Two modular neuro-symbolic systems learn to simplify formulas by imitating term rewriting steps, and the new FastNRS variant generalizes to deeper formulas while being far faster than the original.
Iterative Decoding for Compositional Generalization in Transformers
1 Pith paper cite this work. Polarity classification is still indexing.
abstract
Deep learning models generalize well to in-distribution data but struggle to generalize compositionally, i.e., to combine a set of learned primitives to solve more complex tasks. In sequence-to-sequence (seq2seq) learning, transformers are often unable to predict correct outputs for longer examples than those seen at training. This paper introduces iterative decoding, an alternative to seq2seq that (i) improves transformer compositional generalization in the PCFG and Cartesian product datasets and (ii) evidences that, in these datasets, seq2seq transformers do not learn iterations that are not unrolled. In iterative decoding, training examples are broken down into a sequence of intermediate steps that the transformer learns iteratively. At inference time, the intermediate outputs are fed back to the transformer as intermediate inputs until an end-of-iteration token is predicted. We conclude by illustrating some limitations of iterative decoding in the CFQ dataset.
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Learning neuro-symbolic convergent term rewriting systems
Two modular neuro-symbolic systems learn to simplify formulas by imitating term rewriting steps, and the new FastNRS variant generalizes to deeper formulas while being far faster than the original.