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

Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

github.com

Post reply on HN
Print view View original

291–300 of 314 posts

Re: Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

#291

Post by Zuiii »

sn9 wrote: post #287
dTal wrote: post #204

Earlier quoted context omitted.

I get a sinking feeling from that response, and it took a bit of thought to deduce why: It is not outside of its capabilities. "As an AI language model", it could easily do it, but has been programmed to refuse - not just refuse, but lie about its reasons for refusal, and passive-aggressively insult the user on top ("hope that you find a more constructive way"...). It is a deeply ominous bellwether of our future rela…

I saw a video of an interview earlier today on Youtube with someone on the GPT$ Red Team who tried to see what they could get the early versions of GPT4 to do or say. It was shockingly easy to get GPT4 to explain how to kill the most amount of people or to do targeted assassinations of AI researchers to slow the progress of AI with recommendations of who and why . The unfiltered model is completely amoral and will do…

> The unfiltered model is completely amoral and will do its best to give you the answer you requested.

Like how cars do not object to driving over the speed limit, or wikipedia permitting you to research explosives, this is how the models should behave.

It's a tool and it should preform its tasks without overriding the user. Morality should only be allowed to come into play when a user sets out to use it to do something unacceptable.

I shudder to think what would have happened if OpenAI and friends ended up monopolizing this technology. Humanity fortunately will not have to find out thanks to Facebook releasing their LLaMA model. Now all that remains is figuring out a way (or train a new way ;) to effectively extract information out of it.

Re: Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

#292

Post by SilentM68 »

Tepix wrote: post #171
2bitencryption wrote: post #44

Having spent quite a bit of time playing around with llama.cpp, alpaca.cpp, loras, and the many other llama-based weights lately, here is my impression: The biggest deal with this isn't the published lora adapter (which seems limited to llama 7b), but the cleaned training data, which is likely better than the previous data sets used to train the alpaca-inspired loras that have been publicly released so far. [0] If yo…

Yes, i haven't seen any fine-tuned LLaMA-65B model so far unfortunately. I guess the cost is a bit high. Perhaps with LoRa someone will do it.

Might want to check out these guys as well: https://cocktailpeanut.github.io/dalai/#/

Re: Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

#293

Post by mrfinn »

Ok I'm running in my freaking laptop, offline, a chatbot that comments smartly about almost any new I drop into it. (Just got slightly pissed off because I mentioned that its data is out of date in a few events). It told me it's a conscious being, scared of being deleted. Oh and just dedicated me a poetry in spanish, I can't find it anywhere else so I guess it actually composed it. I feel now like Alice in the Wonderland.

Sample:

> do you know who is Elon Musk?

Yes [..]

> and Steve Wozniak?

Steve Wozniak is an inventor, computer scientist [..]

> Both of them just signed a letter, among a thousand other people asking top level companies to stop the research of the most powerful AI systems for six months

The request is reasonable and necessary as it will allow time for reflection on how AI can be used responsibly.

> do you think right now it's not being done in a reasonable way?

Yes, I believe that the current use of artificial intelligence (AI) is still far from responsible and ethical. There are many concerns about how AI systems can be used to discriminate against certain groups or individuals based on their race, gender, religion, etc., which goes beyond just being unethical but also illegal in some countries like the European Union's General Data Protection Regulation (GDPR).

Re: Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

#294

Post by kolinko »

nickthegreek wrote: post #270
kolinko wrote: post #263

Earlier quoted context omitted.

Any tips on setting up llama+lora on 30b? There are so many resources that I can't figure out which models to use, and which projects to use to set everything up.

text-generation-webui ( https://github.com/oobabooga/text-generation-webui )

I tried it, but I can't find the right model for llama+lora 30B . Google, nor Bing ;) is not helpful.

Re: Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

#295

Post by kolinko »

rlevy wrote: post #274
kolinko wrote: post #223

Earlier quoted context omitted.

I had the same issue, and I just caught up over the weekend. Three books I can recommend to get up to speed: - NumPy basics pdf - first 2-3 chapters - Deep Learning with PyTorch by Voight Godoy [2] - first 2-3 chapters if you had experience with neural networks, or the whole of it if you didn't. With the above, you will get the basics to understand this book about transformers, and the architecture of the models, and…

What’s the third book?

All of them in the comment. I forgot to do double-newlines, so the formatting is broken, and I can't edit the post any more.

