ContractGraphEval

150 contract clauses, each with a reference graph, a model-predicted graph, scores from several judges, and uncertainty estimates. Benchmark of "LLM as a Judge for Evaluating Contract Graphs" (Procedia Computer Science 275, 2026).

Paper · Code · Source dataset

Use

from datasets import load_dataset

ds = load_dataset("moriyad/ContractGraphEval", split="validation")
s = ds[0]
s["clause"]["text"], s["judge_scores"]["hybrid"], s["error_prediction"]["p_error_calibrated"]

Fields

Field	Description
clause, instruction	The clause and the extraction prompt
gold_graph, prediction	Reference graph and the output of the GRPO-trained Llama-3.1-8B model
judge_scores	8 reference-based scores in [0, 1]
reference_free_judges	7 checks that need no reference graph, computed with the contract_linter package
uncertainty, risk_scores, error_prediction	Judge disagreement, per-judge risk, and calibrated error probability
evaluation	has_error, mean_score, structural_f1
source_id	The same clause in the source dataset

Error labels are derived from judge scores against the reference graph.

Notes

• The samples are validation rows of the source dataset, from 3 contracts that were not used for training.

Licence and citation

cc-by-nc-nd-4.0. Clause text comes from CUAD (The Atticus Project, CC BY 4.0).

@article{dechtiar2026llmjudge,
  title   = {LLM as a Judge for Evaluating Contract Graphs: Multi-Judge Benchmarking and Agentic Uncertainty-Aware Refinement},
  author  = {Dechtiar, Moriya and Katz, Daniel Martin and Jaume, Sylvain and Wang, Hongming},
  journal = {Procedia Computer Science},
  volume  = {275},
  pages   = {513--522},
  year    = {2026},
  doi     = {10.1016/j.procs.2026.01.060}
}