package models.optim

Get desktop application:
View/edit binary Protocol Buffers messages

Used in: cnn.ResnetModelConfig

• optional string type = 1

  Optimizer type. Currently allowed values: "sgd", "momentum".

• optional float learn_rate = 2

  Learning rate.

• optional float momentum = 3

  Momentum.

• optional int32 batch_size = 4

  Batch size for training.

• optional int32 max_train_iter = 5

  Max number of training iterations.

• optional string learn_rate_scheduler_type = 6

  Name of the learning rate scheduler. Currently allowed values: "fixed".

• repeated int32 learn_rate_decay_steps = 7

  A list of training steps where learning rate decay happens.

• repeated float learn_rate_list = 8

  A list of learning rate values used by the fixed learning rate scheduler.