Natural Language Understanding on ARC-c
65.8AccuracymPLUG-Owl2
Evaluation Results
| Method | Links | |
|---|---|---|
| mPLUG-Owl2Shot setting=0-shot2023.11 | 65.8 | |
| Vicuna-v1.5Shot setting=0-shot2023.11 | 56.6 | |
| LLaMA-2-ChatShot setting=0-shot2023.11 | 54.9 | |
| WizardLMShot setting=0-shot2023.11 | 47.5 | |
| SpaceByte + SPSequence Reduction=6.3, FLOPs/byte Reduction=2.72026.05 | 41.9 | |
| Fixed (p = 4) + SPSequence Reduction=4.0, FLOPs/byte Reduction=2.12026.05 | 41.6 | |
| H-Net + SPSequence Reduction=5.3†, FLOPs/byte Reduction=2.42026.05 | 40.9 | |
| Fixed (p = 16) + SPSequence Reduction=16.0, FLOPs/byte Reduction=2.92026.05 | 40.8 | |
| Tokenizer-basedSequence Reduction=3.7, FLOPs/byte Reduction=4.12026.05 | 40.7 | |
| Entropy-basedSequence Reduction=6.4, FLOPs/byte Reduction=4.02026.05 | 40.6 | |
| Entropy-based + SPSequence Reduction=6.3†, FLOPs/byte Reduction=2.72026.05 | 40.4 | |
| LLaMA-2Shot setting=0-shot2023.11 | 40.3 | |
| H-NetSequence Reduction=4.8, FLOPs/byte Reduction=3.42026.05 | 40.1 | |
| Fixed (p = 4)Sequence Reduction=4.0, FLOPs/byte Reduction=3.12026.05 | 39.8 | |
| Fixed (p = 8) + SPSequence Reduction=8.0, FLOPs/byte Reduction=2.62026.05 | 39.8 | |
| Fixed (p = 8)Sequence Reduction=8.0, FLOPs/byte Reduction=4.52026.05 | 39.6 | |
| Byte-levelSequence Reduction=1.0, FLOPs/byte Reduction=1.02026.05 | 38.7 | |
| SpaceByteSequence Reduction=6.3, FLOPs/byte Reduction=4.02026.05 | 38.6 | |
| Fixed (p = 16)Sequence Reduction=16.0, FLOPs/byte Reduction=5.72026.05 | 35.2 | |
| ProDiaLProjection Type=Out-Proj, Number of Parameters=0.90M2024.11 | 30.8 | |
| FTProjection Type=In-Proj, Number of Parameters=56.62M2024.11 | 29.61 | |
| ProDiaLProjection Type=In-Proj, Number of Parameters=1.25M2024.11 | 29.61 | |
| FTProjection Type=Both-Proj, Number of Parameters=84.94M2024.11 | 29.52 | |
| LoRAProjection Type=Both-Proj, Number of Parameters=2.36M2024.11 | 29.52 | |
| FTProjection Type=Out-Proj, Number of Parameters=28.31M2024.11 | 29.27 | |
| Full-FTProjection Type=Baseline, Number of Parameters=130.00M2024.11 | 28.84 | |
| ProDiaLProjection Type=Both-Proj, Number of Parameters=2.38M2024.11 | 28.84 | |
| DoRAProjection Type=Both-Proj, Number of Parameters=2.45M2024.11 | 28.75 | |
| StrongProjection Type=Baseline, Number of Parameters=3.80M2024.11 | 28.67 | |
| DoRAProjection Type=Out-Proj, Number of Parameters=0.90M2024.11 | 28.67 | |
| LoRAProjection Type=Out-Proj, Number of Parameters=0.89M2024.11 | 28.54 | |
| DoRAProjection Type=In-Proj, Number of Parameters=1.55M2024.11 | 28.5 | |
| LoRAProjection Type=In-Proj, Number of Parameters=1.48M2024.11 | 28.41 | |
| BitFitProjection Type=Baseline, Number of Parameters=0.07M2024.11 | 26.88 |