commands.train_nn

Train a neural net to model a 3D density map given 2D images with pose assignments.

Example usage

$ cryodrgn train_nn projections.mrcs –poses angles.pkl –ctf ctf.pkl -o output/train_nn -n 10

# Run with more epochs $ cryodrgn train_nn projections.star –poses angles.pkl –ctf.pkl -o outs/003_train-nn –num-epochs 30 –lr 0.01

# Restart after already running the same command with some epochs completed $ cryodrgn train_nn projections.star –poses angles.pkl –ctf.pkl -o outs/003_train-nn –num-epochs 75 –lr 0.01 –load latest

Functions

add_args(parser)
get_latest(args)
main(args)
save_checkpoint(model, lattice, optim, ...)
save_config(args, dataset, lattice, model, ...)
train(model, lattice, optim, y, rot[, ...])