oh wow, something you can roughly model as a diy in a base model. so impressive. yawn. at least NVDA should benefit. i guess.
If there's a way to do something like this with Llama I'd love to hear about it (not being sarcastic)
Learning to Reason with LLMs
31–40 of 1001 posts
Re: Learning to Reason with LLMs
#32Someone give this model an IQ test stat.
Re: Learning to Reason with LLMs
#33> We have found that the performance of o1 consistently improves with more reinforcement learning (train-time compute) and with more time spent thinking (test-time compute).
Wow. So we can expect scaling to continue after all. Hyperscalers feeling pretty good about their big bets right now. Jensen is smiling.
This is the most important thing. Performance today matters less than the scaling laws. I think everyone has been waiting for the next release just trying to figure out what the future will look like. This is good evidence that we are on the path to AGI.
Re: Learning to Reason with LLMs
#34Congrats to OpenAI for yet another product that has nothing to do with the word "open"
And Apple's product line this year? Phones. Nothing to do with fruit. Almost 50 years of lying to people. Names should mean something!
Re: Learning to Reason with LLMs
#35I LOVE the long list of contributions. It looks like the credits from a Christoper Nolan film. So many people involved. Nice care to create a nice looking credits page. A practice worth copying.
Re: Learning to Reason with LLMs
#36What's with this how many r's in a strawberry thing I keep seeing?
What's amazing is that given how LLMs receive input data (as tokenized streams, as other commenters have pointed out) it's remarkable that it can ever answer this question correctly.
Re: Learning to Reason with LLMs
#37> we are releasing an early version of this model, OpenAI o1-preview, for immediate use in ChatGPT Awesome!
I am interpreting "immediate use in ChatGPT" the same way advanced voice mode was promised "in the next few weeks."
Probably 1% of users will get access to it, with a 20/message a day rate limit. Until early next year.
Re: Learning to Reason with LLMs
#38oh wow, something you can roughly model as a diy in a base model. so impressive. yawn. at least NVDA should benefit. i guess.
If there's a way to do something like this with Llama I'd love to hear about it (not being sarcastic)
nurture the model
have patience
and a couple bash scripts
Re: Learning to Reason with LLMs
#39A lot of skepticism here, but these are astonishing results! People should realize we’re reaching the point where LLMs are surpassing humans in any task limited in scope enough to be a “benchmark”. And as anyone who’s spent time using Claude 3.5 Sonnet / GPT-4o can attest, these things really are useful and smart! (And, if these results hold up, O1 is much, much smarter.) This is a nerve-wracking time to be a knowledge worker for sure.
Re: Learning to Reason with LLMs
#40That naming scheme...
Will the next model be named "1k", so that the subsequent models will be named "4o1k", and we can all go into retirement?