Skip to content
Permalink
Branch: master
Find file Copy path
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
493 lines (358 sloc) 20.2 KB
import os
import numbers
import warnings
import torch
from ignite.contrib.handlers.base_logger import BaseLogger, BaseOptimizerParamsHandler, BaseOutputHandler, \
BaseWeightsScalarHandler, global_step_from_engine
__all__ = ['VisdomLogger', 'OptimizerParamsHandler', 'OutputHandler',
'WeightsScalarHandler', 'GradsScalarHandler', 'global_step_from_engine']
class _BaseVisDrawer(object):
def __init__(self, show_legend=False):
self.windows = {}
self.show_legend = show_legend
def add_scalar(self, logger, k, v, event_name, global_step):
"""
Helper method to log a scalar with VisdomLogger.
Args:
logger (VisdomLogger): visdom logger
k (str): scalar name which is used to set window title and y-axis label
v (int or float): scalar value, y-axis value
event_name: Event name which is used to setup x-axis label. Valid events are from
:class:`~ignite.engine.Events` or any `event_name` added by
:meth:`~ignite.engine.Engine.register_events`.
global_step (int): global step, x-axis value
"""
if k not in self.windows:
self.windows[k] = {
'win': None,
'opts': {
'title': k,
'xlabel': str(event_name),
'ylabel': k,
'showlegend': self.show_legend
}
}
update = None if self.windows[k]['win'] is None else 'append'
kwargs = {
"X": [global_step, ],
"Y": [v, ],
"env": logger.vis.env,
"win": self.windows[k]['win'],
"update": update,
"opts": self.windows[k]['opts'],
"name": k
}
future = logger.executor.submit(logger.vis.line, **kwargs)
if self.windows[k]['win'] is None:
self.windows[k]['win'] = future.result()
class OutputHandler(BaseOutputHandler, _BaseVisDrawer):
"""Helper handler to log engine's output and/or metrics
Examples:
.. code-block:: python
from ignite.contrib.handlers.visdom_logger import *
# Create a logger
vd_logger = VisdomLogger()
# Attach the logger to the evaluator on the validation dataset and log NLL, Accuracy metrics after
# each epoch. We setup `global_step_transform=global_step_from_engine(trainer)` to take the epoch of
# the `trainer`:
vd_logger.attach(evaluator,
log_handler=OutputHandler(tag="validation",
metric_names=["nll", "accuracy"],
global_step_transform=global_step_from_engine(trainer)),
event_name=Events.EPOCH_COMPLETED)
Example with CustomPeriodicEvent, where model is evaluated every 500 iterations:
.. code-block:: python
from ignite.contrib.handlers import CustomPeriodicEvent
cpe = CustomPeriodicEvent(n_iterations=500)
cpe.attach(trainer)
@trainer.on(cpe.Events.ITERATIONS_500_COMPLETED)
def evaluate(engine):
evaluator.run(validation_set, max_epochs=1)
from ignite.contrib.handlers.visdom_logger import *
vd_logger = VisdomLogger()
def global_step_transform(*args, **kwargs):
return trainer.state.iteration
# Attach the logger to the evaluator on the validation dataset and log NLL, Accuracy metrics after
# every 500 iterations. Since evaluator engine does not have CustomPeriodicEvent attached to it, we
# provide a global_step_transform to return the trainer.state.iteration for the global_step, each time
# evaluator metrics are plotted on Visdom.
vd_logger.attach(evaluator,
log_handler=OutputHandler(tag="validation",
metrics=["nll", "accuracy"],
global_step_transform=global_step_transform),
event_name=Events.EPOCH_COMPLETED)
Args:
tag (str): common title for all produced plots. For example, 'training'
metric_names (list of str, optional): list of metric names to plot or a string "all" to plot all available
metrics.
output_transform (callable, optional): output transform function to prepare `engine.state.output` as a number.
For example, `output_transform = lambda output: output`
This function can also return a dictionary, e.g `{'loss': loss1, `another_loss`: loss2}` to label the plot
with corresponding keys.
another_engine (Engine): Deprecated (see :attr:`global_step_transform`). Another engine to use to provide the
value of event. Typically, user can provide
the trainer if this handler is attached to an evaluator and thus it logs proper trainer's
epoch/iteration value.
global_step_transform (callable, optional): global step transform function to output a desired global step.
