Fine-grained 9-way ill-posedness classification on CLAMBER 9-way
70.6AccuracyH0 topology
Evaluation results
| Method | Links | ||
|---|---|---|---|
| H0 topologyModel=Mistral-7B-Instruct-v0.32026.06 | 70.6 | 63.6 | |
| H0 topologyModel=Gemma-7B-it2026.06 | 69.1 | 62.2 | |
| H0 topologyModel=Llama-3.1-8B-Instruct2026.06 | 69.1 | 62.1 | |
| AENModel=Mistral-7B-Instruct-v0.32026.06 | 60.2 | 51.3 | |
| AENModel=Llama-3.1-8B-Instruct2026.06 | 57.8 | 50.3 | |
| Few-shotModel=Mistral-7B-Instruct-v0.32026.06 | 57.7 | 44.9 | |
| AENModel=Gemma-7B-it2026.06 | 54.7 | 46.6 | |
| Few-shot CoTModel=Mistral-7B-Instruct-v0.32026.06 | 37.4 | 32.4 | |
| Few-shot CoTModel=Llama-3.1-8B-Instruct2026.06 | 34 | 31 | |
| Few-shot CoTModel=Gemma-7B-it2026.06 | 27.6 | 26.2 | |
| Zero-shotModel=Mistral-7B-Instruct-v0.32026.06 | 27.3 | 14.6 | |
| Zero-shotModel=Gemma-7B-it2026.06 | 26.3 | 6.2 | |
| Few-shotModel=Llama-3.1-8B-Instruct2026.06 | 19.1 | 20.8 | |
| Zero-shotModel=Llama-3.1-8B-Instruct2026.06 | 16.4 | 15.4 | |
| Few-shotModel=Gemma-7B-it2026.06 | 12.2 | 8.6 |