You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have searched to see if a similar issue already exists. Yes, there is no similar question here
Is your feature request related to a problem? Please describe.
Currently, the ChatbotInterface component in Gradio primarily focuses on displaying the final output of language models. This is insufficient for models with reasoning capabilities like gemini-2.0-flash-thinking-exp-1219. Currentlly, these models output both the reasoning process and the final answer in a single, undifferentiated text stream, making it difficult for users to quickly grasp the conclusion while still having access to the reasoning behind it.
Describe the solution you'd like
I propose enhancing the ChatbotInterface to support a structured display of models with reasoning capabilities. The key improvement is to display the reasoning process within a collapsible panel, similar to an Accordion panel. This allows users to focus on the final answer initially, but easily expand the panel to view the detailed reasoning steps when needed.
Additional context
Example using `gemini-2.0-flash-thinking-exp-1219
Imagine a user asks: "What is the capital of France?"
Current Display:
The user is asking for the capital city of France. I know this is a common piece of general knowledge. The answer is Paris.
Paris.
Proposed Enhanced Display (with reasoning shown, initially collapsed as an accordion):
Final Answer:
Paris.
[+ Show Reasoning] (Clicking this expands the accordion panel below)
Reasoning:
1. The user is asking for the capital city of France.
2. I know this is a common piece of general knowledge.
3. The answer is Paris.
This enhancement would significantly improve the usability of Gradio for showcasing and interacting with state-of-the-art language models, providing better explainability. I really hope the team can consider this and add support for mainstream reasoning models, thank you
The text was updated successfully, but these errors were encountered:
Yes, there is no similar question here
Is your feature request related to a problem? Please describe.
Currently, the
ChatbotInterface
component in Gradio primarily focuses on displaying the final output of language models. This is insufficient for models with reasoning capabilities likegemini-2.0-flash-thinking-exp-1219
. Currentlly, these models output both the reasoning process and the final answer in a single, undifferentiated text stream, making it difficult for users to quickly grasp the conclusion while still having access to the reasoning behind it.Describe the solution you'd like
I propose enhancing the
ChatbotInterface
to support a structured display of models with reasoning capabilities. The key improvement is to display the reasoning process within a collapsible panel, similar to anAccordion
panel. This allows users to focus on the final answer initially, but easily expand the panel to view the detailed reasoning steps when needed.Additional context
Example using `gemini-2.0-flash-thinking-exp-1219
Imagine a user asks: "What is the capital of France?"
Current Display:
Proposed Enhanced Display (with reasoning shown, initially collapsed as an accordion):
This enhancement would significantly improve the usability of Gradio for showcasing and interacting with state-of-the-art language models, providing better explainability. I really hope the team can consider this and add support for mainstream reasoning models, thank you
The text was updated successfully, but these errors were encountered: