List your company

Confidence Calibration on MultiNLI Mismatch (test)

0.0071ECEMIR
0.0033280.0287890.054250.079711Jun 7, 2023

Evaluation results

MethodLinks
MIRBackbone=RoBERTa [24], Calibration Protocol=Base2023.060.00710.01070.00260.0258
IRBackbone=RoBERTa [24], Calibration Protocol=Base2023.060.00880.01070.00360.0246
MIR + PROCALBackbone=RoBERTa [24], Calibration Protocol=Ours2023.060.01030.01360.0040.0235
HB + PROCALBackbone=RoBERTa [24], Calibration Protocol=Ours2023.060.01050.0150.00250.0251
HBBackbone=RoBERTa [24], Calibration Protocol=Base2023.060.01110.01170.00360.0376
TS + PROCALBackbone=RoBERTa [24], Calibration Protocol=Ours2023.060.01220.01350.0040.0278
ETS + PROCALBackbone=RoBERTa [24], Calibration Protocol=Ours2023.060.01240.01260.00390.0277
IR + PROCALBackbone=RoBERTa [24], Calibration Protocol=Ours2023.060.01430.01230.00340.0261
Conf + PROCALBackbone=RoBERTa [24], Calibration Protocol=Ours2023.060.01450.01460.00370.0278
ETSBackbone=RoBERTa [24], Calibration Protocol=Base2023.060.01580.01560.00660.0304
TSBackbone=RoBERTa [24], Calibration Protocol=Base2023.060.0170.01760.00410.0303
PTS + PROCALBackbone=RoBERTa [24], Calibration Protocol=Ours2023.060.02420.02460.00930.0356
ConfBackbone=RoBERTa [24], Calibration Protocol=Base2023.060.02470.02620.00850.0354
PTSK + PROCALBackbone=RoBERTa [24], Calibration Protocol=Ours2023.060.04260.0440.02830.0515
PTSBackbone=RoBERTa [24], Calibration Protocol=Base2023.060.07530.0750.04210.0778
PTSKBackbone=RoBERTa [24], Calibration Protocol=Base2023.060.10140.10140.0740.1033