rag_api icon indicating copy to clipboard operation
rag_api copied to clipboard

can't reuse files

Open jerowe opened this issue 1 year ago • 2 comments

Hello. Thanks for the great project!

I'm having an issue where I can upload a file and query it once.

The file persists in the file tab, but if I click on the file to try and reuse it I get an error saying "File size limit exceeded for endpoint : 0MB"

Screenshot 2024-10-01 at 1 14 51 PM

I did a fresh pull on all the docker images today. Any ideas for me?

jerowe avatar Oct 01 '24 17:10 jerowe

Hi thanks for checking out the project. I can’t reproduce the issue, do you have a librechat.yaml file configured? If so could you share what you have?

danny-avila avatar Oct 01 '24 17:10 danny-avila

# For more information, see the Configuration Guide:
# https://docs.librechat.ai/install/configuration/custom_config.html

# Configuration version (required)
version: 1.0.3

# Cache settings: Set to true to enable caching
cache: true

fileConfig:
  endpoints:
    default:
      disabled: false
      fileLimit: 5
      fileSizeLimit: 10
      totalSizeLimit: 50
      supportedMimeTypes:
        - "image/.*"
        - "application/pdf"
    assistants:
      disabled: false
      fileLimit: 5
      fileSizeLimit: 10
      totalSizeLimit: 50
      supportedMimeTypes:
        - "image/.*"
        - "application/pdf"
    openAI:
      disabled: true


# Definition of custom endpoints
endpoints:
  assistants:
    disableBuilder: true

  custom:
    - name: "KNOWLEDGEBASE_1"  # Unique name for the endpoint
      apiKey: "API_KEY"
      baseURL: "http://flask-api:3000/knowledgebase/"
      models:
        default:
          - anthropic.claude-3-5-sonnet-20240620-v1:0
          - anthropic.claude-3-haiku-20240307-v1:0
          - anthropic.claude-3-sonnet-20240229-v1:0
          - anthropic.claude-v2
          - mistral.mistral-large-2402-v1:0
          - meta.llama2-70b-chat-v1
        fetch: false  # Defaults to false.
      titleConvo: true  # Set to true to enable title conversation
      titleMethod: "functions"  # Defaults to "completion" if omitted.
      titleModel: "anthropic.claude-3-haiku-20240307-v1:0"  # Defaults to "gpt-3.5-turbo" if omitted.
      summarize: false
      summaryModel: "anthropic.claude-3-haiku-20240307-v1:0"  # Defaults to "gpt-3.5-turbo" if omitted.
      forcePrompt: false
      addParams:
        stream: false
        knowledge_base_id: 'ABCD'


    - name: "CHAT"  # Unique name for the endpoint"
      apiKey: "API_KEY"
      baseURL: "http://flask-api:3000/chat/"
      models:

        default:
          - anthropic.claude-3-5-sonnet-20240620-v1:0
          - anthropic.claude-3-haiku-20240307-v1:0
          - anthropic.claude-3-sonnet-20240229-v1:0
          - anthropic.claude-v2
          - mistral.mistral-large-2402-v1:0
          - meta.llama2-70b-chat-v1
        fetch: false  # Defaults to false.
      titleConvo: true  # Set to true to enable title conversation

      titleMethod: "functions"  # Defaults to "completion" if omitted.
      titleModel: "anthropic.claude-3-haiku-20240307-v1:0"  # Defaults to "gpt-3.5-turbo" if omitted.
      summarize: false
      #      summaryModel: "kb-1"  # Defaults to "gpt-3.5-turbo" if omitted.
      summaryModel: "anthropic.claude-3-haiku-20240307-v1:0"  # Defaults to "gpt-3.5-turbo" if omitted.
      forcePrompt: false
      #      modelDisplayLabel: "Terremoto KB"  # Default is "AI" when not set.
      addParams:
        safe_prompt: true # This field is specific to Mistral AI: https://docs.mistral.ai/api/
        stream: false
    - name: "KNOWLEDGEBASE_1"
      apiKey: "API_KEY"
      baseURL: "http://flask-api:3000/knowledgebase/"
      models:
        default:
          - anthropic.claude-3-5-sonnet-20240620-v1:0
          - anthropic.claude-3-haiku-20240307-v1:0
          - anthropic.claude-3-sonnet-20240229-v1:0
          - anthropic.claude-v2
          - mistral.mistral-large-2402-v1:0
          - meta.llama2-70b-chat-v1

        fetch: false  # Defaults to false.
      titleConvo: true  # Set to true to enable title conversation
      titleMethod: "functions"  # Defaults to "completion" if omitted.
      titleModel: "anthropic.claude-3-haiku-20240307-v1:0"  # Defaults to "gpt-3.5-turbo" if omitted.
      summarize: false
      summaryModel: "anthropic.claude-3-haiku-20240307-v1:0"  # Defaults to "gpt-3.5-turbo" if omitted.
      forcePrompt: false
      addParams:
        stream: false
        knowledge_base_id: "123456"

jerowe avatar Oct 03 '24 15:10 jerowe

@jerowe Are you still having the issue?

One thing I would check is to ensure that the file embedding is actually stored in whatever vector database you are using (pgvector is default). Librechat may be showing that it is 'stored' even though it is not in vector database. For whatever reason it may be getting deleted after you query it the first time.

Let me know if you any questions, thanks for bringing this up!

FinnConnor avatar Nov 18 '24 16:11 FinnConnor