I don’t know about your specific issue, but I have found that it helps quite a bit to often start new conversations. Also, I have a couple of paragraphs explaining the whole idea of my project that I always paste in at the beginning of each conversation. I’ve not been doing anything terribly complicated or cutting-edge, but I haven’t come across anything yet that Sonnet hasn’t been able to figure out, although sometimes it does take me being very clear and wordy about what I’m doing and starting from a fresh slate. I’ve also found it helps a lot if I specifically tell it to debug with lots of logs. Then I just go back and forth, giving it the outputs and changing code for it.

I was mainly doing python with gpt4, but now im working on an android project, so kotlin. Gpt4 wasn’t much use for kotlin, especially for questions involving more than a couple files. Sonnet is crushing it though, even when I give it 2k+ LoC. I’d say I’ve done about 2 months of pre-llm work in the last week, granted I am no professional, just a hobbyist.

For programming it is Sonnet 3.5, there is no remotely close 2nd place that I have tried or heard of, and I am always looking. I personally don’t really have any interest in measuring them in other ways. But for coding, Sonnet 3.5 is in a distant lead. Abacus.ai is a nice way to try various models for cheap. Really, some sort of agent setup like mixture of agents that uses Claude and got and maybe some others may do better than Claude alone. Matthew Berman shows Mixture of Agents with local models beating gpt4o, so doing it with sonnet3.5 and others of the best closed models would probably be pretty great.

My dude, no, I’m not the creator, settle down. Mixture of agents is free and open to anyone to use. Here is a demo of it by Matthew Berman. It isnt hard to set up.

https://youtu.be/aoikSxHXBYw

Believe it or not, openai is no longer making the best models. Claude Sonnet 3.5 is much better than openai’s best models by a considerable amount.

It takes a lot of energy to train the models in the first place, but very little once you have them. I run mixture of agents on my laptop, and it outperforms anything openai has released on pretty much every benchmark, maybe even every benchmark. I run it quite a bit and have noticed no change in my electricity bill. I imagine inference on gpt4 must almost be very efficient, if not, they should just switch to piping people open sourced llms run through MoA.

Yeah. It’s really interesting because juniors and hobbyist are the ones getting used to how to interact with it. Since it is rapidly improving, it won’t be long until it will outpace the grunt work ability of seniors and the new seniors will be the ones willing and able to use it. Programming is switching away from being able to write tedious code and into being able to come up with ideas and convey them clearly to an llm. There’s going to be a real leveling of the playing field when even the best seniors won’t have any use for most of their grunt work coding skills. The jump up from Opus 3 to Sonnet 3.5 is absolutely insane, and Opus 3.5 should be here before too long.

That’s really interesting. For android studio it’s been absolutely crushing it for me. It’s taken some getting used to, but I’ve had it build an app with about 60 files. I’m no master programmer, but I’ve been a hobbyist for a couple decades. What it’s done in the last 5 days for me would have taken me 2 months easy, and there’s lots of extra touches that I probably wouldn’t have taken time to do if it wasn’t as simple as loading in a few files and telling it what I want.

Usually when I work on something like this, my todo list grows much faster than my ability to actually put it together, but with this project I’m quickly running out of even any features that I can imagine. I’ve not had any of the issues of it running in circles like I would often get it gpt4.

Have you coded with Claude Sonnet 3.5 yet? It is mind-blowingly better than Opus 3, which was already noticeably better than anything openAI has put out yet. Gpt 4 was nice to code with, but this is on a whole other level. I can’t imagine what Opus 3.5 will be able to do.

Good answer, no way AI will possibly ever catch up to such brilliant responses as this. Certainly, there is no reason to want to have our views represented in the next generation of technology.

You could have a much more complex understanding of what they are. It isn’t nearly as simple as you are imagining. If you genuinely are curious about what you’re overlooking, then here is a link.

https://situational-awareness.ai/

If you are genuinely open to understanding the path we are on, the new situational awareness paper would be very eye-opening. It is 160 pages, so it’s probably a bit too much to get through, but there are really good videos that explain it. Matthew Berman has a great video about it. I’m not interested in swaying you and not going to debate, I’m 100s of hours deep into this and have been absolutely obsessed with it. Nobody doubted its impact as much as me. Education on the matter will undeniably change your mind tremendously. The information is there if you want a peak at the future.

https://situational-awareness.ai/

Thanks so much for taking the time to explain this. I was just going to give them a link.

It’s a much much bigger issue than this. Would you rather live in a world where other countries have good AI and you do not? Would you like it if only China has powerful AI? I get the copyright issue, but some things are more important than other things. This is an arms race, and everyone slowing down isn’t exactly an option.

Is it definitely a W that EU perspectives won’t be as represented in the AI programs that we are all using?

Is there a list of the malicious extensions? What should be done if we ha e malicious ones installed?

You complained about the Facebook engagement algorithm. I said they should be allowed to run the code and people use it if they choose. You disagreed.

It is a bit weird that you’ve flipped over to my side, and now you want freedom, and you’re trying to put me over on your original side. It’s nice that we both agree now. Nice chatting.

It’s not only advertisers. It is a need for engagement. Facebook makes money if people are engaged, both from advertisement and selling data. People prefer to use platforms that have lots of money to put into the user experince. Maybe this will change as people become more aware, maybe with things like the fediverse.

Oftentimes, things like murder and insider trading are at least attempted to be stopped, I don’t know what your point is there. This was a discussion on whether or not the government should stop Facebook from having code that keeps users engaged. I said it is better if the government doesn’t verify all the code that makes it on the internet. That is what the government does in places like North Korea.

I prefer an internet where anyone is free to share the code they want to as opposed to an internet where everything has to be submitted to an authority who has to ok it. Imagine all the innovation that would be stifled in a society where such a system was in place. If you think would prefer that, then maybe North Korea is the place where you would be happiest.

The reason websites have things like engagement algorithms is because they are advertisement based, and they sell user data. This seems shitty at first glance, but it is what people prefer. The alternative is subscription based. Both models have been presented, and people chose what they wanted. Nobody forced them. As time goes on, things evolve. I like to think that in the future, people will move more towards decentralized, community run websites. That’s why I am on Lemmy, and I am not on Facebook. I am certainly happy that I have the freedom to choose. I am also happy that anyone has the freedom to make whatever options they want to offer.

And amazingly, despite all these awful things you listed, nearly everything is better. People kill strangers less. People kill their families less. People rape each other less. People torture each other less. Countries go to war less. People live longer. People live healthier. People starve less. These things all used to be much much bigger than they are now.

Pick someone out of history and ask them if they would rather watch all their children die of a horrible plague, or they have to live in a world where some idiots are messaging each other about Q-Anon. What do you think they will say they prefer?

Personally, I’d much rather some idiots running into a special building wearing stupid costumes than deal with the absolute living hell of the pre-technology world absolutely any day. So would you.

Yeah, sure, there are things to dislike about computers. In the same way, ambulances suck because they are noisy. Ovens suck because they can overcook your food…

It is wild how people mocked computers/internet back in the day. Lots of people are too young to remember it. Here is Letterman mocking the internet while a crowd laughs along. It’s basically the same as when people mock LLMs/AI or blockchain or whatever other new technology that they don’t understand.