Time-series Forecasting on Systems (test)
0.14Per-system Difference in KL DivergenceDynaMix – Panda
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DynaMix – PandaPrediction Horizon (L_pred)=1024, KL implementation=Geometric Misalignment2025.05 | 0.14 | — | — | — | — | — | |
| DynaMix – PandaPrediction Horizon (L_pred)=512, KL implementation=Geometric Misalignment2025.05 | 0.23 | — | — | — | — | — | |
| Chronos 20M SFT – PandaPrediction Horizon (L_pred)=512, KL implementation=Geometric Misalignment2025.05 | 0.3 | — | — | — | — | — | |
| Chronos 200M – Chronos 20M SFTPrediction Horizon (L_pred)=2048, KL implementation=Geometric Misalignment2025.05 | 0.38 | — | — | — | — | — | |
| Chronos 200M – Chronos 20M SFTPrediction Horizon (L_pred)=3584, KL implementation=Geometric Misalignment2025.05 | 0.38 | — | — | — | — | — | |
| Chronos 200M – Chronos 20M SFTPrediction Horizon (L_pred)=1024, KL implementation=Geometric Misalignment2025.05 | 0.39 | — | — | — | — | — | |
| Chronos 200M – Chronos 20M SFTPrediction Horizon (L_pred)=3072, KL implementation=Geometric Misalignment2025.05 | 0.39 | — | — | — | — | — | |
| Chronos 200M – Chronos 20M SFTPrediction Horizon (L_pred)=512, KL implementation=Geometric Misalignment2025.05 | 0.44 | — | — | — | — | — | |
| Chronos 20M SFT – PandaPrediction Horizon (L_pred)=1024, KL implementation=Geometric Misalignment2025.05 | 0.49 | — | — | — | — | — | |
| DynaMix – PandaPrediction Horizon (L_pred)=2048, KL implementation=Geometric Misalignment2025.05 | 0.64 | — | — | — | — | — | |
| Chronos 20M SFT – PandaPrediction Horizon (L_pred)=2048, KL implementation=Geometric Misalignment2025.05 | 0.79 | — | — | — | — | — | |
| DynaMix – PandaPrediction Horizon (L_pred)=3072, KL implementation=Geometric Misalignment2025.05 | 0.96 | — | — | — | — | — | |
| Chronos 20M SFT – PandaPrediction Horizon (L_pred)=3072, KL implementation=Geometric Misalignment2025.05 | 1 | — | — | — | — | — | |
| DynaMix – PandaPrediction Horizon (L_pred)=3584, KL implementation=Geometric Misalignment2025.05 | 1.07 | — | — | — | — | — | |
| Chronos 20M SFT – PandaPrediction Horizon (L_pred)=3584, KL implementation=Geometric Misalignment2025.05 | 1.09 | — | — | — | — | — | |
| Chronos 20M – Chronos 20M SFTPrediction Horizon (L_pred)=512, KL implementation=Geometric Misalignment2025.05 | 1.82 | — | — | — | — | — | |
| Chronos 20M – Chronos 20M SFTPrediction Horizon (L_pred)=1024, KL implementation=Geometric Misalignment2025.05 | 1.86 | — | — | — | — | — | |
| Chronos 20M – Chronos 20M SFTPrediction Horizon (L_pred)=2048, KL implementation=Geometric Misalignment2025.05 | 1.94 | — | — | — | — | — | |
| Chronos 20M – Chronos 20M SFTPrediction Horizon (L_pred)=3072, KL implementation=Geometric Misalignment2025.05 | 1.99 | — | — | — | — | — | |
| Chronos 20M – Chronos 20M SFTPrediction Horizon (L_pred)=3584, KL implementation=Geometric Misalignment2025.05 | 2.03 | — | — | — | — | — | |
| Chronos 200M – Chronos 20M SFTContext windows=5, Metric Implementation=GMM-based KL divergence2025.05 | — | 0.39 | 0.41 | 0.43 | 0.44 | 0.42 | |
| Chronos 20M – Chronos 20M SFTContext windows=5, Metric Implementation=GMM-based KL divergence2025.05 | — | 1.27 | 1.09 | 0.89 | 0.84 | 0.89 | |
| Chronos 20M SFT – PandaContext windows=5, Metric Implementation=GMM-based KL divergence2025.05 | — | 0.79 | 0.36 | -0.01 | -0.22 | -0.33 | |
| DynaMix – PandaContext windows=5, Metric Implementation=GMM-based KL divergence2025.05 | — | 0.81 | 0.18 | -0.42 | -0.75 | -0.88 |