Publication/Creation Date

March 14 2023

Creators/Contributors

OpenAI (creator)

Description

Abstract:

We report the development of GPT-4, a large-scale, multimodal model which can accept image and text inputs and produce text outputs. While less capable than humans in many real-world scenarios, GPT-4 exhibits human-level performance on various professional and academic benchmarks, including passing a simulated bar exam with a score around the top 10% of test takers. GPT-4 is a Transformer based model pre-trained to predict the next token in a document. The post-training alignment process results in improved performance on measures of factuality and adherence to desired behavior. A core component of this project was developing infrastructure and optimization methods that behave predictably across a wide range of scales. This allowed us to accurately predict some aspects of GPT-4’s performance based on models trained with no more than 1/1,000th the compute of GPT-4.

Augments

Communicating, Understanding, Learning, Working, Harming, Writing

Technology Keywords

Artificial Intelligence (AI), Artificial General Intelligence (AGI), Large Language Models (LLMs), Large Multimodal Models, Natural Language Generation (NLG), Natural Language Processing (NLP), Text Summarizer, Machine Translation, Reinforcement Learning From Human Feedback (RLHF), Generative AI, Visual Recognition

Keywords

Language, Communication, Translation, Education, Safety, Ethics, Privacy, Bias, Disinformation, Information, Trust, Accuracy, Economics, Work, Security, Research, Cognitive Enhancement, Automation

Marketing Keywords

OpenAI, OpenAI GPT-4, OpenAI GPT-3.5, OpenAI ChatGPT

Source

https://cdn.openai.com/papers/gpt-4.pdf