        models: 'llava-llama3:latest'
      config:
        maxTokens: 4096
        temperature: 0.2
    - name: 'OpenSources CodeGemma'
      label: 'OpenSources'
      preset:
        endpoint: 'custom'
        models: 'codegemma:instruct'
      config:
        maxTokens: 4096
        temperature: 0.2
    - name: "Generate Images using DALLE"
      label: "Image Generation Tool (DALLE)"
      description: "Generate images using DALLE"
      preset:
        default: false
        endpoint: "gptPlugins"
        model: "gpt-4o"
        modelLabel: "Image Generation"
        tools: ["dalle-3"]
        agentOptions:
          model: "gpt-4o"
        greeting: |
          This assistant will generate images based on your query using DALLE3
        promptPrefix: |
          Taking the users input, generate an image using dalle

What is the main subject of your question?

No response

Screenshots

No response

Code of Conduct

• I agree to follow this project's Code of Conduct