Iterative self-training on a model's own correct outputs, with simple length and voting filters, lets transformers generalize to far longer arithmetic and path-finding problems than they saw in training.
Exploring length generalization in large language models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Self-Improving Transformers Overcome Easy-to-Hard and Length Generalization Challenges
Iterative self-training on a model's own correct outputs, with simple length and voting filters, lets transformers generalize to far longer arithmetic and path-finding problems than they saw in training.