Skip to content

Latest commit

 

History

History
21 lines (17 loc) · 1.12 KB

CHANGELOG.md

File metadata and controls

21 lines (17 loc) · 1.12 KB

v0.1.0

Initial Release (v0.1.0)

Features

  • Added preprocessing script (preprocess.py) to convert CSV or EDF files to NumPy .npy files with various preprocessing steps including notch filtering and bandpass filtering.
  • Implemented parallel processing for preprocessing using the --parallel flag.
  • Added support for TUH EEG files in preprocessing.
  • Included detailed README with instructions for using preprocess.py and example usage for Crown CSV files and TUH EEG files.
  • Integrated wandb for experiment tracking.
  • Added CSVLogCallback class in src/trainer/make.py for logging training and evaluation metrics to CSV files.
  • Provided train_parallel.sh script for distributed training using PyTorch with multiple GPUs.
  • Included dependencies in requirements.txt and requirements-dev.txt for easy setup.

Documentation

  • Comprehensive README with setup instructions, preprocessing details, and example usage.
  • Links to external resources for additional help with tools like tmux.

References

  • Based on NeuroGPT by Wenhui Cui et al.
  • Inspired by Neurosity Foundational Model by Jeremy Nixon and AJ Keller.