Inheritance diagram for fastreid.engine.hooks.SWA:

Public Member Functions
	__init__ (self, int swa_start, int swa_freq, float swa_lr_factor, float eta_min, lr_sched=False)

	before_step (self)

	after_step (self)

Public Member Functions inherited from fastreid.engine.train_loop.HookBase
	before_train (self)

	after_train (self)

Public Attributes
	swa_start

	swa_freq

	swa_lr_factor

	eta_min

	lr_sched

	scheduler

Detailed Description

Definition at line 472 of file hooks.py.

Constructor & Destructor Documentation

◆ init()

fastreid.engine.hooks.SWA.__init__	(		self,
		int	swa_start,
		int	swa_freq,
		float	swa_lr_factor,
		float	eta_min,
			lr_sched = False )

Definition at line 473 of file hooks.py.

    def __init__(self, swa_start: int, swa_freq: int, swa_lr_factor: float, eta_min: float, lr_sched=False, ):
        self.swa_start = swa_start
        self.swa_freq = swa_freq
        self.swa_lr_factor = swa_lr_factor
        self.eta_min = eta_min
        self.lr_sched = lr_sched
 

Member Function Documentation

◆ after_step()

fastreid.engine.hooks.SWA.after_step ( self )

Called after each iteration.

Reimplemented from fastreid.engine.train_loop.HookBase.

Definition at line 494 of file hooks.py.

    def after_step(self):
        next_iter = self.trainer.iter + 1
 
        # Use Cyclic learning rate scheduler
        if next_iter > self.swa_start and self.lr_sched:
            self.scheduler.step()
 
        is_final = next_iter == self.trainer.max_iter
        if is_final:
            self.trainer.optimizer.swap_swa_param()

◆ before_step()

fastreid.engine.hooks.SWA.before_step ( self )

Called before each iteration.

Reimplemented from fastreid.engine.train_loop.HookBase.

Definition at line 480 of file hooks.py.

    def before_step(self):
        is_swa = self.trainer.iter == self.swa_start
        if is_swa:
            # Wrapper optimizer with SWA
            self.trainer.optimizer = optim.SWA(self.trainer.optimizer, self.swa_freq, self.swa_lr_factor)
            self.trainer.optimizer.reset_lr_to_swa()
 
            if self.lr_sched:
                self.scheduler = torch.optim.lr_scheduler.CosineAnnealingWarmRestarts(
                    optimizer=self.trainer.optimizer,
                    T_0=self.swa_freq,
                    eta_min=self.eta_min,
                )
 

Member Data Documentation

◆ eta_min

fastreid.engine.hooks.SWA.eta_min

Definition at line 477 of file hooks.py.

◆ lr_sched

fastreid.engine.hooks.SWA.lr_sched

Definition at line 478 of file hooks.py.

◆ scheduler

fastreid.engine.hooks.SWA.scheduler

Definition at line 488 of file hooks.py.

◆ swa_freq

fastreid.engine.hooks.SWA.swa_freq

Definition at line 475 of file hooks.py.

◆ swa_lr_factor

fastreid.engine.hooks.SWA.swa_lr_factor

Definition at line 476 of file hooks.py.

◆ swa_start

fastreid.engine.hooks.SWA.swa_start

Definition at line 474 of file hooks.py.

The documentation for this class was generated from the following file:

smreid/fastreid/engine/hooks.py

Public Member Functions

Public Attributes