Sentiment Classification on Sent140 (Generalization Gap Metric)
23.8Generalization GapIRM
Evaluation Results
| Method | Links | |
|---|---|---|
| IRMprotocol=leave-one-out zero-shot transfer, aggregation=mean over 8 seeds2026.05 | 23.8 | |
| BERTprotocol=leave-one-out zero-shot transfer, aggregation=averaged across all source-target pairs2026.05 | 24 | |
| AGMprotocol=leave-one-out zero-shot transfer, aggregation=mean over 8 seeds2026.05 | 24.4 | |
| Mask-onlyprotocol=leave-one-out zero-shot transfer, aggregation=mean over 8 seeds2026.05 | 24.4 | |
| Fishprotocol=leave-one-out zero-shot transfer, aggregation=mean over 8 seeds2026.05 | 24.7 | |
| DROprotocol=leave-one-out zero-shot transfer, aggregation=mean over 8 seeds2026.05 | 24.8 | |
| DANNprotocol=leave-one-out zero-shot transfer, aggregation=mean over 8 seeds2026.05 | 26.4 | |
| RoBERTaprotocol=leave-one-out zero-shot transfer, aggregation=averaged across all source-target pairs2026.05 | 27.1 |