Skip to content

flatironinstitute/cryoSBI

Repository files navigation

cryo_em_SBI

tests
Testing Status

Collaborators

David Silva-Sánchez, Lars Dingeldein, Roberto Covino, Pilar Cossio

Dependencies

  1. Lampe
  2. SciPy
  3. NumPy
  4. PyTorch
  5. json

Installing

Download this repository

https://github.com/DSilva27/cryo_em_SBI.git

Install the module

::
python3 -m pip install .

Using the code

Generating data from command line

python3 -m cryo_sbi.generate_training_set --config_file experiments/benchmark_hsp90/image_params_snr01_128.json --num_train_samples 10 --num_val_samples 10 --file_name "test1" --n_workers 24

Train posterior from command line

python3 -m cryo_sbi.inference.NPE_train_without_saving \
    --image_config_file experiments/benchmark_hsp90/image_params_training.json \
    --train_config_file experiments/benchmark_hsp90/resnet18_encoder.json\
    --epochs 100 \
    --estimator_file experiments/benchmark_hsp90/posterior_hsp90.estimator \
    --loss_file experiments/benchmark_hsp90/posterior_hsp90.loss