List your company

Natural Language Inference on GLUE and SNLI v1 (test val)

91.22MNLI AccuracyθA fine-tune
36.640850.810464.9879.1496May 27, 2026

Evaluation results

MethodLinks
θA fine-tuneN (Calibration Samples)=All, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0591.2295.3284.8391.0591.2990.74—
θB fine-tuneN (Calibration Samples)=All, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0585.2893.5977.2589.5788.5186.84—
BiCoN (Calibration Samples)=200, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0571.0685.7468.9588.2566.5976.115.01
THESEUSN (Calibration Samples)=200, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0569.9283.6368.2388.0964.3174.833.73
θAN (Calibration Samples)=200, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0562.4981.1865.786.2365.9472.311.21
BiCoN (Calibration Samples)=100, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0561.9479.3367.584.3158.4370.38
THESEUSN (Calibration Samples)=100, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0560.9378.3466.7863.1956.3865.122.82
θBN (Calibration Samples)=200, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0560.8382.4563.5485.5263.1871.10
θAN (Calibration Samples)=100, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0560.1378.4659.9368.1258.3264.992.69
θBN (Calibration Samples)=100, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0558.4875.4361.3762.4453.7862.30
BiCoN (Calibration Samples)=50, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0546.3276.7167.8783.460.867.0212.32
THESEUSN (Calibration Samples)=50, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0546.1272.6566.4263.1854.6160.595.89
θAN (Calibration Samples)=50, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0539.3475.5858.1268.6253.2158.974.27
θBN (Calibration Samples)=50, Source Model=T5-3B, Target Model=T5-Large, Protocol=linear probing2026.0538.7469.557.462.5645.2854.70