Natural Language Inference on MSciNLI
68.27AUROCCALIDIST
Evaluation Results
| Method | Links | |
|---|---|---|
| CALIDISTLLM Backbone=Qwen-3, Style=Assertion-style2026.06 | 68.27 | |
| VanillaLLM Backbone=Qwen-32026.06 | 63.83 | |
| ConsistencyLLM Backbone=Qwen-32026.06 | 52.67 | |
| EntropyLLM Backbone=Qwen-32026.06 | 52.67 | |
| FSDLLM Backbone=Qwen-32026.06 | 52.67 |