Perplexity
Perplexity measures how uncertain a language model is when predicting a sequence, with lower values indicating better prediction.
#NextTokenPrediction
Explore AIstify's latest reporting, research, and expert analysis tagged with "next-token prediction", collected in one continuously updated archive.
Related Posts
Causal Language Modeling (CLM)
Causal language modeling trains a model to predict each next token without accessing future tokens in the sequence.
Autoregressive Model
An autoregressive model predicts each next value or token using values that appeared earlier in the sequence.