Natural Language Inference Distribution Estimation on ChaosNLI
0KL DivergenceChaosNLI HJD
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ChaosNLI HJDLLM=Baseline2024.12 | 0 | 0 | 0 | 1 | |
| GPT-4o + human explanationsLLM=GPT-4o, Explanation Type=Human2024.12 | 0.187 | 0.207 | 0.223 | 0.769 | |
| GPT-4o VariErr Label-GuidedLLM=GPT-4o, Explanation Type=VariErr Label-Guided2024.12 | 0.192 | 0.209 | 0.226 | 0.761 | |
| Llama3 VariErr Label-GuidedLLM=Llama3, Explanation Type=VariErr Label-Guided2024.12 | 0.234 | 0.247 | 0.266 | 0.76 | |
| Llama3 + human explanationsLLM=Llama3, Explanation Type=Human2024.12 | 0.238 | 0.25 | 0.269 | 0.771 | |
| Llama3 MNLI Label-GuidedLLM=Llama3, Explanation Type=MNLI Label-Guided2024.12 | 0.242 | 0.251 | 0.275 | 0.849 | |
| GPT-4o Label-FreeLLM=GPT-4o, Explanation Type=Label-Free2024.12 | 0.252 | 0.242 | 0.275 | 0.752 | |
| Llama3LLM=Llama3, Explanation Type=None2024.12 | 0.259 | 0.262 | 0.284 | 0.689 | |
| GPT-4oLLM=GPT-4o, Explanation Type=None2024.12 | 0.265 | 0.263 | 0.289 | 0.703 | |
| Llama3 Label-FreeLLM=Llama3, Explanation Type=Label-Free2024.12 | 0.295 | 0.278 | 0.31 | 0.744 | |
| MNLI distributionLLM=Baseline2024.12 | 1.242 | 0.281 | 0.295 | 0.795 | |
| VariErr distributionLLM=Baseline2024.12 | 3.604 | 0.282 | 0.296 | 0.688 |