Skip to content

Response format #8

@Raygama

Description

@Raygama

Create a module that format the response from the LLM into the one with emotion and the LLM text

  • Define response schema: e.g. { "text": "...", "emotion": "neutral" }.
  • Instruct LLM (via prompts) to output JSON.
  • Implement parsing logic and fallbacks when the LLM misformats.
  • Map emotion labels (string) using EmotionMapper.

Metadata

Metadata

Assignees

No one assigned

    Labels

    AI-ServiceIssues that relate to the AI side of AURA

    Type

    Projects

    Status

    Ready

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions