Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Alpaca: A strong open-source instruction-following model

crfm.stanford.edu

Post reply on HN
Print view View original

81–90 of 313 posts

Re: Alpaca: A strong open-source instruction-following model

#81

Post by mlyle »

Traubenfuchs wrote: post #56
mlyle wrote: post #49

Earlier quoted context omitted.

On the other hand, LLM's without some degree of safety work will frequently advocate for genocide/eugenics; perpetuate terrible stereotypes; mock ethnic dialects of language; veer randomly into erotica; etc. Not things in "trigger warning" territory but stuff that is broadly considered unacceptable.

In private, you can ask it not to produce this kind of content. If you offer a public API it‘s your responsibility to restrain the LLM or do an automated acceptability analysis before publishing content. But the raw, open source code should not be constrained, castrated and sterilized.

> But the raw, open source code should not be constrained, castrated and sterilized.

Which is what we have now. But they are going to fine-tune it so that we can use it for various purposes without worrying too much it will go on a rant about "the blacks" again, which makes it a lot more useful for many use cases.

Re: Alpaca: A strong open-source instruction-following model

#82

Post by wkat4242 »

nickthegreek wrote: post #14

>We are waiting for huggingface to officially support the llama models (i.e. this PR to be merged) before we release a stable version of the finetuning code. Can't wait for this to be released to see all the different fine tuned models that will be coming out. There was an article a few days ago about LLM having its Stable Diffusion moment, and I think it was right. We went from LLama release > llama.cpp > Dalai and…

Yeah this whole industry seems to suddenly have moved from the pace of a pebble creek to a waterfall.

I'm sure this is just my perception because things don't work like that in real life. Is it really just a ton of slowly developed things happening to mature close together?

Re: Alpaca: A strong open-source instruction-following model

#83

Post by simonw »

loufe wrote: post #40
simonw wrote: post #28

This is why I think we're seeing a Stable Diffusion moment for LLMs: https://simonwillison.net/2023/Mar/11/llama/ Look at the timeline: 24th February 2023: LLaMA is announced, starts being shared with academic partners: https://research.facebook.com/publications/llama-open-and-ef... 2nd March: Someone posts a PR with a BitTorrent link to the models: https://github.com/facebookresearch/llama/pull/73 10th March: First…

Might I suggest looking the story between the 2nd and 10th of march? I've noticed Hacker News hasn't been following certain areas of the effort. A lot of great work had happened and continues to be happen in close conjunction with the text-generation-webui (seriously, most of the cutting edge with 4-bit GPTQ etc. has been closely tied to the project). > https://github.com/oobabooga/text-generation-webui/

Wow, yeah that's a VERY active project: https://github.com/oobabooga/text-generation-webui/graphs/co... - only started Dec 18, 2022 and already 22 contributors and 806 commits!

Re: Alpaca: A strong open-source instruction-following model

#84

Post by bilsbie »

isaacfung wrote: post #57
bilsbie wrote: post #48

What do they mean by instruction? Is it just regular LLM?

LLM just predicts the next token given the previous tokens(this can be trained without manual labelling by humans). Instruct GPT and ChatGPT use reinforcement learning from human feedback to align the model with human intents so it understands instructions. https://huggingface.co/blog/rlhf

Thanks. So what does the output look like without rlhf?

Re: Alpaca: A strong open-source instruction-following model

#85

Post by wkat4242 »

simonw wrote: post #28

This is why I think we're seeing a Stable Diffusion moment for LLMs: https://simonwillison.net/2023/Mar/11/llama/ Look at the timeline: 24th February 2023: LLaMA is announced, starts being shared with academic partners: https://research.facebook.com/publications/llama-open-and-ef... 2nd March: Someone posts a PR with a BitTorrent link to the models: https://github.com/facebookresearch/llama/pull/73 10th March: First…

I know, this is crazy!!

I can't fathom how development has suddenly seemed to accelerate.

Re: Alpaca: A strong open-source instruction-following model

#86

Post by gpm »

2bitencryption wrote: post #46
simonw wrote: post #28

This is why I think we're seeing a Stable Diffusion moment for LLMs: https://simonwillison.net/2023/Mar/11/llama/ Look at the timeline: 24th February 2023: LLaMA is announced, starts being shared with academic partners: https://research.facebook.com/publications/llama-open-and-ef... 2nd March: Someone posts a PR with a BitTorrent link to the models: https://github.com/facebookresearch/llama/pull/73 10th March: First…

Here is one question I have not seen answered yet: All the magic of "7B LLaMA running on a potato" seems to involve lowering precision down to f16 and then further quantizing to int4. Clearly this quantized model still outputs something resembling human language, at the very least. But I haven't seen anyone show what effect this quantizing has on the quality of the output. If the quality of the output is bad, it's un…

> All the magic of "7B LLaMA running on a potato" seems to involve lowering precision down to f16

LLaMa weights are f16s to start out with, no lowering necessary to get to there.

