Explore legal task completion, citation accuracy and contract analysis with model and workflow comparisons.
Rubric criteria met across public legal tasks using Cortega’s legal workflow and a mixture of models.
Automated grading against public task rubrics; no lawyer review. One run was evaluated. The result does not measure lawyer acceptance, review time or performance on every matter. Other evaluations use different tasks and grading.
Harvey LAB tasks graded against their rubrics across 25 practice areas.
Compare citation accuracy with and without source checks.
Measure contract provision extraction and model review against labelled examples.
Explore the results by practice area.