Files
pyannote-audio/pyannote/audio/cli/lr_schedulers/CosineAnnealingWarmRestarts.py
T

67 lines
2.1 KiB
Python

# MIT License
#
# Copyright (c) 2022 CNRS
#
# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.
from typing import Optional
from torch.optim import Optimizer
from torch.optim.lr_scheduler import (
CosineAnnealingWarmRestarts as _CosineAnnealingWarmRestarts,
)
def CosineAnnealingWarmRestarts(
optimizer: Optimizer,
min_lr: float = 1e-8,
max_lr: float = 1e-3,
patience: int = 1,
num_batches_per_epoch: Optional[int] = None,
**kwargs,
):
"""Wrapper around CosineAnnealingWarmRestarts
Parameters
----------
optimizer : Optimizer
Optimizer
min_lr : float, optional
Defaults to 1e-8.
max_lr : float, optional
Defaults to 1e-3
patience : int, optional
Number of epochs per cycle. Defaults to 1.
num_batches_per_epoch : int, optional
Number of batches per epoch.
"""
# initialize optimizer lr to max_lr
for g in optimizer.param_groups:
g["lr"] = max_lr
num_steps = patience * num_batches_per_epoch
return {
"scheduler": _CosineAnnealingWarmRestarts(
optimizer, num_steps, eta_min=min_lr, T_mult=2
),
"interval": "step",
}