Calibration on BEAR (test)
0.083Brier Scoregemma-7b
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| gemma-7bEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.083 | 0.042 | |
| gemma-2bEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.098 | 0.032 | |
| gpt2-mediumEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.11 | 0.071 | |
| opt-6.7bEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.11 | 0.024 | |
| gemma-7bEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.114 | 0.081 | |
| gpt2-largeEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.116 | 0.07 | |
| gpt2Estimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.117 | 0.1 | |
| gpt2-xlEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.121 | 0.068 | |
| gemma-7bEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.124 | — | |
| opt-1.3bEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.126 | 0.056 | |
| opt-350mEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.127 | 0.098 | |
| opt-125mEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.13 | 0.116 | |
| gemma-2bEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.133 | 0.068 | |
| gemma-7bEstimation Method=C_Base2026.01 | 0.133 | 0.077 | |
| gemma-7bEstimation Method=C_Margin2026.01 | 0.134 | 0.046 | |
| gpt2-largeEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.139 | 0.101 | |
| gpt2Estimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.14 | 0.158 | |
| gpt2-mediumEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.14 | 0.116 | |
| gemma-7bEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.144 | — | |
| opt-6.7bEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.144 | 0.029 | |
| opt-125mEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.147 | 0.164 | |
| opt-350mEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.15 | 0.137 | |
| gpt2-xlEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.152 | 0.086 | |
| gemma-2bEstimation Method=C_Margin2026.01 | 0.155 | 0.053 | |
| gemma-2bEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.157 | — | |
| roberta-largeEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.158 | 0.144 | |
| gemma-2bEstimation Method=C_Base2026.01 | 0.162 | 0.113 | |
| roberta-baseEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.162 | 0.165 | |
| gpt2-largeEstimation Method=C_Margin2026.01 | 0.163 | 0.104 | |
| opt-1.3bEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.163 | 0.068 | |
| opt-6.7bEstimation Method=C_Margin2026.01 | 0.168 | 0.069 | |
| opt-350mEstimation Method=C_Margin2026.01 | 0.177 | 0.139 | |
| gpt2-mediumEstimation Method=C_Margin2026.01 | 0.179 | 0.121 | |
| gpt2Estimation Method=C_Margin2026.01 | 0.181 | 0.157 | |
| opt-125mEstimation Method=C_Margin2026.01 | 0.186 | 0.153 | |
| opt-1.3bEstimation Method=C_Margin2026.01 | 0.187 | 0.1 | |
| opt-6.7bEstimation Method=C_Base2026.01 | 0.187 | 0.154 | |
| roberta-largeEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.188 | 0.228 | |
| gemma-2bEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.19 | — | |
| opt-6.7bEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.191 | — | |
| gpt2-xlEstimation Method=C_Margin2026.01 | 0.193 | 0.128 | |
| bert-large-casedEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.196 | 0.198 | |
| roberta-baseEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.196 | 0.268 | |
| bert-base-casedEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.205 | 0.218 | |
| gpt2-largeEstimation Method=C_Base2026.01 | 0.206 | 0.236 | |
| opt-1.3bEstimation Method=C_Base2026.01 | 0.214 | 0.207 | |
| gpt2-mediumEstimation Method=C_Base2026.01 | 0.218 | 0.249 | |
| opt-350mEstimation Method=C_Base2026.01 | 0.224 | 0.265 | |
| xlm-roberta-largeEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.224 | 0.249 | |
| xlm-roberta-baseEstimation Method=C_Average, Aggregation Strategy=Min2026.01 | 0.225 | 0.274 | |
| gpt2-xlEstimation Method=C_Base2026.01 | 0.228 | 0.24 | |
| bert-large-casedEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.23 | 0.278 | |
| gpt2-mediumEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.23 | — | |
| opt-1.3bEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.232 | — | |
| opt-6.7bEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.232 | — | |
| gpt2-xlEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.235 | — | |
| gpt2Estimation Method=C_Base2026.01 | 0.237 | 0.295 | |
| gpt2-largeEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.238 | — | |
| opt-125mEstimation Method=C_Base2026.01 | 0.241 | 0.299 | |
| bert-base-casedEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.244 | 0.298 | |
| roberta-largeEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.254 | — | |
| gpt2Estimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.259 | — | |
| opt-350mEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.263 | — | |
| roberta-baseEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.265 | — | |
| xlm-roberta-largeEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.272 | 0.36 | |
| opt-125mEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.273 | — | |
| bert-base-casedEstimation Method=C_Margin2026.01 | 0.276 | 0.295 | |
| roberta-largeEstimation Method=C_Margin2026.01 | 0.278 | 0.291 | |
| xlm-roberta-baseEstimation Method=C_Average, Aggregation Strategy=Vote2026.01 | 0.283 | 0.395 | |
| gpt2-xlEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.284 | — | |
| bert-large-casedEstimation Method=C_Margin2026.01 | 0.285 | 0.302 | |
| gpt2-mediumEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.286 | — | |
| gpt2-largeEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.29 | — | |
| opt-1.3bEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.29 | — | |
| roberta-baseEstimation Method=C_Margin2026.01 | 0.291 | 0.321 | |
| roberta-largeEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.304 | — | |
| bert-large-casedEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.31 | — | |
| gpt2Estimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.315 | — | |
| opt-125mEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.32 | — | |
| opt-350mEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.32 | — | |
| roberta-baseEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.322 | — | |
| bert-base-casedEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.338 | — | |
| xlm-roberta-largeEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.354 | — | |
| xlm-roberta-largeEstimation Method=C_Margin2026.01 | 0.359 | 0.412 | |
| xlm-roberta-baseEstimation Method=C_Consistency, Aggregation Strategy=Min2026.01 | 0.362 | — | |
| bert-base-casedEstimation Method=C_Base2026.01 | 0.367 | 0.454 | |
| roberta-largeEstimation Method=C_Base2026.01 | 0.367 | 0.441 | |
| xlm-roberta-baseEstimation Method=C_Margin2026.01 | 0.368 | 0.427 | |
| bert-large-casedEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.369 | — | |
| bert-large-casedEstimation Method=C_Base2026.01 | 0.373 | 0.452 | |
| roberta-baseEstimation Method=C_Base2026.01 | 0.387 | 0.482 | |
| bert-base-casedEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.408 | — | |
| xlm-roberta-largeEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.43 | — | |
| xlm-roberta-baseEstimation Method=C_Consistency, Aggregation Strategy=Vote2026.01 | 0.448 | — | |
| xlm-roberta-largeEstimation Method=C_Base2026.01 | 0.461 | 0.559 | |
| xlm-roberta-baseEstimation Method=C_Base2026.01 | 0.481 | 0.585 |