A forward curriculum that gradually increases the number of predicted tokens helps small models retain MTP's decoding speed benefits and often improves generation quality, but does not consistently improve next-token prediction accuracy.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Pre-Training Curriculum for Multi-Token Prediction in Language Models
A forward curriculum that gradually increases the number of predicted tokens helps small models retain MTP's decoding speed benefits and often improves generation quality, but does not consistently improve next-token prediction accuracy.