Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,15 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_0_best.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_0_final.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_1_best.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_1_final.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_2_best.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_2_final.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_3_best.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_3_final.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_4_best.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_4_final.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_5_best.keras filter=lfs diff=lfs merge=lfs -text
+track_prediction_ensemble_model_5_final.keras filter=lfs diff=lfs merge=lfs -text

track_prediction_ensemble_model_0_best.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b1c81cefb79e6e31dd6e3c30a976736e56ba11a8eaeaec5475dfeb83d05e15f0
+size 761521

track_prediction_ensemble_model_0_final.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b054e487bfc09fd121bcf5df5f588807f49ff731b6b04571499df41bea5c43ad
+size 761521

track_prediction_ensemble_model_1_best.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3dba932503dd9231087e6e810b1a3206561aa6680fd9c4815594d0a5718e540
+size 1981033

track_prediction_ensemble_model_1_final.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:43f1d14ff8613391ec29485e20935bf3b7b30234d70bf65e3f50e6052fd1ada1
+size 1981034

track_prediction_ensemble_model_2_best.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5d37a3eb7669bacfda464826494c8fd78d6952652337dc2b263892b9647ccd58
+size 289867

track_prediction_ensemble_model_2_final.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:942da85d8ee131ba5443d90640b84d51afd83629ebc78c8dc4e07eb8b73acaaa
+size 289869

track_prediction_ensemble_model_3_best.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07387e4c74e932c499549f8fe3ba5d7e1019aaf1fddcedab5582abe63f085605
+size 289898

track_prediction_ensemble_model_3_final.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5562331bb575645f0c3c1994cfdbb44a8648dae4803dba17effe9309dee0976a
+size 289898

track_prediction_ensemble_model_4_best.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:479d3c6727c0c16c8ab9762fae3fffd6871131bd4d4fb3f5d0dff0b19351b777
+size 289908

track_prediction_ensemble_model_4_final.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7d7be80e3d0e00ed28875ee4b21dfb3f36f4047b43752749314cf4e8fe40555
+size 289907

track_prediction_ensemble_model_5_best.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77e213baa613dc9b16387dfd54ec9a31a7fdb925a3fcb238d426faa3082e0e3b
+size 289910

track_prediction_ensemble_model_5_final.keras

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ff988aff13661d299e694d11e7d3dcba0118b4de3542995d72bfc3f2e6cf1d7
+size 289910

training_report.md

@@ -0,0 +1,65 @@
+# Training Report - Ensemble
+
+    Generated: 2025-09-06 13:15:26
+
+    ## Overview
+    - **Command**: `ensemble`
+    - **Training Duration**: 1651.54 seconds (27.5 minutes)
+    - **Output Directory**: `output/ensemble_20250906_124755`
+
+    ## Dataset Information
+    - **Total Records**: 24,832
+    - **Training Steps per Epoch**: 310
+    - **Validation Steps per Epoch**: 77
+
+    ### Vocabulary Sizes
+    - **Stations**: 6 unique stations
+    - **Routes**: 13 unique routes  
+    - **Tracks**: 13 unique tracks (prediction targets)
+
+    ## Training Configuration
+    - **Num Models**: 6
+- **Epochs**: 150
+- **Batch Size**: 64
+- **Base Learning Rate**: 0.001
+- **Dataset Size**: 24832
+
+
+## Final Performance Metrics
+- **Average Validation Loss**: 1.2251
+- **Average Validation Accuracy**: 0.5957
+- **Best Individual Accuracy**: 0.6049
+- **Worst Individual Accuracy**: 0.5812
+- **Ensemble Std Accuracy**: 0.0087
+
+## Additional Information
+- **Individual Model Metrics**: {'model_index': 0, 'validation_loss': 1.2142382860183716, 'validation_accuracy': 0.5909090638160706, 'learning_rate': 0.000896730883132793, 'parameters': 53384}, {'model_index': 1, 'validation_loss': 1.2308698892593384, 'validation_accuracy': 0.6049107313156128, 'learning_rate': 0.0011035837511110408, 'parameters': 156552}, {'model_index': 2, 'validation_loss': 1.2147506475448608, 'validation_accuracy': 0.6022727489471436, 'learning_rate': 0.0011871558720145028, 'parameters': 14856}, {'model_index': 3, 'validation_loss': 1.2401705980300903, 'validation_accuracy': 0.5911120176315308, 'learning_rate': 0.0008334328623039442, 'parameters': 14856}, {'model_index': 4, 'validation_loss': 1.2382102012634277, 'validation_accuracy': 0.5811688303947449, 'learning_rate': 0.0009118210339513598, 'parameters': 14856}, {'model_index': 5, 'validation_loss': 1.2121498584747314, 'validation_accuracy': 0.6038960814476013, 'learning_rate': 0.001173992747183349, 'parameters': 14856}
+- **Ensemble Strategy**: Diverse architectures (deep, wide, standard)
+- **Learning Rate Variation**: 0.8x to 1.2x base rate with random variation
+- **Total Parameters**: 269360
+
+
+    ## Dataset Schema
+    The model was trained on MBTA track assignment data with the following features:
+    - **Categorical Features**: station_id, route_id, direction_id
+    - **Temporal Features**: hour, minute, day_of_week (cyclically encoded)
+    - **Target**: track_number (classification with 13 classes)
+
+    ## Model Architecture
+    - Embedding layers for categorical features
+    - Cyclical time encoding (sin/cos) for temporal patterns
+    - Dense layers with dropout regularization
+    - Softmax output for multi-class track prediction
+
+    ## Usage
+    To load and use this model:
+
+    ```python
+    import keras
+    # Load for inference (optimizer not saved):
+    model = keras.models.load_model('track_prediction_ensemble_final.keras', compile=False)
+    ```
+
+    ---
+    *Report generated by imt-ml training pipeline*
+