Development Pathways:

Iteration 1 (23/10/2024):

Making a good UI to interact with LLMs

Iteration 2 (24/10/2024):

Adding a feature to save the model and load it later

Iteration 3 (25/10/2024):

Loading the thread names based on their modification datetime

Iteration 4 (25/10/2024):

Delete threads using button

Iteration 5 (26/10/2024):

Feature to save the last used LLM with the thread
Open it automatically with that thread...

Iteration 6 (16/11/2024):

Changed UI
Added feature to Check and Stop Running Models
Added Modularity to the code
Added Image Support (LLama vision and Llava)

Iteration 7 (17/11/2024):

Completely modularized the threads codes
Refactored the image logic Completely

Iteration 8 (04/01/2024):

Save the images in local file, store new name (path) in thread json
Make the subprocess to copy the parsed image to the local folder

Iteration 9 (05/01/2024):

Updated requirements
Fixed compatibility with latest versions of ollama and streamlit

Future plans:

try except in the write_stream to save the half response at least in case of error
Update the docstring-s of functions in the app_threads file
Move the LLM model sidebar out of app.py
Integrate lang-chain in it