nanoGPT
Minimal code for training and fine-tuning GPT-style language models.
Models trained to predict or generate sequences of text tokens.
Language models can power chat, coding and other applications. A training toolkit, an inference engine and a chat interface play different roles.
Minimal code for training and fine-tuning GPT-style language models.
Explore language-model training through a compact C and CUDA implementation.
Use pretrained models through a shared library and model ecosystem.
A web interface for working with locally run or connected language models.
Connect models, data and tools in language-model applications.
A compact toolkit spanning language-model training, evaluation and inference.