Spatio-temporal Prediction on Milan-Internet (10% train)
0.2537NRMSEST-Vision-LLM
Evaluation Results
| Method | Links | |
|---|---|---|
| ST-Vision-LLMInput sequence length (S)=12, Prediction horizon (K)=36, Few-shot training percentage=10%2025.10 | 0.2537 | |
| GCNGPTInput sequence length (S)=12, Prediction horizon (K)=36, Few-shot training percentage=10%2025.10 | 0.441 | |
| ST-LLMInput sequence length (S)=12, Prediction horizon (K)=36, Few-shot training percentage=10%2025.10 | 0.4555 | |
| GATGPTInput sequence length (S)=12, Prediction horizon (K)=36, Few-shot training percentage=10%2025.10 | 0.5154 | |
| Time-LLMInput sequence length (S)=12, Prediction horizon (K)=36, Few-shot training percentage=10%2025.10 | 0.5639 | |
| GWNETInput sequence length (S)=12, Prediction horizon (K)=36, Few-shot training percentage=10%2025.10 | 0.5881 | |
| STNInput sequence length (S)=12, Prediction horizon (K)=36, Few-shot training percentage=10%2025.10 | 0.6696 | |
| ResLSTMInput sequence length (S)=12, Prediction horizon (K)=36, Few-shot training percentage=10%2025.10 | 0.9958 |