Time Series Forecasting on GIFT-Eval 97 tasks
0.692MASEAME-TS Ultra
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| AME-TS UltraScale=Large, #Params=540M† 133M2026.05 | 0.692 | 0.857 | 0.687 | 0.68 | |
| Chronos-2Scale=Base, #Params=120M2026.05 | 0.698 | 0.886 | 0.704 | 0.679 | |
| AME-TS LargeScale=Large, #Params=162M† 75M2026.05 | 0.7 | 0.852 | 0.69 | 0.68 | |
| TimesFM-2.5Scale=Large, #Params=231M2026.05 | 0.705 | 0.856 | 0.694 | 0.687 | |
| Moirai2Scale=Large, #Params=300M2026.05 | 0.728 | 0.925 | 0.739 | 0.705 | |
| Sundial-128MScale=Base, #Params=128M2026.05 | 0.75 | 0.945 | 0.739 | 0.739 | |
| AME-TS BaseScale=Base, #Params=62M† 29M2026.05 | 0.765 | 0.926 | 0.737 | 0.725 | |
| AME-TS SmallScale=Small, #Params=30M† 7M2026.05 | 0.812 | 0.99 | 0.767 | 0.741 | |
| AME-TSScale=Per-dataset2026.05 | 0.844 | 0.946 | 0.832 | 0.8 | |
| PatchTSTScale=Per-dataset2026.05 | 0.849 | 1.02 | 0.839 | 0.81 | |
| AME-TS TinyScale=Small, #Params=14M† 6M2026.05 | 0.856 | 1.043 | 0.808 | 0.834 | |
| Chronos-SmallScale=Small, #Params=46M2026.05 | 0.892 | 1.059 | 0.866 | 0.847 | |
| iTransformerScale=Per-dataset2026.05 | 0.893 | 1.074 | 0.858 | 0.85 | |
| Moirai-SmallScale=Small, #Params=14M2026.05 | 0.946 | 1.109 | 1.056 | 0.906 |