Contract Understanding on ContractNLI
4.04Expected Calibration Error (ECE)CaliDist(As.)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CaliDist(As.)LLM=Llama-3.1, Distractor Style=Assertion-style2026.06 | 4.04 | 22.59 | |
| CaliDist verbalized (As.)LLM=Llama-3.1, Distractor Style=Assertion-style2026.06 | 4.63 | 23.25 | |
| FSDLLM=Llama-3.1, Distractor Style=Assertion-style2026.06 | 19.16 | 26.1 | |
| ConsistencyLLM=Llama-3.1, Distractor Style=Assertion-style2026.06 | 23.84 | 27.67 | |
| EntropyLLM=Llama-3.1, Distractor Style=Assertion-style2026.06 | 28.7 | 30.58 | |
| Vanilla VerbalLLM=Llama-3.1, Distractor Style=Assertion-style2026.06 | 29.79 | 30.98 |