sara_entailment
- Task Description: Given a statute, a fact pattern, and an assertion, determine if the assertion is 'entailed' by the fact pattern and statute.
- Task Type: Binary classification
- Document Type: fact pattern
- Number of Samples: 276
- Input Length Range: 59-314 tokens
- Evaluation Metrics: accuracy (maximize), balanced_accuracy (maximize), f1_macro (maximize), f1_micro (maximize), valid_predictions_ratio (maximize)
- Tags: interpretation, tax law
- Paper: LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in Large Language Models
- Dataset Download: https://hazyresearch.stanford.edu/legalbench/
7 submissions
| Rank | Model | accuracy | balanced_accuracy | f1_macro | f1_micro | valid_predictions_ratio | Date | Results |
|---|---|---|---|---|---|---|---|---|
| 1 | meta-llama/Meta-Llama-3.1-405B-Instruct-Turbo | 0.906 | 0.700 | 0.759 | 0.906 | 0.235 | 2025-08-04 | View |
| 2 | claude-3-5-haiku-20241022 | 0.883 | 0.883 | 0.883 | 0.883 | 0.971 | 2025-08-01 | View |
| 3 | meta-llama/Meta-Llama-3.1-70B-Instruct-Turbo | 0.875 | 0.833 | 0.855 | 0.875 | 0.029 | 2025-07-25 | View |
| 4 | gpt-4o-mini | 0.732 | 0.732 | 0.729 | 0.732 | 1.000 | 2025-07-02 | View |
| 5 | google/gemma-2-27b-it | 0.655 | 0.650 | 0.612 | 0.655 | 0.522 | 2025-07-24 | View |
| 6 | claude-3-haiku-20240307 | 0.518 | 0.500 | 0.341 | 0.518 | 0.809 | 2025-07-25 | View |
| 7 | gpt-4.1-nano | 0.496 | 0.478 | 0.332 | 0.496 | 0.956 | 2025-07-03 | View |