We’ve created GPT-4, the latest milestone in OpenAI’s effort in scaling up deep learning. GPT-4 is a large multimodal model (accepting image and text inputs, emitting text outputs) that, while worse than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks.
tldw: Brockman showed up some straightforward demos of GPT-4's text & code writing versatility, and some limited demo of its image processing, but you aren't missing anything insightful about the arch/training/scaling/future/etc.
2Quadratic Reciprocity
The really cool bit was when he had a very quick mockup of a web app drawn on a piece of paper and uploaded a photo of it and GPT-4 then used just that to write the HTML and JavaScript for the app based on the drawing.
Full paper available here: https://cdn.openai.com/papers/gpt-4.pdf