Theorem Proving on CombiBench (pass@32)
16pass@32WZ-LLM
Evaluation Results
| Method | Links | |
|---|---|---|
| WZ-LLMModel size=8B, Sample budget=pass@322026.05 | 16 | |
| WZ-ProverModel size=8B, Sample budget=pass@322026.05 | 15 | |
| WZ-uncoveredModel size=8B, Sample budget=pass@322026.05 | 15 | |
| WZ-Sketch + Goedel-Prover-V2Model size=8B, Sample budget=pass@322026.05 | 13 | |
| Goedel-Prover-V2-8BModel size=8B, Sample budget=pass@322026.05 | 12 | |
| DeepSeek-Prover V2-7BModel size=7B, Sample budget=pass@322026.05 | 8 | |
| Kimina-Prover Distill-8BModel size=8B, Sample budget=pass@322026.05 | 6 | |
| WZ-Sketch + WZ-ProverModel size=8B, Sample budget=pass@322026.05 | 1 |