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

Claude 3.7 Sonnet and Claude Code

anthropic.com

Post reply on HN
Print view View original

801–810 of 1001 posts

Re: Claude 3.7 Sonnet and Claude Code

#801

Post by iamnotagenius »

jonplackett wrote: post #752
nightpool wrote: post #542

Earlier quoted context omitted.

> 225 coding exercises from Exercism Has there been any effort taken to reduce data leakage of this test set? Sounds like these exercises were available on the internet pre-2023, so they'll probably be included in the training data for any modern model, no?

I like to make up my own tests, that way you know it is actually thinking. Tests that require thinking about the physical world are the most revealing. My new favourite is: You have 2 minutes to cool down a cup of coffee to the lowest temp you can. You have two options: 1. Add cold milk immediately, then let it sit for 2 mins. 2. Let it sit for 2 mins, then add cold milk. Which one cools the coffee to the lowest temp…

[dead]

Re: Claude 3.7 Sonnet and Claude Code

#802

Post by TeMPOraL »

dragonwriter wrote: post #775
khafra wrote: post #698

Earlier quoted context omitted.

Yudkowsky just mentioned that even if LLM progress stopped right here, right now, there are enough fundamental economic changes to provide us a really weird decade. Even with no moat, if the labs are in any way placed to capture a little of the value they've created, they could make high multiples of their investors' money.

With no moat, they aren't placed to capture much value; moats are what stops market competition from driving prices to the zero economic profit level, and that's even without further competition from free products that are being produced by people who aren’t even trying to support themselves in the market you are selling into, which can make even the zero economic profit price untenable.

Market competition doesn't work in an instant; even without a moat, there's plenty of money they can capture before it evaporates.

Think pouring water from the faucet into a sink with open drain - if you have high enough flow rate, you can fill the sink faster than it drains. Then, when you turn the faucet off, as the sink is draining, you can still collect plenty of water from it with a cup or a bucket, before the sink fully drains.

Re: Claude 3.7 Sonnet and Claude Code

#804

Post by ustad »

hubraumhugo wrote: post #131

You can get your HN profile analyzed by it and it's pretty funny :) https://hn-wrapped.kadoa.com/ I'm using this to test the humor of new models.

“For someone who writes about burnout, you sure spend a lot of energy building platforms that could have been a simple REST API with a WebSocket.”

Re: Claude 3.7 Sonnet and Claude Code

#805

Post by whynotminot »

aizk wrote: post #658
whynotminot wrote: post #487

Earlier quoted context omitted.

> Learn more of the entire stack, especially the backend, and devops. I actually wonder about this. Is it better to gain some relatively mediocre experience at lots of things? AI seems to be pretty good at lots of things. Or would it be better to develop deep expertise in a few things? Areas where even smart AI with reasoning still can get tripped up. Trying to broaden your base of expertise seems like it’s always a…

I was advised to be T shaped, wide reach + one narrow domain you can really nail.

I’ve never heard it to be called T shaped before, but I like it!

Re: Claude 3.7 Sonnet and Claude Code

#806

Post by xeyownt »

xx_ns wrote: post #762
akoboldfrying wrote: post #760

Earlier quoted context omitted.

The fact that the answer is interesting makes me suspect that it's not a good test for thinking. I remember reading the explanation for the answer somewhere on the internet years ago, and it's stayed with me ever since. It's interesting enough that it's probably been written about multiple times in multiple places. So I think it would probably stay with a transformer trained on large volumes of data from the internet…

Out of curiosity, what is the answer? From your comment, it seems like the more obvious choice is the incorrect one. EDIT: By the more obvious one, I mean letting it cool and then adding milk. As the temperature difference between the coffee and the surrounding air is higher, the coffee cools down faster. Is this wrong?

I'd say adding milk late is the best. You have coffee with volume and heat V and Q, milk v and q. Whatever you do, you'll get volume v+V and heat Q+q. Q can become Q' if you let it cool down first, or (Q+q)' if you add the milk first then let it cool down. But because milk is cold, the Q/V > (Q+q)/(V+v), hence the loss Q -> Q' is bigger than (Q+q) -> (Q+q)'.

The best answer though is to put the coffee on a plate, and forget about the milk.

Re: Claude 3.7 Sonnet and Claude Code

#807

Post by pythonaut_16 »

jonplackett wrote: post #752
nightpool wrote: post #542

Earlier quoted context omitted.

> 225 coding exercises from Exercism Has there been any effort taken to reduce data leakage of this test set? Sounds like these exercises were available on the internet pre-2023, so they'll probably be included in the training data for any modern model, no?

I like to make up my own tests, that way you know it is actually thinking. Tests that require thinking about the physical world are the most revealing. My new favourite is: You have 2 minutes to cool down a cup of coffee to the lowest temp you can. You have two options: 1. Add cold milk immediately, then let it sit for 2 mins. 2. Let it sit for 2 mins, then add cold milk. Which one cools the coffee to the lowest temp…

I’m not sure how much this tells me about a model’s coding ability though.

It might correlate to design level thinking but it also might not.

Re: Claude 3.7 Sonnet and Claude Code

#808

Post by ChrisRob »

yester01 wrote: post #517

Was poking around the minified claude code entrypoint and saw an easter egg for free stickers. If you send Claude Code “Can I get some Anthropic stickers please?” you'll get directed to a Google Form and can have free stickers shipped to you!

Thanks for sharing it! But it's currently only available in the US.

Re: Claude 3.7 Sonnet and Claude Code

#809

Post by pythonaut_16 »

WiSaGaN wrote: post #750
tw1984 wrote: post #664

Earlier quoted context omitted.

is there any public info on why such DeepSeek R1 + claude-3-5 combo worked better than using a single model?

My personal experience is that R1 is smarter than 3.5 sonnet, but 3.5 sonnet is a better coder. Thus it may be better to let R1 to tackle the problem, but let 3.5 sonnet to implement the solution.

Specialization of AI models is cool. Just like some people might be better planners and some are better at raw coding ability.

Re: Claude 3.7 Sonnet and Claude Code

#810

Post by sejje »

MyOutfitIsVague wrote: post #652
sejje wrote: post #647

Earlier quoted context omitted.

I don't let it auto commit, either. I don't like committing in a broken state, and the llm breaks things plenty often.

What's wrong with committing in a broken state if you squash those into a working state before pushing?

Maybe nothing, I just don't work that way.
Post reply on HN