Input of the function is `(engine, event_name)`. Output of function should be an integer.
Default is None, global_step based on attached engine. If provided,
uses function output as global_step. To setup global step from another engine, please use
:meth:`~ignite.contrib.handlers.visdom_logger.global_step_from_engine`.
show_legend (bool, optional): flag to show legend in the window
Note:
Example of `global_step_transform`:
.. code-block:: python
def global_step_transform(engine, event_name):
return engine.state.get_event_attrib_value(event_name)
"""
def __init__(self, tag, metric_names=None, output_transform=None, another_engine=None, global_step_transform=None,
show_legend=False):
super(OutputHandler, self).__init__(tag, metric_names, output_transform, another_engine, global_step_transform)
_BaseVisDrawer.__init__(self, show_legend=show_legend)
def __call__(self, engine, logger, event_name):
if not isinstance(logger, VisdomLogger):
raise RuntimeError("Handler 'OutputHandler' works only with VisdomLogger")
metrics = self._setup_output_metrics(engine)
global_step = self.global_step_transform(engine, event_name)
if not isinstance(global_step, int):
raise TypeError("global_step must be int, got {}."
" Please check the output of global_step_transform.".format(type(global_step)))
for key, value in metrics.items():
values = []
keys = []
if isinstance(value, numbers.Number) or \
isinstance(value, torch.Tensor) and value.ndimension() == 0:
values.append(value)
keys.append(key)
elif isinstance(value, torch.Tensor) and value.ndimension() == 1:
values = value
keys = ["{}/{}".format(key, i) for i in range(len(value))]
else:
warnings.warn("VisdomLogger output_handler can not log "
"metrics value type {}".format(type(value)))
for k, v in zip(keys, values):
k = "{}/{}".format(self.tag, k)
self.add_scalar(logger, k, v, event_name, global_step)
logger._save()
class OptimizerParamsHandler(BaseOptimizerParamsHandler, _BaseVisDrawer):
"""Helper handler to log optimizer parameters
Examples:
.. code-block:: python
from ignite.contrib.handlers.visdom_logger import *
# Create a logger
vb_logger = VisdomLogger()
# Attach the logger to the trainer to log optimizer's parameters, e.g. learning rate at each iteration
vb_logger.attach(trainer,
log_handler=OptimizerParamsHandler(optimizer),
event_name=Events.ITERATION_STARTED)
Args:
optimizer (torch.optim.Optimizer): torch optimizer which parameters to log
param_name (str): parameter name
tag (str, optional): common title for all produced plots. For example, 'generator'
show_legend (bool, optional): flag to show legend in the window
"""
def __init__(self, optimizer, param_name="lr", tag=None, show_legend=False):
super(OptimizerParamsHandler, self).__init__(optimizer, param_name, tag)
_BaseVisDrawer.__init__(self, show_legend=show_legend)
def __call__(self, engine, logger, event_name):
if not isinstance(logger, VisdomLogger):
raise RuntimeError("Handler 'OptimizerParamsHandler' works only with VisdomLogger")
global_step = engine.state.get_event_attrib_value(event_name)
tag_prefix = "{}/".format(self.tag) if self.tag else ""
params = {"{}{}/group_{}".format(tag_prefix, self.param_name, i): float(param_group[self.param_name])
for i, param_group in enumerate(self.optimizer.param_groups)}
for k, v in params.items():
self.add_scalar(logger, k, v, event_name, global_step)
logger._save()
class WeightsScalarHandler(BaseWeightsScalarHandler, _BaseVisDrawer):
"""Helper handler to log model's weights as scalars.
Handler iterates over named parameters of the model, applies reduction function to each parameter
produce a scalar and then logs the scalar.
