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

Learning to Reason with LLMs

openai.com

Post reply on HN
Print view View original

401–410 of 1001 posts

Re: Learning to Reason with LLMs

#401

Post by MattHeard »

losvedir wrote: post #383
bartman wrote: post #367

This 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?

It appeared for me about thirty minutes after I first checked.

Re: Learning to Reason with LLMs

#402

Post by tedsanders »

wewtyflakes wrote: post #176

Maybe I missed it, but do the tokens used for internal chain of thought count against the output tokens of the response (priced at spicy level of $60.00 / 1M output tokens)?

Yes. Chain of thought tokens are billed, so requests to this model can be ~10x the price of gpt-4o, or even more.

Re: Learning to Reason with LLMs

#403

Post by m3kw9 »

losvedir wrote: post #383
bartman wrote: post #367

This 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?

Likely phased rollout throughout the day today to prevent spikes

Re: Learning to Reason with LLMs

#404

Post by iammjm »

staplers wrote: post #386
ComputerGuru wrote: post #347

The "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…

"This knowledge is perfectly fine to disseminate via traditional means, but God forbid an LLM share it!" Barrier to entry is much lower.

How is typing a query in a chat window “much lower” vs typing the query in Google?

Re: Learning to Reason with LLMs

#405

Post by zamadatix »

gibsonf1 wrote: post #366

Yes, but it will hallucinate like all other LLM tech making it fully unreliable for anything mission critical. You literally need to know the answer to validate the output, because if you don't, you won't know if output is true or false or in between.

You need to know how to validate the answer to your level of confidence, not necessarily already have the answer to compare itself. In some cases this is the same task or (close enough to) that it's not a useful difference, in other cases the two aren't even from the same planet.

Re: Learning to Reason with LLMs

#406

Post by peppertree »

carabiner wrote: post #359
tylervigen wrote: post #293

Here's the o1-preview answer to the strawberry question: -- There are *three* letter "R"s in the word "strawberry." Let's break down the word to count the occurrences: - *S* - *T* - *R* - *A* - *W* - *B* - *E* - *R* - *R* - *Y* The letter "R" appears in positions 3, 8, and 9.

can you ask it: 9.11 and 9.9, which number is larger

4.9 is larger than 4.11.

Explanation:

• 4.9 is equivalent to 4.90. • 4.11 remains 4.11.

When comparing the numbers:

• 4.90 (which is 4.9) • 4.11

Since 4.90 > 4.11, 4.9 is the larger number.

Re: Learning to Reason with LLMs

#407

Post by dboreham »

ComputerGuru wrote: post #347

The "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…

I think it's about perception of provenance. The information came from some set of public training data. Its output however ends up looking like it was authored by the LLM owner. So now you need to mitigate the risk you're held responsible for that output. Basic cake possession and consumption problem.

Re: Learning to Reason with LLMs

#409

Post by acomjean »

I always think to a professor that was consulting on some civil engineering software. He found a bug in the calculation it was using to space rebar placed in concrete, based on looking at it was spitting out and thinking that looks wrong.

This kind of thing makes me nervous.

Re: Learning to Reason with LLMs

#410

Post by octernion »

carabiner wrote: post #359
tylervigen wrote: post #293

Here's the o1-preview answer to the strawberry question: -- There are *three* letter "R"s in the word "strawberry." Let's break down the word to count the occurrences: - *S* - *T* - *R* - *A* - *W* - *B* - *E* - *R* - *R* - *Y* The letter "R" appears in positions 3, 8, and 9.

can you ask it: 9.11 and 9.9, which number is larger

it claims 9.11 is larger, hah.

https://postimg.cc/KKKLZNjW

Post reply on HN