pytorch-image-models/timm/optim/lamb.py

193 lines
9.0 KiB
Python
Raw Normal View History

2021-08-10 04:13:43 +08:00
""" PyTorch Lamb optimizer w/ behaviour similar to NVIDIA FusedLamb
This optimizer code was adapted from the following (starting with latest)
* https://github.com/HabanaAI/Model-References/blob/2b435114fe8e31f159b1d3063b8280ae37af7423/PyTorch/nlp/bert/pretraining/lamb.py
* https://github.com/NVIDIA/DeepLearningExamples/blob/master/PyTorch/LanguageModeling/Transformer-XL/pytorch/lamb.py
* https://github.com/cybertronai/pytorch-lamb
Use FusedLamb if you can (GPU). The reason for including this variant of Lamb is to have a version that is
similar in behaviour to APEX FusedLamb if you aren't using NVIDIA GPUs or cannot install/use APEX.
In addition to some cleanup, this Lamb impl has been modified to support PyTorch XLA and has been tested on TPU.
2021-08-10 04:13:43 +08:00
Original copyrights for above sources are below.
Modifications Copyright 2021 Ross Wightman
2021-08-10 04:13:43 +08:00
"""
# Copyright (c) 2021, Habana Labs Ltd. All rights reserved.
# Copyright (c) 2019-2020, NVIDIA CORPORATION. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# MIT License
#
# Copyright (c) 2019 cybertronai
#
# 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.
import math
2021-08-10 04:13:43 +08:00
import torch
from torch.optim import Optimizer
class Lamb(Optimizer):
2021-08-10 04:13:43 +08:00
"""Implements a pure pytorch variant of FuseLAMB (NvLamb variant) optimizer from apex.optimizers.FusedLAMB
reference: https://github.com/NVIDIA/DeepLearningExamples/blob/master/PyTorch/LanguageModeling/Transformer-XL/pytorch/lamb.py
LAMB was proposed in `Large Batch Optimization for Deep Learning: Training BERT in 76 minutes`_.
Arguments:
params (iterable): iterable of parameters to optimize or dicts defining parameter groups.
2021-08-10 04:13:43 +08:00
lr (float, optional): learning rate. (default: 1e-3)
betas (Tuple[float, float], optional): coefficients used for computing
running averages of gradient and its norm. (default: (0.9, 0.999))
eps (float, optional): term added to the denominator to improve
numerical stability. (default: 1e-8)
weight_decay (float, optional): weight decay (L2 penalty) (default: 0)
grad_averaging (bool, optional): whether apply (1-beta2) to grad when
calculating running averages of gradient. (default: True)
max_grad_norm (float, optional): value used to clip global grad norm (default: 1.0)
2021-08-19 13:20:35 +08:00
trust_clip (bool): enable LAMBC trust ratio clipping (default: False)
always_adapt (boolean, optional): Apply adaptive learning rate to 0.0
2021-08-10 04:13:43 +08:00
weight decay parameter (default: False)
.. _Large Batch Optimization for Deep Learning - Training BERT in 76 minutes:
https://arxiv.org/abs/1904.00962
.. _On the Convergence of Adam and Beyond:
https://openreview.net/forum?id=ryQu7f-RZ
"""
def __init__(
2021-08-18 12:48:26 +08:00
self, params, lr=1e-3, bias_correction=True, betas=(0.9, 0.999), eps=1e-6,
2021-08-19 13:20:35 +08:00
weight_decay=0.01, grad_averaging=True, max_grad_norm=1.0, trust_clip=False, always_adapt=False):
defaults = dict(
2021-08-18 12:48:26 +08:00
lr=lr, bias_correction=bias_correction, betas=betas, eps=eps, weight_decay=weight_decay,
2021-08-19 13:20:35 +08:00
grad_averaging=grad_averaging, max_grad_norm=max_grad_norm,
trust_clip=trust_clip, always_adapt=always_adapt)
2021-08-10 04:13:43 +08:00
super().__init__(params, defaults)
@torch.no_grad()
2021-08-10 04:13:43 +08:00
def step(self, closure=None):
"""Performs a single optimization step.