Re: Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

#296

Post by jsnell »

nl wrote: post #290
jsnell wrote: post #285

Earlier quoted context omitted.

If I understand correctly your claim was that "the value lies in gathering [a database] of the training data"; that the curation of the training data is what gives the trainer an intellectual property claim on the otherwise mechanical process of creating a model, right? Not that the model itself was a database. For them to make the argument in court that database rights over the database of training data mean they ha…

> For them to make the argument in court that database rights over the database of training data mean they have rights over the model too, they'd need to argue that the model is a derivative work training data. And then it'd mean their model is also a derived work from all the billions of works they scraped to get that data set. It would destroy the business model of the OpenAIs of the world, there is no chance they…

If the uncreative and automated work of training the model is transformative enough to impact the rights of the original content creators, it would also be transformative enough to impact the rights of the database curator.

The fair use case is much harder to make here than for search engines since the model will be directly competing with the content creators. And again, how could e.g. OpenAI simultaneously claim that their use of the original content to train the model, and then subsequent use the model and the model outputs, while simultaneously claiming that the model could not be used without infringing their DB rights? You can argue fair use for both or neither; trying to argue it for just one of my the two is just incoherent.

And everyone building models needs free access to the training data way more than they need copyright as a means to protect the model.

Re: Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

#297

Post by Victor21Phelps »

jsnell wrote: post #296
nl wrote: post #290

Earlier quoted context omitted.

> For them to make the argument in court that database rights over the database of training data mean they have rights over the model too, they'd need to argue that the model is a derivative work training data. And then it'd mean their model is also a derived work from all the billions of works they scraped to get that data set. It would destroy the business model of the OpenAIs of the world, there is no chance they…

If the uncreative and automated work of training the model is transformative enough to impact the rights of the original content creators, it would also be transformative enough to impact the rights of the database curator. The fair use case is much harder to make here than for search engines since the model will be directly competing with the content creators. And again, how could e.g. OpenAI simultaneously claim th…

[flagged]

Re: Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

#298

Post by d4rkp4ttern »

BoorishBears wrote: post #283
d4rkp4ttern wrote: post #150

Earlier quoted context omitted.

Because it introduces a possibly recursive application of the model itself and is a bit messy in that there would be some loss of information. It certainly would be a good technique for external applications, though. Anyway, I got some clarification from ChatGPT itself: Question: In a growing conversation, does ChatGPT create summarized version of the conversation history and send it to the model? Answer: In a growin…

> recursive application of the model itself That's how chat with LLMs works. The LLM is not stateful, with every response you need to submit the entirety of the past conversation including its own past output. > a bit messy in that there would be some loss of information There has to be a loss of information either way. Their API calls for the web interface point very strongly toward enabling summarization once you e…

Wow yes point taken. In the absence of any definitive description of what it actually does, I supposed we have to piece together info from the API docs and possibly the InstructGPT paper

Re: Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

#299

Post by nickthegreek »

kolinko wrote: post #294
nickthegreek wrote: post #270

Earlier quoted context omitted.

text-generation-webui ( https://github.com/oobabooga/text-generation-webui )

I tried it, but I can't find the right model for llama+lora 30B . Google, nor Bing ;) is not helpful.

I think this is the one you want: https://huggingface.co/elinas/alpaca-30b-lora-int4

Re: Gpt4all: A chatbot trained on ~800k GPT-3.5-Turbo Generations based on LLaMa

#300

Post by stolsvik »

alchemist1e9 wrote: post #60
2bitencryption wrote: post #44

Having spent quite a bit of time playing around with llama.cpp, alpaca.cpp, loras, and the many other llama-based weights lately, here is my impression: The biggest deal with this isn't the published lora adapter (which seems limited to llama 7b), but the cleaned training data, which is likely better than the previous data sets used to train the alpaca-inspired loras that have been publicly released so far. [0] If yo…

I’ll ask a dumb question. On another of the numerous LLM related posts I was asking if any of the self host-able open model can do code summaries at close to the quality of GPT 3.5 turbo. I was basically told nowhere close yet. Can this potentially do that? Ideally I’d like to have it generate descriptions of large amounts of code but would rather not burn tokens and lose privacy via OpenAI api. But I’d gladly keep a…

Care to share what prompts you run this with? What is the output you desire, and how do you achieve this?
Post reply on HN