Can this be downloaded like stable diffusion? How big are these models? How long does it take to run a "query" against them on a decent gaming PC?
LLaMA: A foundational, 65B-parameter large language model
61–70 of 209 posts
Re: LLaMA: A foundational, 65B-parameter large language model
#62Can this be downloaded like stable diffusion? How big are these models? How long does it take to run a "query" against them on a decent gaming PC?
you have to contact facebook for a copy, so not as open as stable diffusion
Re: LLaMA: A foundational, 65B-parameter large language model
#63>'We release all our models to the research community' Where release means you fill out a form and wait indefinitely. Also no use for commercial purposes - which means 95% of users are out - certainly doesn't democratize LLMs.
> doesn't democratize LLMs open source always leads to more democratization in the long run
This isn’t open source, at least by the usual definition that you can freely use it for any purpose and redistribute it
Re: LLaMA: A foundational, 65B-parameter large language model
#64Earlier quoted context omitted.
> We release all our models to the research community. This is yet more evidence for the "AI isn't a competitive advantage" thesis. State-of-the-Art is a public resource, so competing with AI offers no "moat".
The research is open, you still need 1) to have a good idea for a product that people want 2) lots of talent and resources to actually build and run everything
Actually no it's not available for commercial purposes or for anyone they do not approve.
Re: LLaMA: A foundational, 65B-parameter large language model
#65The most interesting snippet in the paper I think is this: > For instance, LLaMA-13B outperforms GPT-3 on most bench- marks, despite being 10× smaller. We believe that this model will help democratize the access and study of LLMs, since it can be run on a single GPU.
How about alignment/ability to answer prompt queries and chain of thought reasoning capabilities?
Without the fine tuning RLHF phase to make it like instructgpt I'm assuming it won't be as good as ChatGPT, is that right?
How hard would it be to fine tune the 65B model on commodity hardware?
Found answer here:
> Out-of-scope use cases LLaMA is a base, or foundational, model. As such, it should not be used on downstream applications without further risk evaluation and mitigation. In particular, our model has not been trained with human feedback, and can thus generate toxic or offensive content, incorrect information or generally unhelpful answers.
https://github.com/facebookresearch/llama/blob/main/MODEL_CA...
Re: LLaMA: A foundational, 65B-parameter large language model
#66Earlier quoted context omitted.
I don't see such grand stratagems as being a likely explanation. It seems more likely that a bunch of dorks are running around unsupervised trying their best to make lemonade with whatever budgets they are given as nobody can realistically manage AI researches. Or at least, it seems that much like in political governance, events are suddenly outpacing the ability of corporate to react. In the case of OpenAI it is a "…
this is a pretty biased and uninformed opinion. Pretty condescending to call it a "bunch of dorks...running around unsupervised."
Check out https://news.ycombinator.com/newsguidelines.html
Not sure how calling out comment as "biased", "uninformed" or "condescending" is helping.
Re: LLaMA: A foundational, 65B-parameter large language model
#67Earlier quoted context omitted.
> doesn't democratize LLMs open source always leads to more democratization in the long run
This isn’t open source, at least by the usual definition that you can freely use it for any purpose and redistribute it
Yeah. I was hoping to be able to use the model in koboldai, but no dice.
Re: LLaMA: A foundational, 65B-parameter large language model
#68Can this be downloaded like stable diffusion? How big are these models? How long does it take to run a "query" against them on a decent gaming PC?
This can't, but there's some software, KoboldAi, which lets you download and run other LLMs
Re: LLaMA: A foundational, 65B-parameter large language model
#69LLaMA is short for "Large Language Model Meta AI" -- I don't get how, but I guess the "fakeronym" was a brandable, recognizable word, and close enough that they went with it?
Re: LLaMA: A foundational, 65B-parameter large language model
#70The most interesting snippet in the paper I think is this: > For instance, LLaMA-13B outperforms GPT-3 on most bench- marks, despite being 10× smaller. We believe that this model will help democratize the access and study of LLMs, since it can be run on a single GPU.
How about alignment/ability to answer prompt queries and chain of thought reasoning capabilities? Without the fine tuning RLHF phase to make it like instructgpt I'm assuming it won't be as good as ChatGPT, is that right? How hard would it be to fine tune the 65B model on commodity hardware? Found answer here: > Out-of-scope use cases LLaMA is a base, or foundational, model. As such, it should not be used on downstrea…
Yes but fine tuning for RL is not expected to be hard. You're essentially limited by how much human feedback is available, so it's very different from training the foundational model on random bulk data.