Upload temperature_config.json with huggingface_hub

Upload preprocessor_config.json with huggingface_hub

Upload model_benchmark.json with huggingface_hub

Upload label_mapping.json with huggingface_hub

Upload config.json with huggingface_hub

Upload class_confidence_stats.json with huggingface_hub

Upload calibration_data.json with huggingface_hub

Upload build_model.py with huggingface_hub

Delete best_phase2_swa.keras with huggingface_hub

Delete best_phase2_ema.keras with huggingface_hub

Delete best_phase2.keras with huggingface_hub

Delete best_phase1.keras with huggingface_hub

Upload greyscope-labs-architecture-classification-efficientnetv2.jpg

Delete results/greyscope-labs-architecture-classification-transferlearning-efficientnetv2.jpg

Delete results/greyscope-labs-architecture-classification-cnn-transferlearning.PNG

Upload greyscope-labs-architecture-classification-transferlearning-efficientnetv2.jpg

Update README.md

Upload greyscope-labs-architecture-classification-cnn-transferlearning.PNG

Delete results/greyscope-labs-arch-building-classification-hunggingface-cnn-densenet121.png

Update README.md

Upload V16 results/ — 12 visualization PNGs

Upload V16 tfjs_model/

Upload V16 tflite/

Upload V16 saved_model/

Upload V16 best_phase2_ema.keras

Upload V16 best_phase1.keras

Upload V16 best_phase2.keras

Upload V16 best_phase2_swa.keras

Upload V16 calibration_data.json

Upload V16 class_confidence_stats.json

Upload V16 confusion_pairs.json

Upload V16 model_benchmark.json

Upload V16 label_mapping.json

Upload V16 preprocessor_config.json

Upload V16 config.json

Upload V16 LICENSE

Upload V16 README.md

Remove V15 training_curves.png — replaced by V16 in results/

Remove V15 per_class_accuracy.png — replaced by V16 in results/

Remove V15 misclassification_examples.png — replaced by V16 in results/

Remove V15 confusion_matrix.png — replaced by V16 in results/

Remove V15 confidence_histogram.png — replaced by V16 in results/

Update preprocessor_config.json

Update config.json

Update model README

Enable all 5 images: training curves, confusion matrix, per-class accuracy, confidence histogram, misclassification examples

Add misclassification_examples.png

Add confidence_histogram.png

Add per_class_accuracy.png

Add confusion_matrix.png