How I made a custom GPT | With web search, knowledge database and character personas

Comparing current popular UI frameworks for running LLMs, then picking one to reverse engineer GPTs experience on your local PC with Docker!
Also, for added bonus - side-by-side comparison of LLaMA 2 7b and Zephyr 7b!
Chapters:
0:00 Intro
0:51 Comparing UIs
2:48 Dockerfile and docker compose
9:03 Custom GPTs crafting
15:15 What's next?
Gist:
gist.github.com/AIWintermuteA...

[@exploring-electronic]

Neat! You showed stable diffusion integration - can you make a video on that setup?

[@Hardwareai]

I can do everything xD let's see how many thumbs up this comment can get

[@BogdanMnikov]

Very well explained!

[@Hardwareai]

Glad it was helpful!

[@levbereggelezo]

Thank you for the tutorial! What are the other worthwhile language models for RAG?

[@Hardwareai]

Thank you! Well, for RAG it is important for the model to follow the prompts to the letter and not to be overly creative. Again, based on local LLaMA subreddit, OpenHermes 2.5 (also fine-tune of Mistral) and (larger) Tess Yi 34b 200k are recent recommendations!