commands.abinit_het

Train a heterogeneous NN reconstruction model with hierarchical pose optimization.

Example usage

# the default is to train for thirty epochs; here we train for fifty instead $ cryodrgn abinit_het particles.mrcs -o cryodrgn-outs/003_abinit_het –zdim 4 –ctf ctf.pkl -n 50

# using .star particle input requires datadir argument pointing to image stacks $ cryodrgn abinit_het particles.star –datadir path_to_images/ -o cryodrgn-outs/004_abinit_het.10 –zdim 10 –ctf ctf.pkl -n 50

Functions

add_args(parser)
eval_z(model, lattice, data, batch_size, device)
get_latest(args)
main(args)
make_model(args, lattice, enc_mask, in_dim)
pretrain(model, lattice, optim, minibatch, ...)
save_checkpoint(model, lattice, optim, ...)	Save model weights, latent encoding z, and decoder volumes
save_config(args, dataset, lattice, model, ...)
sort_poses(poses)
train(model, lattice, ps, optim, L, ...[, ...])