v2.7.1 - better-er...er

Consolidated buttons to remove buttons_module.py.

Consolidated metrics scripts to only use metrics_bar.py instead of two scripts, and also started using Qthread instead of multiprocessing and annoying "pipes" between separate processes.

Added "gpu power" metrics if you have an Nvidia GPU. Linux and MacOS still can't display GPU metrics, but CPU/RAM metrics should b displayed properly now.

Added a pulldown for more prompt formatting presents, the default stilling being Llama2/Mistral.

Significant revisions to User Guide so read carefully for helpful tips.

Improved setup.py for people who have CUDA 12.1 installed. Now correctly directs the user.

v2.7 - smooth operations

Improved initialization.
Improved CPU/GPU metrics.
Improved threading/multiprocessing
Improved transcribe file.
Major refactoring
Killed some bugs
MacOS should work better now.

v2.6.5 - test embeddings

Added a new "test embeddings" checkbox which, when checked, will display the actual contexts retrieved from the vector database and NOT connect to LM Studio. This enables one to check the quality of the embeddings depending on the database creation settings such as chunk size etc.

v2.6.4 - color schemes

Added a bunch of custom color schemes, including one for colorblind people.

Updated Embedding Model portion for the User Guide. MAKE SURE AND READ.

Use the setup.py specifically attached to this release instead of the one within the src folder if you have CUDA 12.1 installed and not CUDA 11.8.

I also revised the User Guide after creating this release so used the files specifically attached hereto. Simply place them inside the user guide folder and replace the other files.

v2.6.3 - embedding model details

READ the Embedding Model portion of the revised User Guide.

Also, updated/removed/added awesome embedding models.

v2.6.2 - easy Windows install

Easier Windows installation - just check the updated installation instructions.

v2.6.1 - search audio files!

Added a feature in the "Tools" tab to transcribe audio files to .txt, which are then put into the database to be searched!

Added new Transcribe portion of User Manual to explain how to use new feature.

v2.6 - less vram more file-types

Use the attached config.yaml file instead! It didn't upload correctly the first time.

Fixed an error where 2 instances of the embedding model were sitting in memory and/or not being removed from memory. Overall better memory management now.

Added support for ODT, RTF, and Markdown files.

IMPORTANT: Added the option to change the compute device for creating versus querying the vector database. Read updated "Tips" within the User Manual.

Refactored code.

v2.5 - amazing effort

Update a lot so make sure and review the user manual.

Added settings for the transcription feature. Also, it will only load the whisper model when a user clicks "start recording" and immediately delete it from memory afterwards, addressing the issue that some people don't want it taking up memory.

Added settings for the database itself. Lots of fun to experiment with!

Improved print statements so you can see the number of chunks being created among other inner-workings of the program.

Reduced space on main GUI that buttons took to conserve space overall and display settings better.

Refactored some code for future expansion.

v2.3 - glorious

Added a slew of settings.

Significantly upgraded user guide.

After research, removed some embedding models and added others.

Refactored code to more heavily rely on config.yaml.