• 1 Post
  • 15 Comments
Joined 1 year ago
cake
Cake day: June 16th, 2023

help-circle







  • Ever time i see a post like this i ask the same thing and i have yet to receive answer.

    Why should i care?

    There are so many open source language models, all with different strengths and weaknesses. There are tools to run them on any OS with all kinds of different hardware requirements.

    This has been the case since before chatgpt came out and has exponentially blown up since.

    Gpt4all is just a single recent model. But in recent weeks it always gets the headlight under “run chatgpt at home”

    What does it do to stand out? Why would i use this and not one of the vicuna or llama models?

    Hugging face has a leaderboard for open source large language models.

    https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard

    If you are interested in running this tech at home, familiarize yourself with multiple models because they all will behave differently depending on your hardware and your needs.






  • Seems the model used is gpt4all but i have yet to see a good explanation on what gpt4all does that makes it seem like its a trending for consumers which leads me to believe it really is just people confusing its name as being actually comparable to gpt4.

    If you check the leaderboard on huggingface there are a whopping 37 open source large language models with better quality outputs then gpt4all.

    Any good llm interface that allows you to run them locally should allow you to run any of them and you should play around with multiple cause they might all perform at very different speeds depending on your system. Personally i have have used up to Wizard-Vicuna-13B which is listed more then 10 spots above gpt4all and can provides me a decent but dumb conversation in a reasonably slow speed.

    The biggest (64B) models will probably be too slow to 95% of consumers and will get you good gpt-3 like performance at best.

    Unless someone can tell me otherwise i don’t believe that running these for actual productive goals/more then playing around is something i can advise. And putting the focus on just the interface and not the model that does the work seems a bit salespeech like.