Open-domain task on Arena-Hard (test)
12.61ErrorC-PAC
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| C-PACClustering type=Disjoint clustering, Uncertainty=Verbalized, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 12.61 | 0 | 22.5 | |
| C-PACClustering type=Joint clustering, Uncertainty=Verbalized, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 12.7 | 0 | 23.06 | |
| C-PACClustering type=Disjoint clustering, Uncertainty=Logits, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 13.08 | 0 | 36.63 | |
| PACClustering type=Disjoint clustering, Uncertainty=Router, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 13.16 | 8.67 | 13 | |
| C-PACClustering type=Disjoint clustering, Uncertainty=Router, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 13.18 | 0 | 36.32 | |
| C-PACClustering type=Joint clustering, Uncertainty=Router, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 13.2 | 0 | 35.96 | |
| PACClustering type=Joint clustering, Uncertainty=Verbalized, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 13.23 | 7.1 | 28.95 | |
| PACClustering type=Joint clustering, Uncertainty=Router, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 13.3 | 9.38 | 13.49 | |
| C-PACClustering type=Joint clustering, Uncertainty=Logits, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 13.32 | 0 | 38.05 | |
| PACClustering type=Disjoint clustering, Uncertainty=Verbalized, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 13.34 | 6.35 | 29.49 | |
| PACClustering type=Joint clustering, Uncertainty=Logits, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 13.46 | 3.91 | 42.62 | |
| PACClustering type=Disjoint clustering, Uncertainty=Logits, epsilon=0.15, Backbone=Llama-based LLMs2026.01 | 13.51 | 3.88 | 42.7 |