Natural Language Inference on contract_nli LegalBench
81.7AccuracyAuto-EISR
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Auto-EISRn=71, Executors=deepseek-v3.2, kimi-k2.5, kimi-k2.6 (mean), val_frac=0.35, max_epochs=4, patience=3, seed=422026.06 | 81.7 | — | 1.9 | |
| v1n=71, Executors=deepseek-v3.2, kimi-k2.5, kimi-k2.6 (mean), val_frac=0.35, max_epochs=4, patience=3, seed=422026.06 | 79.8 | 8.9 | — | |
| B1bn=71, Executors=deepseek-v3.2, kimi-k2.5, kimi-k2.6 (mean), val_frac=0.35, max_epochs=4, patience=3, seed=422026.06 | 70.9 | — | — |