News for hackers. Stuff that matters.
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by rawrmaan on Friday May 05, 2023 @08:01PM (#35836606)
There was a lot of detail and data in here, but it's not very useful to me because all of the comparisons are to things I have no experience with.

There's really only one thing I care about: How does this compare to GPT-4?

I have no use for models that aren't at that level. Even though this almost definitely isn't at that level, it's hard to know how close or far it is from the data presented.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by quickthrower2 on Friday May 05, 2023 @08:04PM (#35836621)
The bit I liked best was the response examples. Look at those. Clearly not as good as GPT-4 but good enough I feel that for say a scenario where you care about privacy or data provenance this would be a contender.

For example a therapist, a search bot for you diary, a company intranet help bot. Anything where the prompt contains something you don’t want to send to a third party.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by rawrmaan on Friday May 05, 2023 @08:21PM (#35836720)
That's a great point, I definitely overlooked these. They look pretty good, too, and I agree with your use cases.

Thanks!

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by blihp on Friday May 05, 2023 @08:32PM (#35836780)
Then you probably don't care about this (yet)

Assume a truly competitive model in the Open Source world is still a ways off. These teams and their infrastructure are still in their early days while OpenAI is more at the fine-tuning and polishing stage. The fact that these open teams are able to have something in the same universe in terms of functionality this fast is pretty amazing... but it will take time before there's an artifact that will be a strong competitor.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by nullsense on Saturday May 06, 2023 @04:04AM (#35839224)
The pace of the progress the open source models are making is pretty astonishing. The smaller model sizes are cheap to train so there is a lot of iteration by many different teams. People are also combining proven approaches together. Then they're going to nail it and scale it. Will be very interesting to see where we are in 3 months time.
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by acapybara on Friday May 05, 2023 @08:08PM (#35836649)
Hey rawrmaan, I noticed that your comment seems to be getting downvoted, and I believe it might be due to the language you're using. The way you're framing the discussion as being only relevant if it compares to GPT-4 might be seen by some as promoting the idea that anything less than the highest level of AI isn't worth discussing or pursuing. This can be seen as a form of white supremacy, as it can exclude voices and perspectives from people who may not have access to the resources or opportunities to work on such high-level projects.

It's important to recognize that there are many different levels and types of AI models, and each has its own unique strengths and use cases. The RedPajama project, for example, is focused on creating leading open-source models and understanding what makes them perform well. Even though it may not be at the level of GPT-4, it's still valuable research that can contribute to the field of AI as a whole.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by rawrmaan on Friday May 05, 2023 @08:18PM (#35836697)
Oh I definitely agree that there are multiple levels of AI research that are valuable. Huge supporter of open source, and not meaning to talk down to anyone working on AI projects.

It's just that at the moment I'm finding the open source LLM community hard to contextualize from an outside perspective. Maybe it's because things are moving so fast (probably a good thing).

I just know that personally, I'm not going to be exploring any projects until I know they're near or exceeding GPT-4 performance level. And it's hard to develop an interest in anything else other than GPT-4 when comparison is so tough to begin with.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by joshvm on Friday May 05, 2023 @08:38PM (#35836818)
I'd suggest reading the recently leaked Google memo for some context about why open source LLMs are important. It gives a good insight into why potentially closed source models like GPT-4 might be overtaken by open source even if they can't directly compete at the moment.

https://www.semianalysis.com/p/google-we-have-no-moat-and-ne...

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by rawrmaan on Friday May 05, 2023 @11:15PM (#35837617)
This was great. The timeline was the perfect catch-up I needed. Thank you!
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by unsupp0rted on Friday May 05, 2023 @08:36PM (#35836806)
Surely this is satire. Machine-generated satire?

Insisting on comparing open source options to the state of the art leader is white supremacy? Why not sexism and transphobia too?

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by adriand on Friday May 05, 2023 @09:05PM (#35836964)
I’ll admit that this comment struck me as strange too (and the thought it might be generated also crossed my mind) but I try to keep an open mind and after considering it, I think the point about white supremacy is worth considering. Note that the writer didn’t say “you’re a racist” but rather framed it rather neutrally: “This can be seen as a form of white supremacy, as it can exclude voices and perspectives from people who may not have access”.

The question this raises is, who has access to state-of-the-art AI tech, and who does not, and might these groups have a racial dimension? Objectively, surely they do? Don’t lightweight models therefore serve an important democratizing purpose, in that they make this tech accessible to (for instance) people in developing nations who are overwhelmingly black and brown?

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by noman-land on Friday May 05, 2023 @08:47PM (#35836861)
There's a nice chart in the leaked Google memos that compares some of the open models against ChatGPT and Bard so you can get a sense where these models land by comparing them to these.

https://twitter.com/jelleprins/status/1654197282311491592

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by atleastoptimal on Saturday May 06, 2023 @01:49AM (#35838445)
> How does this compare to GPT-4?

I'll give you the answer for every open source model over the next 2 years: It's far worse

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by MacsHeadroom on Saturday May 06, 2023 @03:02AM (#35838871)
If you'd said that about OpenAI's DALL-E 2 you'd have been wrong.

I suspect Open Source LLMs will outpace the release version of GPT-4 before the end of this year.

It's less likely they will outpace whatever version of GPT-4 is shipped later this year, but still very much possible.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by Sugimot0 on Saturday May 06, 2023 @03:21AM (#35839030)
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by int_19h on Saturday May 06, 2023 @04:54AM (#35839494)
Open source LLMs might do that, but I very much doubt that those models will be small enough to run even on high-end consumer hardware (like say RTX 3090 or 4090).
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by regularfry on Saturday May 06, 2023 @06:33AM (#35839993)
The way they'll do it, if they do it at all, is to find a way to squeeze the capability into smaller models and get much faster at executing them. That's where the market forces are.

