From Wikipedia Quality
Jump to: navigation, search

Generative Pre-trained Transformer 4' (GPT-4) is an autoregressive language model developed by OpenAI. It is the fourth generation in the GPT-n series and represents a significant advancement in natural language processing (NLP) capabilities compared to its predecessor, GPT-3. GPT-4 was introduced in March 2023 and has been widely recognized for its enhanced performance and versatility.

GPT-4 builds on the architecture and capabilities of GPT-3 but includes several key improvements. GPT-4 has significantly more parameters than GPT-3, allowing it to process and generate text with greater accuracy and nuance. While GPT-3 has 175 billion parameters, GPT-4 boasts several trillion parameters, enhancing its ability to understand and generate human-like text.[1]. It has improved contextual understanding, enabling it to generate more coherent and contextually relevant responses. This improvement is particularly noticeable in complex tasks that require deep comprehension and the ability to maintain context over longer interactions. Additionally, GPT-4 offers enhanced fine-tuning capabilities, allowing developers to tailor the model more effectively to specific tasks and domains. This adaptability makes GPT-4 more versatile and applicable to a broader range of applications.

Like GPT-3, GPT-4 was trained on a diverse range of internet text, with a significant portion of its dataset comprising Wikipedia articles. The inclusion of Wikipedia in GPT-4's training data was instrumental in achieving its high performance across various NLP tasks.

See Also


  1. Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F.L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S. and Avila, R., 2023. GPT-4 technical report. arXiv preprint arXiv:2303.08774.