models/ contains two types of model architectures tested and corresponding hyper-param space searched in this study.
my_callbacks.py: defined callbacks used in this part.
Preprocessing.py: some helper functions
randomopt.py: the main script for the hyper-param optimization with a random search approach
python randomopt.py \
--trainfile ../data/ogt_for_hyperopt_original_distribution.fasta \
--modelname Model_ResNetN3.py \
--iterations 1000 \
--patience 100 \
--outdir ../results/opt_hyperparams/ResNetN3_OriDist/
python randomopt.py \
--trainfile ../data/ogt_for_hyperopt_uniform_distribution.fasta \
--modelname Model_ResNetN3.py \
--iterations 1000 \
--outdir ../results/opt_hyperparams/ResNetN3_UniDist/
python randomopt.py \
--trainfile ../data/ogt_for_hyperopt_original_distribution.fasta \
--modelname Model_ResNetRed.py \
--iterations 1000 \
--patience 100 \
--outdir ../results/opt_hyperparams/ResNetRed_OriDist/
python randomopt.py \
--trainfile ../data/ogt_for_hyperopt_uniform_distribution.fasta \
--modelname Model_ResNetRed.py \
--iterations 1000 \
--outdir ../results/opt_hyperparams/ResNetRed_UniDist/
visualization.ipynb