AI Paper Atlas

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

Unsupervised Cross-Task Generalization via Retrieval Augmentation

Bill Yuchen Lin, Kangmin Tan, Chris Scott Miller, Beiwen Tian, Xiang Ren

NeurIPS 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 18 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
retrieve with upstream-model hidden representations (ReCross no rerank) instead of SBERTReCross · unseen tasks meanSoftEM: 40.24 -> 43.57Improved
add re-ranking trained by distant supervision (ablation)ReCross · unseen tasks meanSoftEM: 43.57 -> 44.53Improved
ReCross retrieval augmentation over BART0BART0 · unseen tasks meanSoftEM: 40.38 -> 44.53Improved
ReCross on BIG-bench subsetBART0 · BIG-bench subsetSoftEM: 28.07 -> 35.23Improved
random retrieval instead of no retrievalBART0 · unseen tasks meanSoftEM: 40.38 -> 39.13Hurt
increase query size from 16 to 32 (ablation)ReCross · unseen tasks meanSoftEM: 44.53 -> 42.52Hurt
reduce query size to 1 (ablation)ReCross · unseen tasks meanSoftEM: 44.53 -> 43.20Hurt
reduce retrieved size from 512 to 256 (ablation)ReCross · unseen tasks meanSoftEM: 44.53 -> 40.80Hurt
increase retrieved size from 512 to 1024 (ablation)ReCross · unseen tasks meanSoftEM: 44.53 -> 44.02No clear effect
increase upsampling ratio from 2 to 3 or 4 (ablation)ReCross · unseen tasks meanSoftEM: 44.53 -> 43.92 (mu=3), 43.91 (mu=4)Hurt
perturb template instructions by concatenation (ablation)ReCross · unseen tasks meanSoftEM: 44.53 -> 40.52Hurt
perturb template instructions by random word change (ablation)ReCross · unseen tasks meanSoftEM: 44.53 -> 42.07Hurt

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