List your company
ResearchFormerly WizWand

Generalization gap prediction on QNLI Case 9

0.18Gap Prediction ErrorInconsistency
0.1480.3640.580.796May 31, 2023
Updated 8mo ago

Evaluation results

MethodLinks
InconsistencyTraining scenario=Fine-tuning, Data source=Unlabeled data2023.050.18
1-sharpnessTraining scenario=Fine-tuning, Data source=Test loss landscape (dev data, unfair advantage)2023.050.37
HessianTraining scenario=Fine-tuning, Data source=Test loss landscape (dev data, unfair advantage)2023.050.81
HessianTraining scenario=Fine-tuning, Data source=Training loss landscape2023.050.87
1-sharpnessTraining scenario=Fine-tuning, Data source=Training loss landscape2023.050.98