this post was submitted on 04 Oct 2023
27 points (96.6% liked)

Free Open-Source Artificial Intelligence

2889 readers
1 users here now

Welcome to Free Open-Source Artificial Intelligence!

We are a community dedicated to forwarding the availability and access to:

Free Open Source Artificial Intelligence (F.O.S.A.I.)

More AI Communities

LLM Leaderboards

Developer Resources

GitHub Projects

FOSAI Time Capsule

founded 1 year ago
MODERATORS
[email protected]
[email protected]
27
Mistral 7B Megathread (lemmy.world)
submitted 1 year ago* (last edited 1 year ago) by [email protected] to c/[email protected]
5 comments fedilink hide all child comments
 

Starting a Mistral Megathread to aggregate resources.

This is my new favorite 7B model. It is really good for what it is. I am excited to see what we can tune together. I will be using this thread as a living document, expect a lot of changes and notes, revisions and updates.

Let me know if there's something in particular you want to see here. I will be adding to this thread throughout my fine-tuning journey with Mistral.

Mistral Model Megathread

Key

  • Link #1 - Base Model
  • Link #2 - Instruct Model

Quantized Base Models from TheBloke

GPTQ

GGUF

AWQ

Quantized Samantha Models from TheBloke

GPTQ

GGUF

AWQ

Quantized Kimiko Models from TheBloke

GPTQ

GGUF

AWQ

Quantized Dolphin Models from TheBloke

GPTQ

GGUF

AWQ

Quantized Orca Models from TheBloke

GPTQ

GGUF

AWQ

Quantized Airoboros Models from TheBloke

GPTQ

GGUF

AWQ

If you like to run any of the quantized/optimized models from TheBloke, do visit the full model pages from each of the quantized model cards to see and support the developers of each fine-tuned model.

top 5 comments
sorted by: hot top controversial new old
[–] [email protected] 3 points 1 year ago (1 children)

Nice list, thanks! What are some use-cases people are using 7B models for?

[–] [email protected] 5 points 1 year ago* (last edited 1 year ago) (1 children)

I am actively exploring this question.

So far - it’s been the best performing 7B model I’ve been able to get my hands on. Anyone running consumer hardware could get a GGUF version running on almost any dedicated GPU/CPU combo.

I am a firm believer there is more performance and better quality of responses to be found in smaller parameter models. Not too mention interesting use cases you could apply fine-tuning an ensemble approach.

A lot of people sleep on 7B, but I think Mistral is a little different - there’s a lot of exploring to be had finding these use cases but I think they’re out there waiting to be discovered.

I’ll definitely report back on how the first attempt at fine-tuning this myself goes. Until then, I suppose it would be great for any roleplay or basic chat interaction. Given it’s low headroom - it’s much more lightweight to prototype with outside of the other families and model sizes.

If anyone else has a particular use case for 7B models - let us know here. Curious to know what others are doing with smaller params.

[–] [email protected] 5 points 1 year ago

That's fair, I think chat/roleplay are great use cases.

I also think some of these lightweight models might make for interesting personal recommendation/categorization engines, etc. In my experiments with using models to categorize credit card transaction statements ala Mint, only GPT4 was able to do a good job out of the box. I bet a small model could do quite well with fine tuning though.

Another thought I had was to make some sort of personal recommendation engine, so you could export your Netflix/Spotify likes and have it recommend movies or music that you might enjoy, etc. I suppose it's still early days for those kind of uses for open source models!

[–] [email protected] 3 points 1 year ago (1 children)

Looks like an interesting model. I couldn't find it on their website, do you know what the training data was for this model?

[–] [email protected] 2 points 1 year ago* (last edited 1 year ago)

https://mistral.ai/news/announcing-mistral-7b/

They don't publish the training dataset. It's a secret. There are open bugreports on their Github, HuggingFace #8, #10, #38 and i think someone said so explicitly on their Discord.