Awesome, thank you for sharing these tools. I will setup the materials and test it/use it, I believe using these will be 10000x times better then AI website chatbots )))What kind of code you working with?
I'd suggest you check out some options that you can run locally with easy setup. Theyve come a very long way very quickly on all these LLM's and groundbreaking updates happening daily. I'm using Vicuna 13B right now which outperforms all other models and is comparable to GPT-4 at 90% ( OpenAI gpt is who actually rated it, btw )
Vicuna 13b is only 7gb. I use ooga booga to run the LLM for a nice clean web ui. Uncensored, can train on more custom data sets and all kinds of posibilities. Its also free, and will run for you as is completely offline. Check it out
I run this on a spare laptop purely through cpu, no gpu. This field is changing drastically by the hour. You cant compare the progress to anything else we've ever seen before.
Check out the UI https://sky.cs.berkeley.edu/news/an...-gpt-4-with-90-chatgpt-quality-made-by-lmsys/
Vicuna LLM I use https://huggingface.co/anon8231489123/vicuna-13b-GPTQ-4bit-128g/tree/main
There is also minigpt4 ( which is actually not based on gpt model but everyone is comparing the quality results to it ) I'll be setting up soon. This one can accept images to analyze in many different ways. Even able to do the napkin wire website design and give you code like the openai demo showed.
You can play with their demo here: https://minigpt-4.github.io/
try one of the alternate demo links and the speed is slow because TONS of people testing it, but you can setup local.
Anyone not staying up to snuff on this and following and working with the technology is making a hugemistake imho
Just found it myself. Wow. I tried alpaca back in march - wasn't very impressed and now it's obsolete, the progress is crazy!
It is going so fast... what will it be in 1 year!?
