Query and display multiple LLM responses simultaneously - frontend

[tubamos]

Dependencies

• Integrate chat in the main app #210
• Implement multiselect option for LLM selection dropdown #211
• Present references in agent answers #174
• Create option to change LLM #177
• Integrate LLM4: Llama #188

Domain

app frontend

Description

To fulfill the fundamental requirement of comparing responses coming from multiple LLMs, we can implement the display of parallel chat bubbles, one for each response from a different LLM.

The chat bubbles with responses that will be created will be as many as the number of LLMs that have been selected using the checkbox in the dropdown.

Example

OpenAI has a similar UI in the web interface where there are two simultaneus responses produced that let the user select which is the best one. We could implement it in a similar way but for responses coming from different LLMs altogether.

Alternative:

Integrate multiple responses in the same chat response bubble.

Resources

• This might be relevant to the horizontal scrolling
  https://www.youtube.com/watch?v=cdP210MwnLw

• This might be relavant to the implementation in a single bubble:
  https://www.cometchat.com/docs/ui-kit/react-native/interactive-card-bubble

User Story

• As a user,
• I want to select multiple LLMs and receive their responses in parallel chat bubbles,
• so that I can analyze the differences in their answers.

Acceptance Criteria

• The chat interface displays multiple responses from different LLMs in separate parallel chat bubbles.
• Each LLM response is presented in a distinct bubble, clearly indicating the originating LLM.
• The user can scroll horizontally to navigate between the chat bubbles that host the responses.

Definition of Done

• The feature has been fully implemented.
• The feature has been manually tested and works as expected without critical bugs.
• The feature code is documented with clear explanations of its functionality and usage.
• The feature code has been reviewed and approved by at least one team member.
• The feature branches have been merged into the main branch and closed.
• The feature utility, function and usage have been documented in the respective project wiki on github.