Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
parquet
Sub-tasks:
natural-language-inference
Languages:
English
Size:
100K - 1M
License:
Ctrl+K
evaluate.py: premise-only truncation (the hypothesis is never cut; longest-first only if the hypothesis alone cannot fit)
bd78ad9 verified - anli_r1
- anli_r2
- anli_r3
- beetle
- bionli
- biosift
- calibration
- conjnli
- contractnli_a
- contractnli_b
- control
- crossingguard
- dialfact
- docnli
- fin_topics
- findver
- fol
- frank_sent
- glue_diagnostic
- hans
- healthver
- legal_domain
- lingnli
- med
- mnli
- mnli_mismatched
- nan_nli
- nli4ct
- nmonli_test
- nope_adv
- pubhealth
- ragtruth
- scientsbank
- scifact
- scitail
- scone_test
- sdoh_nli
- seahorse
- sick
- snli
- vast
- vitaminc
- wanli
- wice
- zs_meta
- 2.5 kB
- 7.7 kB
- 18.2 kB
- 8.48 kB