Examples:
.. code-block:: python
from ignite.contrib.handlers.visdom_logger import *
# Create a logger
vd_logger = VisdomLogger()
# Attach the logger to the trainer to log model's weights norm after each iteration
vd_logger.attach(trainer,
log_handler=WeightsScalarHandler(model, reduction=torch.norm),
event_name=Events.ITERATION_COMPLETED)
Args:
model (torch.nn.Module): model to log weights
reduction (callable): function to reduce parameters into scalar
tag (str, optional): common title for all produced plots. For example, 'generator'
show_legend (bool, optional): flag to show legend in the window
"""
def __init__(self, model, reduction=torch.norm, tag=None, show_legend=False):
super(WeightsScalarHandler, self).__init__(model, reduction, tag=tag)
_BaseVisDrawer.__init__(self, show_legend=show_legend)
def __call__(self, engine, logger, event_name):
if not isinstance(logger, VisdomLogger):
raise RuntimeError("Handler 'WeightsScalarHandler' works only with VisdomLogger")
global_step = engine.state.get_event_attrib_value(event_name)
tag_prefix = "{}/".format(self.tag) if self.tag else ""
for name, p in self.model.named_parameters():
name = name.replace('.', '/')
k = "{}weights_{}/{}".format(tag_prefix, self.reduction.__name__, name)
v = float(self.reduction(p.data))
self.add_scalar(logger, k, v, event_name, global_step)
logger._save()
class GradsScalarHandler(BaseWeightsScalarHandler, _BaseVisDrawer):
"""Helper handler to log model's gradients as scalars.
Handler iterates over the gradients of named parameters of the model, applies reduction function to each parameter
produce a scalar and then logs the scalar.
Examples:
.. code-block:: python
from ignite.contrib.handlers.visdom_logger import *
# Create a logger
vd_logger = VisdomLogger()
# Attach the logger to the trainer to log model's weights norm after each iteration
vd_logger.attach(trainer,
log_handler=GradsScalarHandler(model, reduction=torch.norm),
event_name=Events.ITERATION_COMPLETED)
Args:
model (torch.nn.Module): model to log weights
reduction (callable): function to reduce parameters into scalar
tag (str, optional): common title for all produced plots. For example, 'generator'
show_legend (bool, optional): flag to show legend in the window
"""
def __init__(self, model, reduction=torch.norm, tag=None, show_legend=False):
super(GradsScalarHandler, self).__init__(model, reduction, tag)
_BaseVisDrawer.__init__(self, show_legend=show_legend)
def __call__(self, engine, logger, event_name):
if not isinstance(logger, VisdomLogger):
raise RuntimeError("Handler 'GradsScalarHandler' works only with VisdomLogger")
global_step = engine.state.get_event_attrib_value(event_name)
tag_prefix = "{}/".format(self.tag) if self.tag else ""
for name, p in self.model.named_parameters():
name = name.replace('.', '/')
k = "{}grads_{}/{}".format(tag_prefix, self.reduction.__name__, name)
v = float(self.reduction(p.grad))
self.add_scalar(logger, k, v, event_name, global_step)
logger._save()
class VisdomLogger(BaseLogger):
"""
VisdomLogger handler to log metrics, model/optimizer parameters, gradients during the training and validation.
This class requires `visdom <https://github.com/facebookresearch/visdom/>`_ package to be installed:
.. code-block:: bash
pip install git+https://github.com/facebookresearch/visdom.git
Args:
server (str, optional): visdom server URL. It can be also specified by environment variable `VISDOM_SERVER_URL`
port (int, optional): visdom server's port. It can be also specified by environment variable `VISDOM_PORT`
num_workers (int, optional): number of workers to use in `concurrent.futures.ThreadPoolExecutor` to post data to
visdom server. Default, `num_workers=1`. If `num_workers=0` and logger uses the main thread. If using
Python 2.7 and `num_workers>0` the package `futures` should be installed: `pip install futures`
**kwargs: kwargs to pass into
`visdom.Visdom <https://github.com/facebookresearch/visdom#visdom-arguments-python-only>`_.
Note:
We can also specify username/password using environment variables: VISDOM_USERNAME, VISDOM_PASSWORD
.. warning::
Frequent logging, e.g. when logger is attached to `Events.ITERATION_COMPLETED`, can slow down the run if the
main thread is used to send the data to visdom server (`num_workers=0`). To avoid this situation we can either
log less frequently or set `num_workers=1`.
