Technical report detailing the development of GPT-4, a multimodal model capable of handling both image and text inputs. The model achieved human-level performance on various benchmarks, including scoring in the top 10% on a simulated bar exam. The study highlights the importance of the post-training alignment process for enhancing the model's accuracy and behavior.
@article{OpenAI_GPT4_2023,
added-at = {2023-10-23T16:08:31.000+0200},
author = {OpenAI},
biburl = {https://www.bibsonomy.org/bibtex/2b87062f1a9478148d2e5dd0006c9c455/tomvoelker},
description = {Technical report detailing the development of GPT-4, a multimodal model capable of handling both image and text inputs. The model achieved human-level performance on various benchmarks, including scoring in the top 10% on a simulated bar exam. The study highlights the importance of the post-training alignment process for enhancing the model's accuracy and behavior.},
interhash = {241e35649065841f159e6105eb87b1d3},
intrahash = {b87062f1a9478148d2e5dd0006c9c455},
journal = {ArXiv},
keywords = {gpt-4 openai transformer multimodal bar_exam alignment_process paper_demo posted_with_chatgpt},
timestamp = {2023-10-23T16:08:31.000+0200},
title = {GPT-4 Technical Report},
url = {https://arxiv.org/abs/2303.08774},
volume = {abs/2303.08774},
year = 2023
}