You can stream weights from RAM to the GPU pretty efficiently. If you have >= 32GB ram and >=2GB vram my code here should work for you: https://github.com/gmorenz/llama/tree/gpu_offload

There's probably a cleaner version of it somewhere else. Really you should only need >= 16 GB ram, but the (meta provided) code to load the initial weights is completely unnecessarily making two copies of the weights in RAM simultaneously. You could also lower vram requirements a bit more with a bit more work (I just made the smallest change possible to make it work)

Re: Alpaca: A strong open-source instruction-following model

#87

Post by londons_explore »

byteknight wrote: post #73
gamegoblin wrote: post #62

Earlier quoted context omitted.

Decrease in accuracy is negligible and decreases as model size increases. That is, larger models quantize even better than smaller models. https://arxiv.org/abs/2210.17323

Is this because averages are weighed less (less sensitive) as the total sample size increases?

Yes. In a dense everything to everything neural network layer, the number of 'inputs' to a node is proportional to the square root of the number of weights.

Therefore, assuming quantization noise is uncorrelated, as the number of weights doubles, the number of inputs goes up by sqrt(2), and the (normalized) noise goes down by a factor of 2*(sqrt(2)).

So, as a rule of thumb, you can remove 1 bit of precision of the weights for every 4x increase in the number of weights.

All this assumes weights and activations are uncorrelated random variables - which may not hold true.

Re: Alpaca: A strong open-source instruction-following model

#88

Post by ben_w »

margorczynski wrote: post #63
numeri wrote: post #54

Earlier quoted context omitted.

I personally think this is important completely aside from any cultural propriety/censoring, as it's the one of the only parts of AI ethics that we can currently directly work on and test. If we can't get models not to say racist or otherwise terrible things, we can't make any guarantees about our ability to control or guide some future AGI. A very much secondary reason I appreciate these (admittedly annoying) attemp…

> If we can't get models not to say racist or otherwise terrible things, we can't make any guarantees about our ability to control or guide some future AGI. This is a very bold assumption that the current LLMs function and "think" in the same way some future AGI would. They do not even reason, just make up words that fit some context - thus they "hallucinate". There is no reason the approach taken here by injecting s…

Paperclipping is just one example of one of many ways it can go wrong.

While I agree LLMs are unlikely to be the last word on AI, the fact we understand alignment so poorly that they spew random things, let alone any arguments about which words are acceptable[0], is a sign we have much foundational work to do.

Indeed, as I recall, one of the main researchers in this topic describes it as "pre paradigmatic" because we don't have a way to even compare the relative alignment of any two AI.

[0] personally, I suspect but cannot prove that tabooing certain words is a Potemkin village solution to the underlying social problems

Re: Alpaca: A strong open-source instruction-following model

#89

Post by wkat4242 »

nickthegreek wrote: post #34

Quick compare with this vs 7B LLama. Input: Tell me something about owls. Output 7B LLama: Asked by kimberly at 12:54 PM on May. 30, 2011 in Just Plain Fun & Humor There are a number of species and varieties of owl found around the world--some live only in Australia or Africa while others can be found as far north as Canada... The one thing that all have in common is their ability to adapt. Owls will find homes in pl…

That's really impressive. Thanks for doing this test.

Re: Alpaca: A strong open-source instruction-following model

#90

Post by famouswaffles »

margorczynski wrote: post #63
numeri wrote: post #54

Earlier quoted context omitted.

I personally think this is important completely aside from any cultural propriety/censoring, as it's the one of the only parts of AI ethics that we can currently directly work on and test. If we can't get models not to say racist or otherwise terrible things, we can't make any guarantees about our ability to control or guide some future AGI. A very much secondary reason I appreciate these (admittedly annoying) attemp…

> If we can't get models not to say racist or otherwise terrible things, we can't make any guarantees about our ability to control or guide some future AGI. This is a very bold assumption that the current LLMs function and "think" in the same way some future AGI would. They do not even reason, just make up words that fit some context - thus they "hallucinate". There is no reason the approach taken here by injecting s…

It's not a bold assumption. t's the only assumption. We can't control the output of llms completely because we don't know how they generate. Nobody on earth has the faintest clue how all those 175 billion paramters are shaping the response to input.

It doesn't matter bout "thinking" or whatever. Any black box system will be uncontrollable in essence. You can not make inviolable rules for a system you don't understand.

And saying LLMs hallucinate because they don't understand anything is stupid. And just shows ignorance on your part. Models hallucinate because they're rewarded for plausibly guessing during training when knowledge fails. Plausibly guessing is a much better strategy to reducing loss.

And the conclusion is obvious enough. Bugger smarter models hallucinate less because they guess less. That holds true.

https://crfm.stanford.edu/helm/latest/?group=core_scenarios

All the instruct tuned models on this list follow that trend.

From Ada to Babbage to Curie to Claude to Davinci-002/003. Greater size equals Greater truthfulness (evaluated on TruthfulQA)

Post reply on HN