Upload train_full.py with huggingface_hub

Upload uncertainty.py with huggingface_hub

Upload model.py with huggingface_hub

Upload data_pipeline.py with huggingface_hub

Training: val_loss=3.5843

Training: val_loss=3.8311

Update train_cpu.py - fix heteroscedastic loss clamping

Update uncertainty.py - fix heteroscedastic loss clamping

Update model.py - fix heteroscedastic loss clamping

Add ARCHITECTURE.md

Add train_full.py

Add uncertainty.py

Add train.py

Add model.py

Add data_pipeline.py

Add README with architecture overview

initial commit