Changelog
All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
Unreleased
0.4.3 - 2024-08-29
Added
- Added a new model with higher accuracy
0.4.2 - 2024-08-27
Fixed
- Fixed cli arguments
- Fixed readme and documentation errors.
0.4.1 - 2024-08-23
Added
- Option to use time-warped augmented data during training
Fixed
- Padding and truncation algorithm is now fixed such that equal amounts of time samples are padded or truncated from both ends of the classifier examples
0.4.0 - 2024-08-19
Fixed
- Fixed bugs that caused slow training pipeline
- Fixed logic for uploading large dataset objects to Comet ML in small chunks
0.3.9 - 2024-08-16
Fixed
- Fixed a bug where all the class data files were not being used to make the dataset
0.3.8 - 2024-08-16
Fixed
- Fixed a bug where if the dataset dir had no dataset previously, new dataset was not being created.
- Increased CSV field size in train_etl to fix capfinder crashing during when encoutering large csv fields
Added
- Added logging info to standalone train ETL pipeline
0.3.7 - 2024-08-14
Fixed
- Fixed bugs in collate function that caused hogging of memory for large BAM file
0.3.6 - 2024-08-13
Fixed
- Fixed missing headers from some commands in the cli
Added
- Added option to specify custom models during inference
- Added more documentation
0.3.5 - 2024-08-11
Fixed
- Fixed string formatting issue
0.3.4 - 2024-08-11
Added
- Added ability to add more cap types to training
- Added a new train ETL pipeline that can handle larger than memory datasets
- Added improved interface for training pipeline
Fixed
- Old CLI app to reflect changes in API
0.3.3 - 2024-08-08
Fixed
- Issues with pip installation by removing the yanked package (types-pkg-resources)
0.3.2 - 2024-08-08
Fixed
- Issues with pip installation
0.3.1 - 2024-08-08
Fixed
- Issues with pip installation
0.3.0 - 2024-08-08
Fixed
- Issues with pip installation
0.2.9 - 2024-08-02
Fixed
- Fixed issue with batch inference not working
0.2.8 - 2024-08-02
Fixed
- Fixed loading the entire inference dataset in memory
Added
- Added more information to the README file
0.2.7 - 2024-08-01
Fixed
- Problems with cli not displaying capfinder version info
- Fixed issue with API docs generation
- Added max_examples parameter to ETL to limit the number of examples to process in a dataset (use during training)
0.2.6 - 2024-08-01
Fixed
- Slow report generation
- Problems with refreshing of cache
0.2.5 - 2024-07-31
Added
- Report generation
- Cli for inference
0.2.4 - 2024-07-26
Added
- Added functions for performaing inference
0.2.3 - 2024-07-17
Added
- Cosine annealing cyclic learning rate scheduler with resets, decay, and progressive lengthing of cycles
0.2.2 - 2024-07-15
Added
- Added Cyclical learning rate scheduler
Removed
- Removed reduce learning rate on plateau
0.2.1 - 2024-07-11
Added
- Added an attention-augmented CNN-LSTM model
0.2.0 - 2024-07-10
Fixed
- Fixed issue with ml_libs module not found
0.1.9 - 2024-07-10
Fixed
- Fixing issue with ml_libs module not found
0.1.8 - 2024-07-10
Added
- Added support for resnet model
0.1.7 - 2024-07-08
Fixed
- Fixed encoder model hogging all available GPU memory and crashing
0.1.6 - 2024-07-07
Fixed
- Fixed bug with not writing dataset version to the file
0.1.5 - 2024-07-07
Fixed
- Bugs with using the train config functionality
0.1.4 - 2024-07-07
Added
- Added pipeline for making training data
- Added CNN-LSTM and encoder models
0.1.3 - 2023-09-23
Added
- Functions to align OTE with reads in FASTQ
- Function for pretty printing alignment for debugging purposes
- Function for finding the start and end location of ROI in training datasets
0.1.2 - 2023-08-18
Changed
- Changed function arg names again for testing version bump
0.1.1 - 2023-08-16
Changed
- Changed function arg names for testing version bump
0.1.0 - 2023-08-15
Added
- Basic skeleton of the package and tested it