CUAD

CUAD: Measures legal reasoning, contract review, statute interpretation, or legal-domain QA.

10rows
precision_at_80_recallprimary metric
2026-05-27sampled

Metadata

Metrics

Precision at 80% recall, AUPR, Precision at 90% recall

Latest Results

Rows are parsed from the CUAD paper arXiv LaTeX model results table.

Rank Subject Precision at 80% recall Model Match Provenance Sampled
1 DeBERTa-xlarge 44.0 — Imported 2026-05-27
2 RoBERTa-large 38.1 — Imported 2026-05-27
3 RoBERTa-base + Contracts Pretraining 34.1 — Imported 2026-05-27
4 RoBERTa-base 31.1 — Imported 2026-05-27
5 ALBERT-xxlarge 31.0 — Imported 2026-05-27
6 ALBERT-large 20.9 — Imported 2026-05-27
7 ALBERT-xlarge 20.5 — Imported 2026-05-27
8 ALBERT-base 11.1 — Imported 2026-05-27
9 BERT-base 8.2 — Imported 2026-05-27
10 BERT-large 7.6 — Imported 2026-05-27