Mistral 7B Model is out - Apache Licensed

So Mistral 7B is out

https://mistral.ai/news/announcing-mistral-7b/In case you missed it, they announced this on Twitter with just a Torrent Magnet link


//

AFAIK, only OpenAI seems to have productivised their models with the Playground, API and ChatGPT

I got a chance to play with Databricks' Dolly - integrated into my workflow. Can't wait to get my hands dirty with Mistral.

Quick Tip - If you have a model at your disposal, you can use it effectively to reverse engineer every "algorithm" - ahem... ranking... etc.

Have you tried to run it locally or on the Cloud, how are the speed/costs?
Any suggestions on how to better deploy Mistral or Llama 2 on the Cloud?

I am looking at DigitalOcean Paperspace, their new AI/ML-oriented cloud hosting service + https://huggingface.co/TheBloke/Mistral-7B-Instruct-v0.1-GGUF

But everything is still in the early stages and changing fast... One day Llama2 is best, now Mistral, a new open-source project comes along, etc.
All I want is as powerful as an OpenAI model running on my own server, that I can utilize for my AI apps...

Just found out about mistral 7b and ready to get my hands dirty.

I'm try the 8x7b with lm studio, its impressive

Wow Mistral 8x7b... I want to take my OpenAI gpt4 assistant conversations and fine tune 8x7b with it. Anyone know a like no code platform that fine tuned convoes like that?