REVIEW 1 cited by
Iterative Decoding for Compositional Generalization in Transformers
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Iterative Decoding for Compositional Generalization in Transformers
read the original 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.
Forward citations
Cited by 1 Pith paper
-
Distance-Preserving Embeddings in Inhomogeneous Random Graphs
On supercritical inhomogeneous random graphs, multi-scale landmark embeddings achieve (1±ε)-distortion of shortest paths at dimension Ω(n^{1-ε} log n), far below worst-case, with universal kernel extensions and transf...
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.