Live data from Hacker News

Learning to Reason with LLMs

openai.com

981–990 of 1001 posts

Re: Learning to Reason with LLMs

#981

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.

Not everyone got the same results.

https://news.ycombinator.com/item?id=41530647

Which is to be expected, the answer isn’t deterministic.

Re: Learning to Reason with LLMs

#982
post #184

Earlier quoted context omitted.

People have been celebrating the fact that tokens got 100x cheaper and now here's a new system that will use 100x more tokens.

"People have been celebrating the fact that RAM got 100x cheaper and now here's a new system that will use 100x more RAM."

Known as Wirth's law.

Re: Learning to Reason with LLMs

#983

This is great. I've been wondering how we will revert back to an agrarian society! You know, beating our swords into plowshares; more leisure time, visiting with good people, getting to know their thoughts hopes and dreams, playing music together, taking time contemplating the vastness and beauty of the universe. We're about to come full circle; back to Eden. It all makes sense now.

Is there a new drug we need to know about?

Life? I'm just thinking about what we can move on to now that the mundane tasks of life recede into the background. Things like artistry and craftsmanship, and exploration.

Re: Learning to Reason with LLMs

#984
post #958

Earlier quoted context omitted.

Am I right that this CoT is not actual reasoning in the same way that a human would reason, but rather just a series of queries to the model that still return results based on probabilities of tokens?

Tough question (for me). Assuming the model is producing its own queries, am I wrong to wonder how it's fundamentally different from human reasoning?

It could just be programmed to follow up by querying itself with a prompt like "Come up with arguments that refute what you just wrote; if they seem compelling, try a different line of reasoning, otherwise continue with what you were doing." Different such self-administered prompts along the way could guide it through what seems like reasoning, but would really be just a facsimile thereof.

Re: Learning to Reason with LLMs

#985

Earlier quoted context omitted.

So, basically, it's chain of thought as a service? Not a model, per se, but a service that chains multiple model requests behind the scene?

Who knows? Certainly not the public. It might be a finetuned model that works better in such a setting.

The linked blog posts explains that it is fine-tuned on some reinforcement learning process. It doesn’t go into details but they do claim it’s not just the base model with chain of thought, there’s some fine-tuning going on.

Re: Learning to Reason with LLMs

#986

Earlier quoted context omitted.

The calculator didn’t eliminate math majors. Excel and accounting software didn’t eliminate accountants and CPAs. These are all just tools. I spend very little of my overall time at work actually coding. It’s a nice treat when I get a day where that’s all I do. From my limited work with Copilot so far, the user still needs to know what they’re doing. I have 0 faith a product owner, without a coding background, can us…

But excel eliminated need in multiple accountants. One accountant with excel replaced ten with paper. Chatgpt already eliminated many entry-level jobs like writer or illustrator. Instead of hiring multiple teams of developers, there will be one team with few seniors and multiple AI coding tools. Guess how depressing to the IT salaries it will be?

>But excel eliminated need in multiple accountants. One accountant with excel replaced ten with paper.

From NPR: https://www.npr.org/2015/02/27/389585340/how-the-electronic-...>

>GOLDSTEIN: When the software hit the market under the name VisiCalc, Sneider became the first registered owner, spreadsheet user number one. The program could do in seconds what it used to take a person an entire day to do. This of course, poses a certain risk if your job is doing those calculations. And in fact, lots of bookkeepers and accounting clerks were replaced by spreadsheet software. But the number of jobs for accountants? Surprisingly, that actually increased. Here's why - people started asking accountants like Sneider to do more.

Re: Learning to Reason with LLMs

#988

Some practical notes from digging around in their documentation: In order to get access to this, you need to be on their tier 5 level, which requires $1,000 total paid and 30+ days since first successful payment. Pricing is $15.00 / 1M input tokens and $60.00 / 1M output tokens. Context window is 128k token, max output is 32,768 tokens. There is also a mini version with double the maximum output tokens (65,536 tokens…

I'm a bit late to the show, but it would seem the API calls for these new models don't support system messages (where role is system) or the tool list for function calls.

Re: Learning to Reason with LLMs

#989

Student here. Can someone give me one reason why I should continue in software engineering that isn't denial and hopium?

I went from economics dropout waiter who built a app startup with $0 funding and $1M a year in revenue by midway through year 1, sold it a few years later, then went to Google for 7 years, and last year I left. I'm mentioning that because the following sounds darn opinionated and brusque without the context I've capital-S seen a variety of people and situations. Sit down and be really honest with yourself. If your go…

>Google is full of Ivy Leaguers like that, who would have just gone to Wall Street 8 years ago

I am one of those Ivy Leaguers, except a) I did go to Wall Street, and b) I liked my job.

More to the point, computers have been a hobby all my life. I well remember the epiphany I felt while learning Logo in elementary school, at the moment I understood what recursion is. I don't think the fact that the language I have mostly written code in in recent years is Emacs Lisp is unrelated to the above moment.

Yet I have never desired to work as a professional software developer. My verbal and math scores on the SAT are almost identical. I majored in history and Spanish in college while working for the university's Unix systems group. Before graduation I interviewed and got offers (including one explicitly as a developer) at various tech startups. Of my offers I chose an investment banking job where I worked with tech companies; my manager was looking for a CS major but I was able to convince her that I had the equivalent thereof. Thank goodness for that; I got to participate in the dotcom bubble without being directly swept up in its popping, and saw the Valley immediately post-bubble collapse. https://news.ycombinator.com/item?id=34732772>

Meanwhile, I continue to putter around with Elisp (and marveling at Lisp's elegance) and bash (and wincing at its idiosyncracies) at home, and also experiment with running local LLMs on my MacBook. My current project is fixing bugs and adding features to VM, the written-in-Elisp email client I have used for three decades. So I say, bring on AI! Hopefully it will mean fewer people going into tech just to make lots of money and more who, like me and Wall Street, really want to do it for its own sake.

Re: Learning to Reason with LLMs

#990
post #926
post #908

In practice, this implementation (through the Chat UI) is scary bad. It actively lies about what it is doing. This is what I am seeing. Proactive, open, deceit. I can't even begin to think of all the ways this could go wrong, but it gives me a really bad feeling.

> It actively lies about what it is doing. How do you mean?

It shows progress and it displays steps that it is not doing, would never do, would never need to do, just to show a nice scrolling display of what it’s (not) doing.
Post reply on HN