AI Paper Atlas

Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning

Co-training Improves Prompt-based Learning for Large Language Models

Hunter Lang, Monica N Agrawal, Yoon Kim, David Sontag

ICML 2022 · Label-efficient and robust learning › Zero-/few-shot and meta-learning · approach: Robust fine-tuning of zero-shot models · 17 citations (OpenAlex)

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 co-training to the GPT-3 label model (ablation)GPT-3 label model · RTEaccuracy: 62.8 -> 64.9Improved
Add co-training to the GPT-3 label model (ablation)GPT-3 label model · CBaccuracy: 76.8 -> 83.5Improved
Add co-training to the GPT-3 label model (ablation)GPT-3 label model · TRECaccuracy: 77.2 -> 78.3Improved
Add co-training to the T0-3B soft prompt (ablation)T0-3B · RTEaccuracy: 68.9 -> 87.0Improved
Add co-training to the T0-3B soft prompt (ablation)T0-3B · CBaccuracy: 58.9 -> 67.9Improved
Add co-training to the T0-3B soft prompt (ablation)T0-3B · BoolQaccuracy: 56.4 -> 49.1Hurt
Use GPT-3 prompt outputs to co-train DeBERTa-largeDeBERTa-large · RTEaccuracyImproved
Use GPT-3 prompt outputs to co-train DeBERTa-largeDeBERTa-large · CBaccuracyImproved
Use GPT-3 prompt outputs to co-train DeBERTa-largeDeBERTa-large · TRECaccuracyImproved
Use T0-3B prompt outputs to train DeBERTa-largeDeBERTa-large · BoolQaccuracy: 56.4 -> 48.9Hurt
Use T0-3B prompt outputs to train DeBERTa-largeDeBERTa-large · CBaccuracy: 58.9 -> 67.9Mixed
Use a cut statistic to select confident data for the T0 viewT0-3B co-training · CBpseudo-label accuracy: 89%No clear effect

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