136 lines
4.8 KiB
Python
136 lines
4.8 KiB
Python
# Copyright The PyTorch Lightning team.
|
|
#
|
|
# 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.
|
|
import math
|
|
from typing import List, Tuple, Union
|
|
|
|
import torch
|
|
from torch.optim.optimizer import Optimizer
|
|
|
|
from pytorch_lightning.core.lightning import LightningModule
|
|
from pytorch_lightning.plugins.precision_plugin import PrecisionPlugin
|
|
from pytorch_lightning.utilities.distributed import rank_zero_warn
|
|
from pytorch_lightning.utilities import AMPType
|
|
|
|
try:
|
|
from apex import amp
|
|
except ImportError:
|
|
amp = None
|
|
|
|
|
|
class ApexPlugin(PrecisionPlugin):
|
|
|
|
def __init__(self, trainer=None):
|
|
self.trainer = trainer
|
|
|
|
def connect(self, model, optimizers):
|
|
model, optimizers = self.configure_apex(amp, model, optimizers, self.trainer.amp_level)
|
|
self.trainer.reinit_scheduler_properties(optimizers, self.trainer.lr_schedulers)
|
|
return model, optimizers
|
|
|
|
def training_step(self, fx, args):
|
|
output = fx(args)
|
|
return output
|
|
|
|
def backward(self, closure_loss, optimizer, opt_idx, *args, **kwargs):
|
|
closure_loss = amp.scale_loss(closure_loss, optimizer)
|
|
|
|
# enter apex context
|
|
self.trainer.dev_debugger.track_event('AMP', str(AMPType.APEX))
|
|
context = closure_loss
|
|
closure_loss = closure_loss.__enter__()
|
|
|
|
# do backward pass
|
|
if self.trainer.train_loop.automatic_optimization:
|
|
model = self.trainer.get_model()
|
|
model.backward(closure_loss, optimizer, opt_idx)
|
|
else:
|
|
closure_loss.backward(*args, **kwargs)
|
|
|
|
# exit amp context
|
|
a, b, c = None, None, None
|
|
error = context.__exit__(a, b, c)
|
|
if error:
|
|
rank_zero_warn(a, b, c)
|
|
raise Exception('apex unscale error')
|
|
|
|
# once backward has been applied, release graph
|
|
closure_loss = closure_loss.detach()
|
|
return closure_loss
|
|
|
|
def configure_apex(
|
|
self,
|
|
amp: object,
|
|
model: LightningModule,
|
|
optimizers: List[Optimizer],
|
|
amp_level: str,
|
|
) -> Tuple[LightningModule, List[Optimizer]]:
|
|
r"""
|
|
Override to init AMP your own way.
|
|
Must return a model and list of optimizers.
|
|
|
|
Args:
|
|
amp: pointer to amp library object.
|
|
model: pointer to current :class:`LightningModule`.
|
|
optimizers: list of optimizers passed in :meth:`configure_optimizers`.
|
|
amp_level: AMP mode chosen ('O1', 'O2', etc...)
|
|
|
|
Return:
|
|
Apex wrapped model and optimizers
|
|
|
|
Examples:
|
|
.. code-block:: python
|
|
|
|
# Default implementation used by Trainer.
|
|
def configure_apex(self, amp, model, optimizers, amp_level):
|
|
model, optimizers = amp.initialize(
|
|
model, optimizers, opt_level=amp_level,
|
|
)
|
|
|
|
return model, optimizers
|
|
"""
|
|
model, optimizers = amp.initialize(model, optimizers, opt_level=amp_level)
|
|
return model, optimizers
|
|
|
|
def clip_gradients(self, grad_clip_val: Union[int, float], optimizer: Optimizer, norm_type: float):
|
|
"""
|
|
This code is a modification of :meth:`torch.nn.utils.clip_grad_norm_` using a higher epsilon for fp16 weights.
|
|
This is important when setting amp_level to O2, and the master weights are in fp16.
|
|
Args:
|
|
grad_clip_val: Maximum norm of gradients.
|
|
optimizer: Optimizer with gradients that will be clipped.
|
|
norm_type: (float or int): type of the used p-norm. Can be ``'inf'`` for
|
|
infinity norm.
|
|
"""
|
|
model = self.trainer.get_model()
|
|
parameters = model.parameters()
|
|
max_norm = float(grad_clip_val)
|
|
|
|
if isinstance(parameters, torch.Tensor):
|
|
parameters = [parameters]
|
|
parameters = [p for p in parameters if p.grad is not None]
|
|
|
|
if len(parameters) == 0:
|
|
return torch.tensor(0.)
|
|
device = parameters[0].grad.device
|
|
total_norm = torch.norm(
|
|
torch.stack([torch.norm(p.grad.detach(), norm_type).to(device) for p in parameters]), norm_type)
|
|
clip_coef = max_norm / (total_norm + self.norm_clipping_epsilon)
|
|
if clip_coef < 1:
|
|
for p in parameters:
|
|
p.grad.detach().mul_(clip_coef.to(p.grad.device))
|
|
|
|
@property
|
|
def norm_clipping_epsilon(self):
|
|
return 1e-5
|