As a result, our GPT-4 training run was (for us at least!) unprecedentedly stable, becoming our first large model whose training performance we were able to accurately predict ahead of time. We found and fixed some bugs and improved our theoretical foundations.
A year ago, we trained GPT-3.5 as a first “test run” of the system. Over the past two years, we rebuilt our entire deep learning stack and, together with Azure, co-designed a supercomputer from the ground up for our workload. We’ve spent 6 months iteratively aligning GPT-4 using lessons from our adversarial testing program as well as ChatGPT, resulting in our best-ever results (though far from perfect) on factuality, steerability, and refusing to go outside of guardrails. For example, it passes a simulated bar exam with a score around the top 10% of test takers in contrast, GPT-3.5’s score was around the bottom 10%. GPT-4 is a large multimodal model (accepting image and text inputs, emitting text outputs) that, while less capable than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks.
We’ve created GPT-4, the latest milestone in OpenAI’s effort in scaling up deep learning.
0 kommentar(er)
