Autoregressive rollout prediction on TheWell-ASM (steps 21-60, test)
0.0268VRMSEWaveLiT-9.5M
Evaluation Results
| Method | Links | |
|---|---|---|
| WaveLiT-9.5MTraining Protocol=Rollout-finetuned2026.05 | 0.0268 | |
| WaveLiTVariant=bespoke2026.05 | 0.0268 | |
| WaveLiT-9.5MTraining Protocol=Pretrained2026.05 | 0.0406 | |
| DPOT-HModel=DPOT-H2026.05 | 0.0543 | |
| WalrusModel=Walrus2026.05 | 0.056 | |
| WalrusVariant=external2026.05 | 0.056 | |
| WaveLiT-1.2MTraining Protocol=Rollout-finetuned2026.05 | 0.0693 | |
| WaveLiT (Fnd-FT)Variant=multi-task + FT2026.05 | 0.0765 | |
| WaveLiT (Fnd-PT)Variant=multi-task2026.05 | 0.105 | |
| WaveLiT-1.2MTraining Protocol=Pretrained2026.05 | 0.1166 | |
| MPP-AViT-LModel=MPP-AViT-L2026.05 | 0.139 | |
| Poseidon-LModel=Poseidon-L2026.05 | 0.2429 | |
| WalrusModel=Walrus2026.05 | 0.6441 | |
| MPP-AViT-LModel=MPP-AViT-L2026.05 | 0.7468 | |
| Poseidon-LModel=Poseidon-L2026.05 | 0.9586 | |
| WaveLiT-9.5MTraining Protocol=Pretrained2026.05 | 1.169 | |
| DPOT-HModel=DPOT-H2026.05 | 1.2664 | |
| WaveLiT-9.5MTraining Protocol=Rollout-finetuned2026.05 | 1.4803 | |
| WaveLiT-1.2MTraining Protocol=Rollout-finetuned2026.05 | 1.908 | |
| WaveLiT-1.2MTraining Protocol=Pretrained2026.05 | 2.7116 |