GPT
Family of decoder-only Transformer language models trained with a generative objective.
Обзор
GPT is a family of decoder-only Transformer language models trained to predict the next token on large text corpora. It solves the problem of producing fluent, contextually appropriate text without hand-built linguistic rules. It exists because the generative pre-training objective turned out to transfer well to tasks the model was never explicitly trained for, which removed the need for a labelled dataset per task. Later members of the family are the models most people mean when they say "large language model".
Граф технологий
Как GPT связана с остальным стеком.
Использует и зависит от
- Transformer architecture Algorithms GPT uses Transformer High ✓ source
- Transformer architecture Algorithms GPT based on Transformer High ✓ source
Инструменты
ПО, которое реализует GPT или интегрируется с ним.