v1.2.0

New Features and Improvements

• Save Chat:

  • Users can now save their chat sessions in a *_chat_history.json file (based on Horizontal or Vertical view) located in the script's directory. This feature enhances usability by allowing users to reference past conversations easily.
  • Users can back up the *_chat_history.json file by renaming it with the current date. This allows for easy organization and retrieval of past chat sessions. Users can rename it back to Horizontal_chat_history.json or Vertical_chat_history.json as needed for future use. When you run the app, it reads from the json file and show you all the previous chats from that particular json file.
  • Privacy Assurance: All chat history is stored locally on the user's machine, ensuring that conversations are not saved on any remote server or LLM server, thereby protecting user privacy.

• Read More:

  • Added a "Read More" feature that allows users to expand and view additional content or context within the chat interface.

• Delete This Response:

  • Introduced functionality that enables users to delete specific responses from the chat, providing more control over the conversation history.

• Last Response Deletes Whole Conversation:

  • Implemented a feature where deleting the last response will clear the entire conversation, allowing users to remove a particular conversation completely.

• Adjusted Buttons:

  • Improved the layout and functionality of buttons within the interface for a more intuitive user experience.

• Removed Thinking Block:

  • The <think> tags have been removed from the output to reduce clutter and save space, resulting in a cleaner and more concise chat response.

v1.1.0

New Features and Improvements

• Multi-LLM Query Capability:

  • The script now supports querying multiple Large Language Models simultaneously. This enhancement allows users to obtain diverse responses and insights from different models in a single execution, improving the overall efficiency and effectiveness of the querying process.

• Color Coding for Model Names:

  • Added color coding to the output for model names, enhancing readability and allowing users to easily distinguish between different models in the results.

• Remove Model Functionality:

  • Introduced a feature to remove specific models from the querying process, providing users with greater control over which models are utilized.

• Display Running Model Name:

  • The script now displays the name of the currently running model, allowing users to track which model is being queried at any given time.

v1.0.0

Features

• Dynamic Model Selection: Users can select from available models fetched from a local server and add more models as needed.
• Prompt Input: A text area for entering prompts to be processed by the selected models.
• Response Display: After generating responses, the app displays the duration, evaluation count, and evaluation rate for each model, along with the generated text.
• Output Comparison: Users can easily compare the outputs of different models side by side, enabling a clear analysis of their performance on the same input prompt.