Control parameters for dma

Usage

dma_control(
  crossfit_folds = 10L,
  zprime_folds = 1L,
  noise_dim = 5L,
  hidden_dim = 100L,
  num_layer = 3L,
  num_epochs = 500L,
  lr = 0.001,
  riesz_epochs = 100L,
  riesz_lr = 0.01,
  batch_size = 64L,
  riesz_batch_size = 64L,
  device = "cpu"
)

Arguments

crossfit_folds: [numeric(1)]
Number of crossfit folds.
zprime_folds: [numeric(1)]
Number of folds for calculating Z'.
noise_dim: [numeric(1)]
Engression noise dimension.
hidden_dim: [numeric(1)]
NN hidden layer dimension.
num_layer: [numeric(1)]
Number of NN layers for engression.
num_epochs: [numeric(1)]
Training epochs for outcome regressions.
lr: [numeric(1)]
Learning rate for outcome regressions.
riesz_epochs: [numeric(1)]
Training epochs for Riesz representer.
riesz_lr: [numeric(1)]
Learning rate for Riesz representer.
batch_size: [numeric(1)]
Batch size for outcome regressions.
riesz_batch_size: [numeric(1)]
Batch size for Riesz representer.
device: [character(1)]
Torch device: "cpu", "cuda", or "mps".

Value

A list of control parameters.