lr_scheduler.py

# -*- encoding: utf-8 -*-
# ----------------------------------------------
# filename        :lr_scheduler.py
# description     :NomMer: Nominate Synergistic Context in Vision Transformer for Visual Recognition
# date            :2021/12/28 17:43:21
# author          :clark
# version number  :1.0
# ----------------------------------------------


import torch
from timm.scheduler.cosine_lr import CosineLRScheduler
from timm.scheduler.step_lr import StepLRScheduler
from timm.scheduler.scheduler import Scheduler


def build_scheduler(config, optimizer, n_iter_per_epoch):
    """[build lr scheduler]

    Args:
        config ([]): []
        optimizer ([torch.optim]): []
        n_iter_per_epoch ([int]): []

    Returns:
        [timm.scheduler]: []
    """
    num_steps = int(config.TRAIN.EPOCHS * n_iter_per_epoch)
    warmup_steps = int(config.TRAIN.WARMUP_EPOCHS * n_iter_per_epoch)
    decay_steps = int(config.TRAIN.LR_SCHEDULER.DECAY_EPOCHS * n_iter_per_epoch)

    lr_scheduler = None
    if config.TRAIN.LR_SCHEDULER.NAME == 'cosine':
        lr_scheduler = CosineLRScheduler(
            optimizer,
            t_initial=num_steps,
            t_mul=1.0,
            lr_min=config.TRAIN.MIN_LR,
            warmup_lr_init=config.TRAIN.WARMUP_LR,
            warmup_t=warmup_steps,
            cycle_limit=1,
            t_in_epochs=False,
        )
    elif config.TRAIN.LR_SCHEDULER.NAME == 'linear':
        lr_scheduler = LinearLRScheduler(
            optimizer,
            t_initial=num_steps,
            lr_min_rate=0.01,
            warmup_lr_init=config.TRAIN.WARMUP_LR,
            warmup_t=warmup_steps,
            t_in_epochs=False,
        )
    elif config.TRAIN.LR_SCHEDULER.NAME == 'step':
        lr_scheduler = StepLRScheduler(
            optimizer,
            decay_t=decay_steps,
            decay_rate=config.TRAIN.LR_SCHEDULER.DECAY_RATE,
            warmup_lr_init=config.TRAIN.WARMUP_LR,
            warmup_t=warmup_steps,
            t_in_epochs=False,
        )

    return lr_scheduler


class LinearLRScheduler(Scheduler):
    def __init__(
        self,
        optimizer: torch.optim.Optimizer,
        t_initial: int,
        lr_min_rate: float,
        warmup_t=0,
        warmup_lr_init=0.0,
        t_in_epochs=True,
        noise_range_t=None,
        noise_pct=0.67,
        noise_std=1.0,
        noise_seed=42,
        initialize=True,
    ) -> None:
        """[crate linear LR Scheduler]

        Args:
            optimizer (torch.optim.Optimizer): []
            t_initial (int): []
            lr_min_rate (float): []
            warmup_t (int, optional): []. Defaults to 0.
            warmup_lr_init ([type], optional): []. Defaults to 0..
            t_in_epochs (bool, optional): []. Defaults to True.
            noise_range_t ([type], optional): []. Defaults to None.
            noise_pct (float, optional): []. Defaults to 0.67.
            noise_std (float, optional): []. Defaults to 1.0.
            noise_seed (int, optional): []. Defaults to 42.
            initialize (bool, optional): []. Defaults to True.
        """
        super().__init__(
            optimizer,
            param_group_field="lr",
            noise_range_t=noise_range_t,
            noise_pct=noise_pct,
            noise_std=noise_std,
            noise_seed=noise_seed,
            initialize=initialize,
        )

        self.t_initial = t_initial
        self.lr_min_rate = lr_min_rate
        self.warmup_t = warmup_t
        self.warmup_lr_init = warmup_lr_init
        self.t_in_epochs = t_in_epochs
        if self.warmup_t:
            self.warmup_steps = [(v - warmup_lr_init) / self.warmup_t for v in self.base_values]
            super().update_groups(self.warmup_lr_init)
        else:
            self.warmup_steps = [1 for _ in self.base_values]

    def _get_lr(self, t):
        if t < self.warmup_t:
            lrs = [self.warmup_lr_init + t * s for s in self.warmup_steps]
        else:
            t = t - self.warmup_t
            total_t = self.t_initial - self.warmup_t
            lrs = [v - ((v - v * self.lr_min_rate) * min(1.0, t / total_t)) for v in self.base_values]
        return lrs

    def get_epoch_values(self, epoch: int):
        if self.t_in_epochs:
            return self._get_lr(epoch)
        else:
            return None

    def get_update_values(self, num_updates: int):
        if not self.t_in_epochs:
            return self._get_lr(num_updates)
        else:
            return None