Few-shot classification on Synthetic Classification Few-shot (P_bursty=0.5, alpha=0)
90.2ICL ScoreCoQE
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CoQEE=512, L=4, #Param=2.0M2026.03 | 90.2 | 91 | |
| CoQEE=256, L=4, #Param=1.0M2026.03 | 88 | 82 | |
| Transformer + RegularizationE=64, L=12, #Param=0.9M2026.03 | 67.8 | 36.4 | |
| TransformerE=64, L=12, #Param=0.9M2026.03 | 64.5 | 17.8 | |
| Transformer + NoiseE=64, L=12, #Param=0.9M2026.03 | 59.8 | 8.9 | |
| TransformerE=64, L=8, #Param=0.6M2026.03 | 57.5 | 21.4 | |
| Transformer + Temporary ForgettingE=64, L=12, #Param=0.9M2026.03 | 50.5 | 7.2 | |
| TransformerE=64, L=4, #Param=0.4M2026.03 | 50.2 | 27.3 | |
| TransformerE=256, L=12, #Param=1.4M2026.03 | 50.2 | 58.1 | |
| TransformerE=512, L=8, #Param=2.2M2026.03 | 50.2 | 98.9 | |
| TransformerE=512, L=12, #Param=2.4M2026.03 | 50.1 | 98.4 | |
| Linear TransformerE=64, L=12, #Param=0.8M2026.03 | 50.1 | 21 | |
| TransformerE=256, L=8, #Param=1.2M2026.03 | 49.9 | 60.2 | |
| TransformerE=256, L=4, #Param=1.0M2026.03 | 49.9 | 63.4 | |
| TransformerE=512, L=4, #Param=2.0M2026.03 | 49.5 | 99.4 |