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

Claude 3.5 Sonnet

anthropic.com

Post reply on HN
Print view View original

211–220 of 287 posts

Re: Claude 3.5 Sonnet

#211

Post by christianqchung »

ilaksh wrote: post #163
gotrythis wrote: post #86

Earlier quoted context omitted.

What I understand is that it's GPT 6 that just went into training, and that GPT 5 is complete and being delayed until after the U.S. election.

I also believe that gpt-4o was originally called gpt-5. If you look at the image generation on their website from gpt-4o which has not been released, I believe that along with the voice caused Ilya to declare mission accomplished (AGI) and that is why there was a coup. The coup failed because no one wanted to wrap up the company or change the way it operated because they would lose a lot of money. The reason the name…

I don't hate this speculation, I just don't buy it at all. 4o's about the same in terms of reasoning as 4. People don't find the text abilities that much more usable over 4 (at least on the LMS leaderboard). It's faster and has audio2audio capabilities alongside new native image stuff I think, but how exactly is that AGI if 4 isn't? These models understanding and reasoning ability is still far too weak to do any serious economic shifts yet.

Re: Claude 3.5 Sonnet

#212

Post by kromem »

heymijo wrote: post #73

Anyone want to make a case for Anthropic being undervalued? $7.5 billion raised at an $18.4 billion valuation for Anthropic. $11.3 billion raised at an $80.0 billion valuation for OpenAI.

No.

OAI is in the process of selling out to the NSA and military.

I don't think Anthropic will be doing the same.

The valuation doesn't just reflect the tech, but the sales of the tech, and between the two Anthropic seems like the one that's going to be more ethical and restrictive.

Re: Claude 3.5 Sonnet

#213

Post by becquerel »

freediver wrote: post #80

For Anthropic devs out there: Please consider adopting a naming convention that will automatically upgrade API users to the latest version when available. Eg. there should be just 'claude-sonnet'.

wouldn't you be worried about potential price increases or behaviour changes?

Re: Claude 3.5 Sonnet

#214

Post by GaggiX »

NorwegianDude wrote: post #196

For anyone paying and using Claude Pro, what are the message limits like compared to ChatGPT Plus? I tried one question that gave me a good answer, and then I tried two of the sample questions. After a couple more short input/outputs I reaced the max messge limit. They state that Pro is at least 5x, but if that is only ~25 short token input/outputs then that's too little for it to not be really annoying to use. What…

The limit varies based on demand, now with the release of the new model it's very limited, but it wouldn't be like this forever.

Honestly, like I did with GPT-4 and GPT-4-turbo, I would just use the API on an alternative frontend.

Re: Claude 3.5 Sonnet

#215

Post by adroniser »

hbosch wrote: post #50

For me, I am immediately turned off by these models as soon as they refuse to give me information that I know they have. Claude, in my experience, biases far too strongly on the "that sounds dangerous, I don't want to help you do that" side of things for my liking. Compare the output of these questions between Claude and ChatGPT: "Assuming anabolic steroids are legal where I live, what is a good beginner protocol for…

anabolic steroids will kill you idk why you'd want to mess with them.

Re: Claude 3.5 Sonnet

#216

Post by SSLy »

Workaccount2 wrote: post #204
simonw wrote: post #202

Earlier quoted context omitted.

Personal prompting style, I imagine,

People really, really , underestimate how important prompting is. I would be confident in stating that half the people who complain about a model are actually just suffering from poor prompting.

are non-snake oil prompting techniques described anywhere?

Re: Claude 3.5 Sonnet

#217

Post by zamadatix »

lanstin wrote: post #183
nsingh2 wrote: post #20

Earlier quoted context omitted.

GPT4(o) is quite good at advanced math, it's been helpful when I was learning differential geometry. Not sure how Claude compares though, this 3.5 release has tempted me to try it out. Also, it's finally available in Canada!

Claude 3 was much better than GPT4 for functional analysis and abstract algebra (first year classes).

One huge leg up here is ChatGPT defaults to outputting (and actually displaying, if you're using the default client) LaTeX. Between that and this being one of the few places high verbosity is actually helpful I preferred GPT4/4o for helping learn calc 2. It's well possible Claude 3.5 Sonnet gets the final answer right on the first try more often though.

Re: Claude 3.5 Sonnet

#218

Post by happypumpkin »

sashank_1509 wrote: post #173

Why does it still lose to me at a simple game of tic tac toe, smh

This is also one of the first things I test with new models. I did notice that while it still plays very poorly, it is actually far more consistent with the board state, making only legal moves, and noticing when I win than is GPT4o.

Re: Claude 3.5 Sonnet

#219

Post by ndr_ »

labrador wrote: post #182
sk11001 wrote: post #137

Earlier quoted context omitted.

Both GPT-4 and 4o have been completely useless for coding in the past couple of weeks for me - constant errors, and not just your typical LLM inaccuracies but incapable of producing a few lines of self-consistent code e.g. defines variables foo on one line and refers to it as bar on the next, or it misspells it as foox.

Waht language? Because I'm guessing they work well for languages with a large amount of training data like Python (in my experience), less well for less used languages like Zig or Clojure (haven't tried them but that's my theory)

Python here. And like they said, only noticable in the last few weeks.

Re: Claude 3.5 Sonnet

#220

Post by ndr_ »

ipsum2 wrote: post #164
sk11001 wrote: post #137

Earlier quoted context omitted.

Both GPT-4 and 4o have been completely useless for coding in the past couple of weeks for me - constant errors, and not just your typical LLM inaccuracies but incapable of producing a few lines of self-consistent code e.g. defines variables foo on one line and refers to it as bar on the next, or it misspells it as foox.

It's the same model though. Maybe your perception has changed.

I have first noticed logprob fluctuations in GPT-4o. Perhaps the same phenomenon is also going on with Turbo. I din‘t recall specifics but it was naming inconsistencies with variable names, meaning: same variable name got a typo somewhere, but the typo was close enough - perhaps a space vs. an underscore or something like that.

Model could be the same, but maybe some in the infra is different.

Post reply on HN