Teknosys learning library

GPT: autoregressive text generation

GPT-style models use a decoder-based transformer with causal attention: each new token is predicted from the preceding context. Pretraining learns language patterns, while instruction tuning can adapt the model to follow requests. These models can generate drafts, explanations, summaries, translations, and code. Their output is probabilistic, not a guaranteed fact lookup, and should be checked for errors or invented details.

Educational material. Verify examples against current documentation and applicable licenses. Report a correction.