whoisyurii
5 minutes
Next-Token Prediction in Large Language Model Text Generation
How a language model produces text by repeatedly predicting the most probable next token from distributions learned during pre-training.