Training 2B/7B language models on right-to-left text factorization yields a reverse posterior scorer that improves math reasoning when combined with forward likelihood.
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
-
LEDOM: Reverse Language Model
Training 2B/7B language models on right-to-left text factorization yields a reverse posterior scorer that improves math reasoning when combined with forward likelihood.