Learning rate to use for gradient descent training.
end_learning_rate
End learning rate for linear decay. Defaults to 0.
batch_size
Batch size for training. Defaults to 48.
epochs
Number of training iterations over the dataset. Defaults to 2.
optimizer
Optimizer to use for training. Supported values are defined in
BertOptimizer enum: ADAMW and LAMB.
weight_decay
Weight decay of the optimizer. Defaults to 0.01.
desired_precisions
If specified, adds a RecallAtPrecision metric per
desired_precisions[i] entry which tracks the recall given the constraint
on precision. Only supported for binary classification.
desired_recalls
If specified, adds a PrecisionAtRecall metric per
desired_recalls[i] entry which tracks the precision given the constraint
on recall. Only supported for binary classification.
gamma
Gamma parameter for focal loss. To use cross entropy loss, set this
value to 0. Defaults to 2.0.
tokenizer
Tokenizer to use for preprocessing. Must be one of the enum
options of SupportedBertTokenizers. Defaults to FULL_TOKENIZER.
checkpoint_frequency
Frequency(in epochs) of saving checkpoints during
training. Defaults to 0 which does not save training checkpoints.
[[["Easy to understand","easyToUnderstand","thumb-up"],["Solved my problem","solvedMyProblem","thumb-up"],["Other","otherUp","thumb-up"]],[["Missing the information I need","missingTheInformationINeed","thumb-down"],["Too complicated / too many steps","tooComplicatedTooManySteps","thumb-down"],["Out of date","outOfDate","thumb-down"],["Samples / code issue","samplesCodeIssue","thumb-down"],["Other","otherDown","thumb-down"]],["Last updated 2024-05-07 UTC."],[],[]]