Question Answering on TriviaQA (ECE, BS, AUC)
0.044ECEDINCO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DINCOBackbone=Llama-3.2-3B-Instruct2025.09 | 0.044 | 0.148 | 86.4 | |
| SCBackbone=Llama-3.2-3B-Instruct2025.09 | 0.065 | 0.177 | 80.8 | |
| NVCBackbone=Qwen3-1.7B2025.09 | 0.084 | 0.164 | 80.6 | |
| DINCOBackbone=Qwen3-8B2025.09 | 0.097 | 0.155 | 87.9 | |
| MSPBackbone=Qwen3-1.7B2025.09 | 0.104 | 0.186 | 77.4 | |
| DINCOBackbone=Gemma-3-4B-IT2025.09 | 0.121 | 0.191 | 81.7 | |
| SC-VCBackbone=Llama-3.2-3B-Instruct2025.09 | 0.122 | 0.211 | 76.1 | |
| K-VCBackbone=Llama-3.2-3B-Instruct2025.09 | 0.146 | 0.228 | 67.8 | |
| MSPBackbone=Qwen3-8B2025.09 | 0.149 | 0.203 | 81.9 | |
| NVCBackbone=Llama-3.2-3B-Instruct2025.09 | 0.168 | 0.192 | 84.5 | |
| NVCBackbone=Qwen3-8B2025.09 | 0.171 | 0.19 | 85.3 | |
| DINCOBackbone=Qwen3-1.7B2025.09 | 0.177 | 0.179 | 83.5 | |
| VCBackbone=Llama-3.2-3B-Instruct2025.09 | 0.189 | 0.208 | 82.6 | |
| NVCBackbone=Gemma-3-4B-IT2025.09 | 0.218 | 0.236 | 79.1 | |
| SCBackbone=Qwen3-1.7B2025.09 | 0.233 | 0.229 | 78 | |
| SCBackbone=Qwen3-8B2025.09 | 0.236 | 0.244 | 78.5 | |
| VCBackbone=Qwen3-8B2025.09 | 0.24 | 0.242 | 81.7 | |
| MSPBackbone=Llama-3.2-3B-Instruct2025.09 | 0.243 | 0.253 | 76.4 | |
| MSPBackbone=Gemma-3-4B-IT2025.09 | 0.252 | 0.268 | 79 | |
| K-VCBackbone=Gemma-3-4B-IT2025.09 | 0.254 | 0.262 | 78.6 | |
| SC-VCBackbone=Qwen3-8B2025.09 | 0.299 | 0.325 | 70.4 | |
| VCBackbone=Gemma-3-4B-IT2025.09 | 0.3 | 0.299 | 70.2 | |
| SCBackbone=Gemma-3-4B-IT2025.09 | 0.303 | 0.304 | 71.3 | |
| K-VCBackbone=Qwen3-8B2025.09 | 0.341 | 0.348 | 60.4 | |
| SC-VCBackbone=Gemma-3-4B-IT2025.09 | 0.362 | 0.378 | 65.3 | |
| VCBackbone=Qwen3-1.7B2025.09 | 0.387 | 0.383 | 72 | |
| SC-VCBackbone=Qwen3-1.7B2025.09 | 0.451 | 0.474 | 55.9 | |
| K-VCBackbone=Qwen3-1.7B2025.09 | 0.538 | 0.524 | 59.6 |