Rollout Prediction on Euler (test)
0.028CRPSOurs (CRPS)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Ours (CRPS)Backbone=Lola, Training Mode=CRPS retrofitted2026.03 | 0.028 | 0.25 | 1.103 | |
| Ours (CRPS)Backbone=Walrus, Training Mode=CRPS retrofitted2026.03 | 0.037 | 0.273 | 0.909 | |
| Deterministic (Base)Backbone=Lola, Training Mode=Deterministic baseline2026.03 | 0.045 | 0.284 | — | |
| Deterministic (FT)Backbone=Lola, Training Mode=Fine-tuned with pre-training loss2026.03 | 0.045 | 0.285 | — | |
| Deterministic (Base)Backbone=Walrus, Training Mode=Deterministic baseline2026.03 | 0.079 | 0.382 | — | |
| Deterministic (FT)Backbone=Walrus, Training Mode=Fine-tuned with pre-training loss2026.03 | 0.082 | 0.389 | — | |
| Ours (CRPS)Backbone=Poseidon, Training Mode=CRPS retrofitted2026.03 | 0.088 | 0.499 | 1.117 | |
| Deterministic (Base)Backbone=Poseidon, Training Mode=Deterministic baseline2026.03 | 0.131 | 0.558 | — | |
| Deterministic (FT)Backbone=Poseidon, Training Mode=Fine-tuned with pre-training loss2026.03 | 0.131 | 0.532 | — |