AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Defending Against Neural Fake News

Rowan Zellers, Ari Holtzman, Hannah Rashkin, Yonatan Bisk, Ali Farhadi, Franziska Roesner, Yejin Choi

NeurIPS 2019 · Large language models › Language, safety and interpretability · approach: Controlled text generation · 90 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 metadata as context when generating article bodiesGROVER · April 2019 test setperplexity: between .6 to .9 perplexity pointsImproved
Increase model size to GROVER-MegaGROVER · April 2019 test set, conditionalperplexity: 8.7Improved
Use GROVER as a discriminator rather than the best existing discriminatorsNeural fake news detection · Unpairedaccuracy: 73% -> 92%Improved
Rewrite propaganda articles with GROVERHuman Propaganda · Propaganda articlesoverall trustworthiness: 2.19 to 2.42Improved
Use a larger generator than the discriminatorGROVER detection · Neural fake news detectionaccuracy: below 81%Hurt
Use a larger discriminator than the generatorGROVER detection · Neural fake news detectionaccuracy: above 98%Improved
Add weakly supervised examples generated by GROVER-Large when few GROVER-Mega examples are availableGROVER discriminator · Detection of GROVER-Mega generationsaccuracy: around 50% -> around 78%Improved
Use random sampling without variance restrictionGROVER generation · Generated article bodiesperplexityHurt
Use nucleus sampling to restrict generation varianceGROVER generation · Generated article bodiesperplexityImproved
Use an excessively low top-p thresholdGROVER generation · Neural fake news detectiondiscrimination difficultyHurt

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