Thank You - Local Models for Translation

[stephanhixen]

Thank you for all the hard work.

I prefer to run the app locally and have used the install script to come up with a "DistroShelf" process that lets me download/deploy and run it on in more compatible environment (host OS is a immutable linux distro)

Along those lines, so far I do like the results of running AI translation through a local model rather then a cloud service. For now I just tweak the providers.py to add a entry pointing to localhost: (i like the results of the gemma3:12b-it-q4_K_M with my limited VRAM). I tried to figure out a more elegant way to update the code but its just not my expertise but I'd like to vote for official local model support.

[meizhong986]

Just release 1.8.0.
Give it a try and let me know.
The llm implementation is courtesy of @hyiip .

[stephanhixen]

WOW that implementation is great... we don't even need to run models manually with all integrated.
Categorizing LLM by vram and implementing a auto mode was also a really nice touch.

thanks to both of you. Works as expected.