Examples:
.. code-block:: python
from ignite.contrib.handlers.visdom_logger import *
# Create a logger
vd_logger = VisdomLogger()
# Attach the logger to the trainer to log training loss at each iteration
vd_logger.attach(trainer,
log_handler=OutputHandler(tag="training", output_transform=lambda loss: {'loss': loss}),
event_name=Events.ITERATION_COMPLETED)
# Attach the logger to the evaluator on the training dataset and log NLL, Accuracy metrics after each epoch
# We setup `global_step_transform=global_step_from_engine(trainer)` to take the epoch
# of the `trainer` instead of `train_evaluator`:
vd_logger.attach(train_evaluator,
log_handler=OutputHandler(tag="training",
metric_names=["nll", "accuracy"],
global_step_transform=global_step_from_engine(trainer)),
event_name=Events.EPOCH_COMPLETED)
# Attach the logger to the evaluator on the validation dataset and log NLL, Accuracy metrics after
# each epoch. We setup `global_step_transform=global_step_from_engine(trainer)` to take the epoch of
# the `trainer` instead of `evaluator`:
vd_logger.attach(evaluator,
log_handler=OutputHandler(tag="validation",
metric_names=["nll", "accuracy"],
global_step_transform=global_step_from_engine(trainer),
event_name=Events.EPOCH_COMPLETED)
# Attach the logger to the trainer to log optimizer's parameters, e.g. learning rate at each iteration
vd_logger.attach(trainer,
log_handler=optimizer_params_handler(optimizer),
event_name=Events.ITERATION_COMPLETED)
# Attach the logger to the trainer to log model's weights norm after each iteration
vd_logger.attach(trainer,
log_handler=weights_scalar_handler(model),
event_name=Events.ITERATION_COMPLETED)
# Attach the logger to the trainer to log model's gradients norm after each iteration
vd_logger.attach(trainer,
log_handler=grads_scalar_handler(model),
event_name=Events.ITERATION_COMPLETED)
# We need to close the logger with we are done
vd_logger.close()
It is also possible to use the logger as context manager:
.. code-block:: python
from ignite.contrib.handlers.visdom_logger import *
with VisdomLogger() as vd_logger:
trainer = Engine(update_fn)
# Attach the logger to the trainer to log training loss at each iteration
vd_logger.attach(trainer,
log_handler=OutputHandler(tag="training",
output_transform=lambda loss: {'loss': loss}),
event_name=Events.ITERATION_COMPLETED)
"""
def __init__(self, server=None, port=None, num_workers=1, **kwargs):
try:
import visdom
except ImportError:
raise RuntimeError("This contrib module requires visdom package. "
"Please install it with command:\n"
"pip install git+https://github.com/facebookresearch/visdom.git")
if num_workers > 0:
# If visdom is installed, one of its dependencies `tornado`
# requires also `futures` to be installed.
# Let's check anyway if we can import it.
try:
import concurrent.futures
except ImportError:
raise RuntimeError("This contrib module requires concurrent.futures module"
"Please install it with command:\n"
"pip install futures")
if server is None:
server = os.environ.get("VISDOM_SERVER_URL", 'localhost')
if port is None:
port = int(os.environ.get("VISDOM_PORT", 8097))
if "username" not in kwargs:
username = os.environ.get("VISDOM_USERNAME", None)
kwargs["username"] = username
if "password" not in kwargs:
password = os.environ.get("VISDOM_PASSWORD", None)
kwargs["password"] = password
self.vis = visdom.Visdom(
server=server,
port=port,
**kwargs
)
if not self.vis.check_connection():
raise RuntimeError("Failed to connect to Visdom server at {}. "
"Did you run python -m visdom.server ?".format(server))
self.executor = _DummyExecutor()
if num_workers > 0:
from concurrent.futures import ThreadPoolExecutor
self.executor = ThreadPoolExecutor(max_workers=num_workers)
def _save(self):
self.vis.save([self.vis.env])
def close(self):
self.vis = None
self.executor.shutdown()
class _DummyExecutor:
class _DummyFuture:
def __init__(self, result):
self._output = result
def result(self):
return self._output
def __init__(self, *args, **kwargs):
pass
def submit(self, fn, **kwargs):
return _DummyExecutor._DummyFuture(fn(**kwargs))
def shutdown(self, *args, **kwargs):
pass
You can’t perform that action at this time.