AI Paper Atlas

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

Variational Information Bottleneck for Effective Low-Resource Fine-Tuning

Rabeeh Karimi mahabadi, Yonatan Belinkov, James Henderson

ICLR 2021 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 38 citations (OpenAlex)

This paper tried a common change: Add dropout

Use dropout regularization · Low-resource GLUE: MRPC, STS-B, and RTE · Accuracy, F1, Pearson, and Spearman: MRPC 87.80 -> 87.33 accuracy; RTE 67.93 -> 65.80 accuracy Hurt

What 795 accepted papers report for this change

See this change in detailMatched by the wording of each paper's own result, read by AI from the full text. Papers report gains more often than failures, so this shows what gets published, not the chance a change works.

Read the paper

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
Add a VIB compression bottleneck during fine-tuningBERT Base · Low-resource GLUE: MRPC, STS-B, and RTEAccuracy, F1, Pearson, and Spearman: +1.43 accuracy, +2.03 F1, +2.7 Pearson, +2.97 Spearman, +2.6 accuracyImproved
Add a VIB compression bottleneckBERT · SNLI, 200 training samplesAccuracy: +2.72Improved
Add a VIB compression bottleneckBERT · MNLI, 200 training samplesAccuracy: +3.65Improved
Add a VIB compression bottleneckBERT · IMDB, 200 training samplesAccuracy: +4.09Improved
Add a VIB compression bottleneckBERT · YELP, 200 training samplesAccuracy: +0.7Improved
Add a VIB compression bottleneckBERT · SNLI, 6000 training samplesAccuracy: +0.44No clear effect
Use dropout regularizationBERT Base · Low-resource GLUE: MRPC, STS-B, and RTEAccuracy, F1, Pearson, and Spearman: MRPC 87.80 -> 87.33 accuracy; RTE 67.93 -> 65.80 accuracyHurt
Add a VIB bottleneck to suppress task-irrelevant featuresBERT trained on SNLI · 15 NLI transfer datasetsAverage accuracy improvement over BERT: +5.51 points; improves transfer on 13 out of 15 datasetsImproved
Add a VIB bottleneck to suppress task-irrelevant featuresBERT trained on MNLI · 15 NLI transfer datasetsAverage accuracy improvement over BERT: +3.83 points; improves transfer on 13 datasetsImproved
Add VIB regularization to the sentence representationBERT trained on SNLI · SNLI hypothesis-only probe, testAccuracy: 53.17 -> 37.10Improved
Remove the VIB compression loss by setting β=0 (ablation)VIBERT · MRPCAccuracy: 89.23 (0.1) -> 88.57 (0.6)Hurt
Remove the VIB compression loss by setting β=0 (ablation)VIBERT · MRPCF1: 85.23 (0.2) -> 84.27 (0.7)Hurt

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