Skip to contents

The counterpart of fit_density_estimator(). Arguments are forwarded explicitly rather than through ... and formals(), because there are only two targets and they take disjoint arguments: the neural classifiers take the whole training block, the closed-form one takes none of it.

Usage

fit_ratio_estimator(
  classifier,
  theta_z,
  x_z,
  hidden,
  n_blocks,
  num_atoms,
  embedding,
  max_epochs,
  batch_size,
  lr,
  validation_fraction,
  patience,
  n_restarts,
  clip_grad_norm,
  seed,
  verbose,
  device
)