qml.labs.phox.TrainingOptions

class TrainingOptions(unroll_steps=1, val_kwargs=None, convergence_interval=100, random_state=666, opt_jit=False)

Bases: object

Configuration options for training.

Parameters:

• unroll_steps (int) – How many optimization steps to run on the GPU before yielding control back to Python. Higher = Faster. Lower = More interactive/granular logging. Defaults to 1 (slow, good for debugging).

• val_kwargs (dict[str, Any] | None) – Arguments for the loss function to be used during validation.

• convergence_interval (int) – Number of steps over which to check for convergence. Defaults to 100.

• random_state (int) – Seed for PRNGKey.

• opt_jit (bool) – Whether to JIT the optimizer creation (usually False is fine).

Attributes

convergence_interval
opt_jit
random_state
unroll_steps
val_kwargs

convergence_interval = 100

opt_jit = False

random_state = 666

unroll_steps = 1

val_kwargs = None

code/api/api/pennylane.labs.phox.TrainingOptions

 

Download Python script

 

Download Notebook

 

View on GitHub