AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

GAPX: Generalized Autoregressive Paraphrase-Identification X

Yifei Zhou, Renyu Li, Hayden Housen, Ser-Nam Lim

NeurIPS 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use only the positive model (drop negative model) at inference (ablation)autoregressive paraphrase model · OOD averagemacro F1: 52.9 -> 69.7Improved
use only the positive model (drop negative model) in-distribution (ablation)autoregressive paraphrase model · in-distribution averagemacro F1: 75.6 -> 60.4Hurt
replace fixed OODP with perplexity-weighted automatic ensembling (GAP) (ablation)OODP · OOD averagemacro F1: 69.7 -> 70.0No clear effect
replace fixed OODP with perplexity-weighted ensembling (GAP) (ablation)OODP · in-distribution averagemacro F1: 60.4 -> 68.8Improved
ensemble GAP with RoBERTa discriminative model (GAPX) (ablation)GAP · in-distribution averagemacro F1: 68.8 -> 78.1Improved
ensemble GAP with RoBERTa discriminative model (GAPX) (ablation)GAP · OOD averagemacro F1: 70.0 -> 67.7Hurt
use GAPX instead of best transformer baseline BARTGAPX · OOD averagemacro F1: 56.2 -> 67.7Improved
add expert-product debiasing to BERTBERT · OOD averagemacro F1: 54.4 -> 48.7Hurt
replace perplexity with ENERGY OOD score for ensembling (ablation)GAPX · OOD averagemacro F1: 67.7 -> 60.6Hurt
replace perplexity with SOFTMAX OOD score for ensembling (ablation)GAPX · OOD averagemacro F1: 67.7 -> 55.3Hurt
replace perplexity with negative log-likelihood (ablation)GAPX · OOD averagemacro F1: 67.7 -> 66.8No clear effect
ensemble GAP with IDP instead of RoBERTa (ablation)GAP · OOD averagemacro F1: 67.7 -> 64.2Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy