Upload STGformer model trained on METR-LA

README.md

@@ -0,0 +1,74 @@
+---
+tags:
+- traffic-forecasting
+- time-series
+- graph-neural-network
+- transformer
+- stgformer
+datasets:
+- metr-la
+---
+
+# STGformer Model - METR-LA
+
+Spatio-Temporal Graph Transformer (STGformer) trained on METR-LA dataset for traffic speed forecasting.
+
+## Model Description
+
+This model uses a transformer-based graph neural network architecture that combines:
+- Self-attention mechanisms for capturing temporal dependencies
+- Spatial graph convolution for modeling spatial relationships
+- Adaptive embeddings for learning node-specific patterns
+- Time-of-day embeddings for capturing daily patterns
+
+## Evaluation Metrics
+
+- **Test MAE (15 min)**: 2.5637
+- **Test MAPE (15 min)**: 0.0654
+- **Test RMSE (15 min)**: 4.8755
+
+
+## Dataset
+
+**METR-LA**: Traffic speed data from highway sensors.
+
+## Usage
+
+```python
+from utils.stgformer import load_from_hub
+
+# Load model from Hub
+model, scaler = load_from_hub("METR-LA")
+
+# Get predictions
+import numpy as np
+x = np.random.randn(10, 12, 207, 2)  # (batch, seq_len, nodes, [value, tod])
+predictions = model.predict(x)
+```
+
+## Training
+
+Model was trained using the STGformer implementation with configuration:
+- Input features: 2 [speed, time-of-day]
+- Time-of-day embedding dimension: 24
+- Day-of-week embedding dimension: 0 (disabled)
+- Adaptive embedding dimension: 80
+- Number of attention heads: 4
+- Number of layers: 3
+
+## Citation
+
+If you use this model, please cite the STGformer paper:
+
+```bibtex
+@article{stgformer,
+  title={STGformer: Spatio-Temporal Graph Transformer for Traffic Forecasting},
+  author={Author names},
+  journal={Conference/Journal},
+  year={Year}
+}
+```
+
+## License
+
+This model checkpoint is released under the same license as the training code.

config.json

@@ -0,0 +1,32 @@
+{
+  "num_nodes": 207,
+  "in_steps": 12,
+  "out_steps": 12,
+  "input_dim": 2,
+  "output_dim": 1,
+  "steps_per_day": 288,
+  "input_embedding_dim": 24,
+  "tod_embedding_dim": 24,
+  "dow_embedding_dim": 0,
+  "adaptive_embedding_dim": 80,
+  "num_heads": 4,
+  "num_layers": 3,
+  "dropout": 0.1,
+  "dropout_a": 0.3,
+  "kernel_size": [
+    1
+  ],
+  "epochs": 100,
+  "batch_size": 64,
+  "learning_rate": 0.001,
+  "weight_decay": 0.0003,
+  "milestones": [
+    20,
+    30
+  ],
+  "lr_decay_rate": 0.1,
+  "early_stop": 10,
+  "clip_grad": 0,
+  "device": "cuda",
+  "verbose": 1
+}

hub_metadata.json

@@ -0,0 +1,11 @@
+{
+  "dataset": "METR-LA",
+  "upload_date": "2025-11-10T18:22:38.458773",
+  "metrics": {
+    "Test MAE (15 min)": 2.5637319087982178,
+    "Test MAPE (15 min)": 0.06541310995817184,
+    "Test RMSE (15 min)": 4.875480432556589
+  },
+  "framework": "PyTorch",
+  "model_type": "STGformer"
+}

metadata.json

@@ -0,0 +1,36 @@
+{
+  "config": {
+    "num_nodes": 207,
+    "in_steps": 12,
+    "out_steps": 12,
+    "input_dim": 2,
+    "output_dim": 1,
+    "steps_per_day": 288,
+    "input_embedding_dim": 24,
+    "tod_embedding_dim": 24,
+    "dow_embedding_dim": 0,
+    "adaptive_embedding_dim": 80,
+    "num_heads": 4,
+    "num_layers": 3,
+    "dropout": 0.1,
+    "dropout_a": 0.3,
+    "kernel_size": [
+      1
+    ],
+    "epochs": 100,
+    "batch_size": 64,
+    "learning_rate": 0.001,
+    "weight_decay": 0.0003,
+    "milestones": [
+      20,
+      30
+    ],
+    "lr_decay_rate": 0.1,
+    "early_stop": 10,
+    "clip_grad": 0,
+    "device": "cuda",
+    "verbose": 1
+  },
+  "scaler_mean": 54.40592575073242,
+  "scaler_std": 19.49374008178711
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d596a483045a7187645aef67a34283469b1d8cc0964f8218f54465e3c79dd052
+size 3530912