zlacker

[parent] [thread] 6 comments
1. vinter+(OP)[view] [source] 2023-11-20 07:45:54
I'm also pretty suspicious of people in forums like these who say nothing can compare to GPT4 and they're miles ahead of everyone else etc. How much of that is venture capital speaking?

It's not quite where it is (or was) with Tesla, where it was hopeless to know what was sincere and what was just people talking up their investment/talking down their short, but it's getting there.

replies(4): >>tempus+32 >>lajawf+42 >>xvecto+h4 >>exizt8+a5
2. tempus+32[view] [source] 2023-11-20 07:58:33
>>vinter+(OP)
There are concrete benchmarks like “how good is it at answering multiple choice questions accurately or “how good is it at producing valid code to solve a particular coding problem”.

There’s also a chatbot Elo ranking which crowd sources model comparisons https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboar...

GPT-4 is the king right now

3. lajawf+42[view] [source] 2023-11-20 07:58:36
>>vinter+(OP)
The crypto bros switched to AI hype and are now hyping OpenAI/GPT4 hoping to pump MSFT/NVDA. In every HN conversation where someone mentions competing products, there are people talking it down and saying GPT4 is miles ahead, and in a tone to undermine the competition. I see a pattern and it is definitely not sincere.
replies(1): >>xvecto+G4
4. xvecto+h4[view] [source] 2023-11-20 08:07:53
>>vinter+(OP)
I mean, try and compare for yourself. It is quite obviously miles ahead of everything else.

I want OpenAI to be absolutely crushed in the free market after this move. But it will take years for anyone to catch up with GPT-4, if even Anthropic is nowhere close.

replies(1): >>Occams+t6
◧◩
5. xvecto+G4[view] [source] [discussion] 2023-11-20 08:08:46
>>lajawf+42
You clearly haven't tried GPT-4 if you think people are lying about how much better it is.
6. exizt8+a5[view] [source] 2023-11-20 08:10:48
>>vinter+(OP)
Anyone who works with text generation will tell you that GPT-4 is far, far beyond anything anyone else has put out for general purpose text gen. The benchmarks don’t really tell you the whole picture. It’s impossible to prompt other models for anything as complex as what GPT-4 can do, both semantically and stylistically.
◧◩
7. Occams+t6[view] [source] [discussion] 2023-11-20 08:16:31
>>xvecto+h4
Why do you want them to be crushed? They decided that Sam didn't represent the charter and acted accordingly. Do I think it was a boneheaded move? Sure. But maybe it was the right move for them even in spite of the optics?
[go to top]