[Feature] Integration with custom/local/private LLMs (llama/alpaca/vicuna/etc)

[deep-learning-dynamo]

As a java developer, I want to be able to use private/local LLM such as llama/alpaca/vicuna/etc:

• to save the cost (they are free to use)
• to not depend on external providers
• I want to fine-tune/customize it for my purposes and host it myself
• they are becoming better and faster
• they can run on commodity hardware
• this will enable summarizing/embedding large amounts of information (think big knowledge bases) cheaply/for free

See if this is a suitable solution: https://github.com/Nuked88/alpaca.http