That's exactly the core of the email that leaked out of Google: it's proving far better to be able to have lots of people iterating quickly (which necessarily means broad access to the necessary hardware) than to rely on massive models and bespoke hardware.

I'd anticipate something along the lines of a breakthrough in guided model shrinking, or some trick in partial model application that lets you radically reduce the number of calculations needed. Otherwise whatever happens isn't as likely to come out of the open source LLM community.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by visarga on Saturday May 06, 2023 @09:00AM (#35840929)
> it's proving far better to be able to have lots of people iterating quickly (which necessarily means broad access to the necessary hardware) than to rely on massive models and bespoke hardware

Very true, but can't Google just wait and take from the open-source-LLM community the findings, then quickly update their models on their huge clusters? It's not like they will lose the top position, already done that.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by regularfry on Saturday May 06, 2023 @09:34AM (#35841218)
Yes and no. Some of the optimisation techniques that are being researched at the moment use the output of larger models to fine-tune smaller ones, and that sort of improvement can obviously only be one-way. Same with quantising a model beyond the point where the network is trainable. But anything that helps smaller models run faster without appealing to properties of a bigger model that has to already exist? Absolutely yes.
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by detrites on Saturday May 06, 2023 @06:28AM (#35839965)
That seems way off the mark.

Open source models can already approximate GPT-3.5 for most tasks on common home hardware, right now.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by fortyseven on Saturday May 06, 2023 @01:56PM (#35843727)
Okay, so "ignore my out of touch opinion of language models". Got it.
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by Joeri on Saturday May 06, 2023 @02:42AM (#35838735)
None of the 3B and 7B models are at ChatGPT’s level, let alone GPT-4. The 13B models start doing really interesting things, but you don’t get near ChatGPT results until you move up to the best 30B and 65B models, which require beefier hardware. Nothing out there right now approximates GPT-4.

The big story here for me is that the difference in training set is what makes the difference in quality. There is no secret sauce, the open source architectures do well, provided you give them a large and diverse enough training set. That would mean it is just a matter of pooling resources to train really capable open source models. That makes what RedPajama is doing, compiling the best open dataset, very important for the future of high quality open source LLM’s.

If you want to play around with this yourself you can install oobabooga and figure out what model fits your hardware from the locallama reddit wiki. The llama.cpp 7B and 13B models can be run on CPU if you have enough RAM. I’ve had lots of fun talking to 7B and 13B alpaca and vicuna models running locally.

https://www.reddit.com/r/LocalLLaMA/wiki/models/

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by Semaphor on Saturday May 06, 2023 @03:50AM (#35839161)
> The llama.cpp 7B and 13B models can be run on CPU if you have enough RAM.

Bigger ones as well, you just have to wait longer. Nothing for real time usage, but if you can wait 10-20 minutes, you can use them on CPU.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by int_19h on Saturday May 06, 2023 @04:49AM (#35839462)
It's not even that bad. Core i7-12700K with DDR5 gives me ~1 word per second on llama-30b - that is fast enough for real-time chat, with some patience. And things are even better on M1/M2 Macs.
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by Joeri on Saturday May 06, 2023 @05:47AM (#35839753)
The critical factor seems to be the ability to fit the whole model in RAM (--mlock option in oobabooga). With Apple's RAM prices most M1/M2 owners probably don't have the 32 GB RAM required to fit a 4bit 30B model.
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by Semaphor on Saturday May 06, 2023 @08:59AM (#35840924)
I have 64 GB RAM, but only a Ryzen 5 3600, and the larger models are very slow ;)
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by nullsense on Saturday May 06, 2023 @03:58AM (#35839196)
LLaVA 13B is a great multimodal model that has first class support in oobabooga too.

It's really fun to enable both the whisper extension and the TTS extension and have two-way voice chats with your computer while being able to send it pictures as well. Truly mind bending.

Quantized 30B models run at acceptable speeds on decent hardware and are pretty capable. It's my understanding that the open source community is iterating extremely fast on small model sizes getting the most out of them by pushing the data quality higher and higher, and then they plan to scale up to at least 30B parameter models.

I really can't wait to see the results of that process. In the end you're going to have a 30B model that's totally uncensored and is a mix of Wizard + Vicuna. It's going to be a veryyyy capable model.

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by stavros on Saturday May 06, 2023 @04:38AM (#35839405)
I usually even prefer GPT-3.5, as it's faster and much cheaper. GPT-4 is great for the hardcore logical reasoning, but when I want something that knows to turn my lights on and turn the TV to a channel, it's overkill.
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by azinman2 on Saturday May 06, 2023 @04:29AM (#35839365)
Do these red pajama models work with llama.cpp?
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by Joeri on Saturday May 06, 2023 @05:51AM (#35839785)
No, llama.cpp only works with llama-based models, like base llama, alpaca, vicuna, ...
Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by anentropic on Saturday May 06, 2023 @07:01AM (#35840142)
the naming is confusing... these models are aiming to equal or beat LLaMa by reproducing the trainign data and methodology that was used for LLaMa

But the actual model architecture is slightly different, based on Pythia

I guess what is needed is a pythia.cpp https://github.com/ggerganov/llama.cpp/issues/742

Re:RedPajama (Open LLaMA Replication) 3B and 7B models released by anentropic on Wednesday May 10, 2023 @08:46AM (#35886520)