Revision History
Natural language inference (NLI) · 5 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: Replaced fabricated and mislabeled 'SNLI paper' examples with the paper's real examples, and corrected human-accuracy figures (SNLI 89.0% annotator agreement, not 87.7%; MultiNLI 92.0% from Nangia & Bowman 2019; ANLI 84-86% verifier agreement, not ~92%), BERT's MultiNLI score (86.7%, not over 90%), inflated SNLI/MNLI accuracy claims (high/mid 90s), the DeBERTa SuperGLUE model (original 1.5B DeBERTa, not V3), RTE-3's top accuracy (80%), XNLI's construction (new data, not translated MultiNLI sets), ANLI's size, a misattributed Yin et al. quote, and HELM's coverage of ANLI (plus 13 minor fixes)
Version 5 · Nov 5, 2026, 03:09 AM
- Improve internal links (2026-10-13)
Version 4 · Oct 13, 2026, 10:19 AM
- Added 6 contextual internal links
Version 3 · Jul 23, 2026, 03:29 PM