I have actively tried to incorporate ChatGPT in my everyday life as a dev and architect. ChatGPT is mostly a Litmus test when it comes to coding. If you are impressed by the version before this you are most likely a beginner. ChatGPT is mostly wrong when it comes to any advanced qs in maths or software development. It often gives code that uses features, options, responses in APIs that simple does not exists. Would love to try this version out... It will probably suck too.
GPT-4
141–150 of 1001 posts
Re: GPT-4
#142Since it’s trained on a specialized supercomputer I doubt we’ll be seeing an open source or non-OpenAI version of this for the next couple years at least. Sad to say it but OpenAI has successfully privatized AI
They're using specialized hardware to accelerate their development feedback loop. Without a doubt researchers and hackers will find ways to cut down model sizes and complexity, to run on consumer hardware, soon enough. Just use stable diffusion as an example: 4GB for the whole model. Even if text models are 16GB that'd be great.
Re: GPT-4
#143> What are the implications for society when general thinking, reading, and writing becomes like Chess? I think going from LSAT to general thinking is still a very, very big leap. Passing exams is a really fascinating benchmark but by their nature these exams are limited in scope, have very clear assessment criteria and a lot of associated and easily categorized data (like example tests). General thought (particularl…
Peak denialism? Answering LSAT questions requires general intelligence. They present real life scenarios that test-taker has to understand. It requires "common sense" knowledge about the world and reasoning ability. It's not something you can memorize answers to or solve by following prescribed patterns or templates. And GPT-4 wasn't trained specifically to solve LSAT questions.
Re: GPT-4
#144I'd really like to use the openai API for personal projects, but it seems they only offer paying via credit/debit card. Don't really want to get one just for that... :-(
How else would you pay?
Re: GPT-4
#145This is huge: "Rather than the classic ChatGPT personality with a fixed verbosity, tone, and style, developers (and soon ChatGPT users) can now prescribe their AI’s style and task by describing those directions in the 'system' message."
System message is available today (and has been) in the playground under the chat setting.
Re: GPT-4
#146> Given both the competitive landscape and the safety implications of large-scale models like GPT-4, this report contains no further details about the architecture (including model size), hardware, training compute, dataset construction, training method, or similar. "Open"
at least they admit the competitive landscape is a factor rather than going 100% with "it's for safety reasons". I'm sure somebody will release an equivalent soon, the way open source has completely surpassed OpenAI when they try to keep things closed like DALLE vs Stable Diffusion shows that OpenAI really isn't that special, they just have a sweetheart deal with Microsoft
Re: GPT-4
#147“GPT-4 can also be confidently wrong in its predictions, not taking care to double-check work when it’s likely to make a mistake. Interestingly, the base pre-trained model is highly calibrated (its predicted confidence in an answer generally matches the probability of being correct). However, through our current post-training process, the calibration is reduced.”
Interesting that the post-training has that effect.
Re: GPT-4
#148Since it’s trained on a specialized supercomputer I doubt we’ll be seeing an open source or non-OpenAI version of this for the next couple years at least. Sad to say it but OpenAI has successfully privatized AI
I'm drawn to disliking OpenAI for not being open, but on the other hand, as long as the architectures and techniques are public, progress will continue fast. If OpenAI drops the ball and stops improving, another company would just take their place.
Edit: never mind. "Given both the competitive landscape and the safety implications of large-scale models like GPT-4, this report contains no further details about the architecture (including model size), hardware, training compute, dataset construction, training method, or similar."
Re: GPT-4
#149OpenAI is located in the same building as Musk's Neuralink. Can't wait for this to be implanted in babies at birth!
https://www.youtube.com/watch?v=O2RIvJ1U7RE
Re: GPT-4
#150I cant wait for this to do targeted censorship! It already demonstrates it has strong biases deliberately programmed in: > I cannot endorse or promote smoking, as it is harmful to your health. But it would likely happily promote or endorse driving, skydiving, or eating manure - if asked in the right way.
Can't wait till they inject ads am disguised as product biases into the responses in order to monetize it.
User: What should I use to water my plants?
ChatGPT: Brawndo's got what plants crave. It's got electrolytes.
User: But what are electrolytes?
CharGPT: They're what plants crave. You know, the stuff Brawndo has.