Pith. sign in

REVIEW

DoT: An efficient Double Transformer for NLP tasks with tables

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

arxiv 2106.00479 v1 pith:DTED5ZYB submitted 2021-06-01 cs.CL

DoT: An efficient Double Transformer for NLP tasks with tables

classification cs.CL
keywords transformeraccuracymodelpruningtask-specifictokensdeepdouble
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Transformer-based approaches have been successfully used to obtain state-of-the-art accuracy on natural language processing (NLP) tasks with semi-structured tables. These model architectures are typically deep, resulting in slow training and inference, especially for long inputs. To improve efficiency while maintaining a high accuracy, we propose a new architecture, DoT, a double transformer model, that decomposes the problem into two sub-tasks: A shallow pruning transformer that selects the top-K tokens, followed by a deep task-specific transformer that takes as input those K tokens. Additionally, we modify the task-specific attention to incorporate the pruning scores. The two transformers are jointly trained by optimizing the task-specific loss. We run experiments on three benchmarks, including entailment and question-answering. We show that for a small drop of accuracy, DoT improves training and inference time by at least 50%. We also show that the pruning transformer effectively selects relevant tokens enabling the end-to-end model to maintain similar accuracy as slower baseline models. Finally, we analyse the pruning and give some insight into its impact on the task model.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.