Skip to main content
Version: 24.8.0

TrainingSettings​

CLASS - TrainingSettings(

  • stats_steps: int =  20
  • random_seed: Optional[int] =  int(time.time())
  • is_training_from_scratch: bool =  False

)

This Dataclass represents the Training Settings for the Compression.

Class Variables​

  • stats_steps (int) - Number of steps to run and collect statistics of the Model. Default is 20. Should not be set to more than 50 for good performance. 20 is more than enough.
  • random_seed (Optional[int]) - Setting a Random Seed.
  • is_training_from_scratch (bool) - Whether or not the model is training from scratch. It is not recommended at this point to use the SDK for training from scratch large models.