Arguments:
closure (callable, optional): A closure that reevaluates the model
and returns the loss.
"""
loss = None
if closure is not None:
with torch.enable_grad():
loss = closure()
2021-08-10 04:13:43 +08:00
2021-08-19 13:20:35 +08:00
device = self.param_groups[0]['params'][0].device
one_tensor = torch.tensor(1.0, device=device) # because torch.where doesn't handle scalars correctly
2021-08-10 04:13:43 +08:00
global_grad_norm = torch.zeros(1, device=device)
for group in self.param_groups:
for p in group['params']:
if p.grad is None:
continue
grad = p.grad
2021-08-10 04:13:43 +08:00
if grad.is_sparse:
raise RuntimeError('Lamb does not support sparse gradients, consider SparseAdam instad.')
global_grad_norm.add_(grad.pow(2).sum())
global_grad_norm = torch.sqrt(global_grad_norm)
# FIXME it'd be nice to remove explicit tensor conversion of scalars when torch.where promotes
# scalar types properly https://github.com/pytorch/pytorch/issues/9190
max_grad_norm = torch.tensor(self.defaults['max_grad_norm'], device=device)
clip_global_grad_norm = torch.where(
global_grad_norm > max_grad_norm,
global_grad_norm / max_grad_norm,
one_tensor)
2021-08-10 04:13:43 +08:00
for group in self.param_groups:
bias_correction = 1 if group['bias_correction'] else 0
beta1, beta2 = group['betas']
grad_averaging = 1 if group['grad_averaging'] else 0
beta3 = 1 - beta1 if grad_averaging else 1.0
2021-08-10 04:13:43 +08:00
# assume same step across group now to simplify things
# per parameter step can be easily support by making it tensor, or pass list into kernel
if 'step' in group:
group['step'] += 1
else:
group['step'] = 1
if bias_correction:
bias_correction1 = 1 - beta1 ** group['step']
bias_correction2 = 1 - beta2 ** group['step']
else:
bias_correction1, bias_correction2 = 1.0, 1.0
for p in group['params']:
if p.grad is None:
continue
grad = p.grad.div_(clip_global_grad_norm)
2021-08-10 04:13:43 +08:00
state = self.state[p]
# State initialization
if len(state) == 0:
# Exponential moving average of gradient valuesa
state['exp_avg'] = torch.zeros_like(p)
2021-08-10 04:13:43 +08:00
# Exponential moving average of squared gradient values
state['exp_avg_sq'] = torch.zeros_like(p)
2021-08-10 04:13:43 +08:00
exp_avg, exp_avg_sq = state['exp_avg'], state['exp_avg_sq']
2021-08-10 04:13:43 +08:00
# Decay the first and second moment running average coefficient
exp_avg.mul_(beta1).add_(grad, alpha=beta3) # m_t
exp_avg_sq.mul_(beta2).addcmul_(grad, grad, value=1 - beta2) # v_t
denom = (exp_avg_sq.sqrt() / math.sqrt(bias_correction2)).add_(group['eps'])
update = (exp_avg / bias_correction1).div_(denom)
2021-08-18 12:48:26 +08:00
weight_decay = group['weight_decay']
if weight_decay != 0:
update.add_(p, alpha=weight_decay)
2021-08-19 13:20:35 +08:00
if weight_decay != 0 or group['always_adapt']:
# Layer-wise LR adaptation. By default, skip adaptation on parameters that are
# excluded from weight decay, unless always_adapt == True, then always enabled.
w_norm = p.norm(2.0)
g_norm = update.norm(2.0)
# FIXME nested where required since logical and/or not working in PT XLA
trust_ratio = torch.where(
w_norm > 0,
torch.where(g_norm > 0, w_norm / g_norm, one_tensor),
one_tensor,
)
2021-08-19 13:20:35 +08:00
if group['trust_clip']:
# LAMBC trust clipping, upper bound fixed at one
trust_ratio = torch.minimum(trust_ratio, one_tensor)
update.mul_(trust_ratio)
p.add_(update, alpha=-group['lr'])
2021-08-10 04:13:43 +08:00
return loss