Conceptnext-token generation / 自回归生成
Next-Token Prediction
A paradigm in which language models generate text serially token by token, predicting the probability distribution of the next word based on preceding context; it is the underlying generation method of current mainstream large language models.
Timeline (last 90 days)
Oct 23
The serial generation nature of autoregressive next-token prediction means each token requires a forward computation, resulting in significant latency when generating long texts.
Unverified50%