How could it fail to solve some maths problems if it has a method for reasoning through things?
I have a method for reasoning through things but I'm pretty sure I'd fail some of those tough math problems too.
Learning to Reason with LLMs
391–400 of 1001 posts
Re: Learning to Reason with LLMs
#392Are we ready yet to admit Turing test has been passed?
Extremely basic agency would be required to pass the Turing test as intended. Like, the ability to ask a new unrelated question without being prompted. Of course you can fake this, but then you're not testing the LLM as an AI, you're testing a dumb system you rigged up to create the appearance of an AI.
> Turing proposed that a human evaluator would judge natural language conversations between a human and a machine designed to generate human-like responses. The evaluator would be aware that one of the two partners in conversation was a machine, and all participants would be separated from one another. The conversation would be limited to a text-only channel, such as a computer keyboard and screen, so the result would not depend on the machine's ability to render words as speech.
I don't see agency mentioned or implied anywhere: https://en.wikipedia.org/wiki/Turing_test
What definition or setup are you taking it from?
Re: Learning to Reason with LLMs
#393Feels like a lot of commenters here miss the difference between just doing chain-of-thought prompting, and what is happening here, which is learning a good chain of thought strategy using reinforcement learning. "Through reinforcement learning, o1 learns to hone its chain of thought and refine the strategies it uses." When looking at the chain of thought (COT) in the examples, you can see that the model employs diffe…
I'd be curious how this compared against "regular" CoT experiments. E.g. were the gpt4o results done with zero shot or was it asked to explain it's solution step by step.
Re: Learning to Reason with LLMs
#394This video[1] seems to give some insight into what the process actually is, which I believe is also indicated by the output token cost. Whereas GPT-4o spits out the first answer that comes to mind, o1 appears to follow a process closer to coming up with an answer, checking whether it meets the requirements and then revising it. The process of saying to an LLM "are you sure that's right? it looks wrong" and it coming…
So it's like the coding agent of gpt4. But instead of actually running the script and fix if it gets error, this one check with something similar to "are you sure". Thank for the link.
Re: Learning to Reason with LLMs
#395The generated chain of thought for their example is incredibly long! The style is kind of similar to how a human might reason, but it's also redundant and messy at various points. I hope future models will be able to optimize this further, otherwise it'll lead to exponential increases in cost.
I know my thoughts are never redundant or messy, that's for sure.
Fair enough, but you’re a human - not an AI which costs massive GPU hours.
Re: Learning to Reason with LLMs
#396Earlier quoted context omitted.
I appreciated that, too! FWIW, I could get Claude 3.5 to tell me how many rs a python program would tell you there are in strawberry. It didn't like it, though.
I was able to get GPT-4o to calculate characters properly using following prompt: """ how many R's are in strawberry? use the following method to calculate - for example Os in Brocolli. B - 0 R - 0 O - 1 C - 1 O - 2 L - 2 L - 2 I - 2 Where you keep track after each time you find one character by character """ And also later I asked it to only provide a number if the count increased. This also worked well with longer…
At that point just ask it "Use python to count the number of O's in Broccoli". At least then it's still the one figuring out the "smarts" needed to solve the problem instead of being pure execution.
Re: Learning to Reason with LLMs
#397Reading through the Chain of Thought for the provided Cipher example (go to the example, click "Show Chain of Thought") is kind of crazy...it literally spells out every thinking step that someone would go through mentally in their head to figure out the cipher (even useless ones like "Hmm"!). It really seems like slowing down and writing down the logic it's using and reasoning over that makes it better at logic, simi…
Seriously. I actually feel as impressed by the chain of thought, as I was when ChatGPT first came out. This isn't "just" autocompletion anymore, this is actual step-by-step reasoning full of ideas and dead ends and refinement, just like humans do when solving problems. Even if it is still ultimately being powered by "autocompletion". But then it makes me wonder about human reasoning, and what if it's similar? Just fo…
When an AI makes a silly math mistake we say it is bad at math and laugh at how dumb it is. Some people extrapolate this to "they'll never get any better and will always be a dumb toy that gets things wrong". When I forget to carry a 1 when doing a math problem we call it "human error" even if I make that mistake an embarrassing number of times throughout my lifetime.
Do I think LLM's are alive/close to ASI? No. Will they get there? If it's even at all possible - almost certainly one day. Do I think people severely underestimate AI's ability to solve problems while significantly overestimating their own? Absolutely 10,000%.
If there is one thing I've learned from watching the AI discussion over the past 10-20 years its that people have overinflated egos and a crazy amount of hubris.
"Today is the worst that it will ever be." applies to an awful large number of things that people work on creating and improving.
Re: Learning to Reason with LLMs
#398https://openai.com/index/introducing-openai-o1-preview/ > ChatGPT Plus and Team users will be able to access o1 models in ChatGPT starting today. Both o1-preview and o1-mini can be selected manually in the model picker, and at launch, weekly rate limits will be 30 messages for o1-preview and 50 for o1-mini. We are working to increase those rates and enable ChatGPT to automatically choose the right model for a given p…
The human brain uses 20 watts, so yeah we figured out a way to run better than human brain computation by using many orders of magnitude more power. At some point we'll need to reject exponential power usage for more computation. This is one of those interesting civilizational level problems. There's still a lack of recognition that we aren't going to be able to compute all we want to, like we did in the pre-LLM days…
For 20 watts of work on stuff like this for about 4 hours a day counting vacations and weekends and attention span. So 20 hours of rest, relaxation, distraction, household errands and stuff, so that maybe bumps it up to 120 watts per work hour. Then 22.5 years of training or so per worker, 45 year work period, 22.5 year retirement. So double it there to 240 watts. We can't run brains without bodies, so multiply that by 6 giving 1440 watts + the air conditioning, commuting to school and work, etc., maybe 2000 watts?
We're getting close to parity if things keep getting more efficient as fast as they have been. But that's without accounting for the AI training, which can on the plus side be shared among multiple agents, but on the down side can't really do continuous learning very well without catastrophic forgetting.
Re: Learning to Reason with LLMs
#399The "safety" example in the "chain-of-thought" widget/preview in the middle of the article is absolutely ridiculous. Take a step back and look at what OpenAI is saying here "an LLM giving detailed instructions on the synthesis of strychnine is unacceptable, here is what was previously generated vs our preferred, neutered content " What's this obsession with "safety" when it comes to LLMs? "This knowledge is perfectly…
ML companies must pre-anticipate legislative and cultural responses prior to them happening. ML will absolutely be used to empower criminal activity just as it is used to empower legit activity, and social media figures and traditional journalists will absolutely attempt to frame it in some exciting way.
Just like Telegram is being framed as responsible for terrorism and child abuse.
Re: Learning to Reason with LLMs
#400This is incredible. In April I used the standard GPT-4 model via ChatGPT to help me reverse engineer the binary bluetooth protocol used by my kitchen fan to integrate it into Home Assistant. It was helpful in a rubber duck way, but could not determine the pattern used to transmit the remaining runtime of the fan in a certain mode. Initial prompt here [0] I pasted the same prompt into o1-preview and o1-mini and both c…
Wow, that is impressive! How were you able to use o1-preview? I pay for ChatGPT, but on chatgpt.com in the model selector I only see 4o, 4o-mini, and 4. Is o1 in that list for you, or is it somewhere else?
Yes, o1-preview is on the list, as is o1-mini for me (Tier 5, early 2021 API user), under "reasoning".