GenericLinearSchedulerWithLinearWarmup

class lightning_ir.schedulers.schedulers.GenericLinearSchedulerWithLinearWarmup(num_warmup_steps: int, num_training_steps: int, *args, final_value: float = 0.0, num_delay_steps: int = 0, **kwargs)[source]

Bases: LinearSchedulerWithLinearWarmup, GenericScheduler

__init__(num_warmup_steps: int, num_training_steps: int, *args, final_value: float = 0.0, num_delay_steps: int = 0, **kwargs) → None

Scheduler for linearly decreasing values with linear warmup.

Parameters:

  • num_warmup_steps (int) – Number of warmup steps

  • num_training_steps (int) – Number of training steps

  • final_value (float, optional) – The final value that should be reached at the end of decay, defaults to 0.0

  • num_delay_steps (int, optional) – Number of steps to delay warmup / decay, defaults to 0

Methods

Attributes

on_train_batch_start(trainer: Trainer, pl_module: LightningModule, batch: Any, batch_idx: int) → None

Called when the train batch begins.

on_train_end(trainer: Trainer, pl_module: LightningModule) → None

Called when the train ends.

on_train_start(trainer: Trainer, pl_module: LightningIRModule) → None

Called when the train begins.

value_lambda(current_step: int) → float

Lambda function for linearly decreasing values with linear warmup.

Parameters:

current_step (int) – Current step

Returns:

Value at the current step

Return type:

float