AI Paper Atlas

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

Data Augmentations for Improved (Large) Language Model Generalization

Amir Feder, Yoav Wald, Claudia Shi, Suchi Saria, David Blei

NeurIPS 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 3 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
add counterfactual augmentation (CATO B) instead of observational traininglogistic regression · CeBAB-Spuriousaccuracy: 0.64 -> 0.75Improved
add naive LLM augmentationlogistic regression · CeBAB-Spuriousaccuracy: 0.64 -> 0.62Hurt
add naive LLM augmentationlogistic regression · CeBABaccuracy: 0.85 -> 0.80Hurt
use conditional augmentation without diff-in-difflogistic regression · CeBAB-Spuriousaccuracy: 0.64 -> 0.70Improved
add diff-in-diff counterfactual augmentation (Causal Aug.)PubMED BERT · i2b2-2006 OOD demographic traitsF1: 66.92 -> 71.88Improved
add causal augmentationPubMED BERT · i2b2-2006 OOD demographic traitsprecision: 53.32 -> 59.76Improved
add naive augmentationPubMED BERT · i2b2-2006 OOD demographic traitsF1: 66.92 -> 66.52No clear effect
add naive augmentationPubMED BERT · MIMIC-III ID demographic traitsF1: 79.34 -> 80.39Improved
add causal augmentationPubMED BERT · MIMIC-III ID demographic traitsF1: 79.34 -> 79.73No clear effect
add re-weightingPubMED BERT · i2b2-2006 OOD demographic traitsF1: 66.92 -> 70.02Improved
replace PubMED BERT with GPT3 as baseline modelobservational baseline · i2b2-2006 OOD demographic traitsF1: 66.92 -> 63.98Hurt
add causal augmentationPubMED BERT · i2b2-2010 OOD clinical conditionprecision: more than 7% absoluteImproved

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