Implementing Contrastive Self-Supervised Learning with Radiation Augmentations, SimCLR, PyTorch Lightning, and Hyperparameter Optimization#52

Draft

stompsjo wants to merge 57 commits intomainfrom

Commits on Oct 31, 2022

adding hyperopt functions
Jordan Stomps
committed
add supervised logistic regression model function
Jordan Stomps
committed
adding cotraining model function
Jordan Stomps
committed
adding code for Label Prop model function
Jordan Stomps
committed
adding shadow fully connected NN model function
Jordan Stomps
committed
adding shadow eaat cnn function model
Jordan Stomps
committed
abstracting MINOS to Spectra
Jordan Stomps
committed
removing duplicate device in eaat-cnn
Jordan Stomps
committed
revamping design of ssl models, starting with logreg
Jordan Stomps
committed
adding save function to logreg class and renaming hyperopt.py

Jordan Stomps
authored and
Jordan Stomps
committed
commenting logistic regression class and methods

Jordan Stomps
authored and
Jordan Stomps
committed
scripts/utils.py pep8 changes

Jordan Stomps
authored and
Jordan Stomps
committed
implementing LabelProp with hyperopt functionality

Jordan Stomps
authored and
Jordan Stomps
committed
implementing co-training with hyperopt functionality

Jordan Stomps
authored and
Jordan Stomps
committed
implementing Shadow fully-connected NN with hyperopt

Jordan Stomps
authored and
Jordan Stomps
committed
implementing Shadow EAAT CNN with hyperopt

Jordan Stomps
authored and
Jordan Stomps
committed
adding functions for pca analysis

Jordan Stomps
authored and
Jordan Stomps
committed
rearranging model files

Jordan Stomps
authored and
Jordan Stomps
committed
adding unit test for LogReg

Jordan Stomps
authored and
Jordan Stomps
committed
updating dependencies

Jordan Stomps
authored and
Jordan Stomps
committed
correcting pytorch package name

Jordan Stomps
authored and
Jordan Stomps
committed
adding unit test for CoTraining

Jordan Stomps
authored and
Jordan Stomps
committed
adding unit test for LabelProp

Jordan Stomps
authored and
Jordan Stomps
committed
adding unit test for ShadowNN

Jordan Stomps
authored and
Jordan Stomps
committed
including utils scripts in unit tests coverage

Jordan Stomps
authored and
Jordan Stomps
committed
error: training NNs takes too long for a unit test, let alone hyperopt

Jordan Stomps
authored and
Jordan Stomps
committed
error: these cnns are so bad that they can't even make predictions
Jordan Stomps
committed
correcting cnn parameter calculation to include max_pool1d
Jordan Stomps
committed
adding tests for more coverage
Jordan Stomps
committed
adding a test for util plots
Jordan Stomps
committed
adding seed test to co-training
Jordan Stomps
committed
removing old commented line

Jordan Stomps
authored and
Jordan Stomps
committed
changing fresh_start methods of models to use class train method instead

Jordan Stomps
authored and
Jordan Stomps
committed
adding an EarlyStopper class for managing that functionality
Jordan Stomps
committed
adding cross validation implementation
Jordan Stomps
committed

Commits on Nov 1, 2022

investigating ray.tune for better hyperparameter optimization
Jordan Stomps
committed

Commits on Nov 2, 2022

refactoring hyperopt->raytune; todo: update test_models.py
Jordan Stomps
committed

Commits on Nov 3, 2022

fixing errors in unit tests for hyperopt->raytune
Jordan Stomps
committed

Commits on Aug 8, 2023

unifying .gitignore
Jordan Stomps
committed
parent be771462d0188b9f98fdf470907270929662b958
Jordan Stomps
committed
rearranging folders for relative importation
Jordan Stomps
committed

Commits on Aug 10, 2023

functional implementation with extra args and unfinished checkpointing
u9f
committed

Commits on Aug 18, 2023

attempting to debug parallelized ray tune
u9f
committed

Commits on Aug 19, 2023

HyperOpt working in serial
u9f
committed

Commits on Aug 21, 2023

Commits on Aug 23, 2023

adding functionality for storing and restoring pre-existing trials
u9f
committed

Commits on Aug 29, 2023

correcting for output of SSLHyperOpt.py
Jordan Stomps
committed
adding AdamW parameters to dry run
Jordan Stomps
committed

Commits on Sep 5, 2023

adding projection head hyperparameter optimization script
Jordan Stomps
committed
chtc bugfixes
Jordan Stomps
committed
removing extranneous -p
Jordan Stomps
committed
adjusting other hyperparameter inputs
Jordan Stomps
committed

Commits on Dec 21, 2023

correcting import statements
u9f
committed
cleaning up some commented code
Jordan Stomps
committed
removing hardcoded license
Jordan Stomps
committed