pytorch_pfn_extras.training.extensions.PlotReport¶

class pytorch_pfn_extras.training.extensions.PlotReport(y_keys, x_key='iteration', trigger=(1, 'epoch'), postprocess=None, filename='plot.png', marker='x', grid=True)¶

An extension to output plots.

This extension accumulates the observations of the manager to DictSummary at a regular interval specified by a supplied trigger, and plot a graph with using them.

There are two triggers to handle this extension. One is the trigger to invoke this extension, which is used to handle the timing of accumulating the results. It is set to 1, 'iteration' by default. The other is the trigger to determine when to emit the result. When this trigger returns True, this extension appends the summary of accumulated values to the list of past summaries, and writes the list to the log file. Then, this extension makes a new fresh summary object which is used until the next time that the trigger fires.

It also adds 'epoch' and 'iteration' entries to each result dictionary, which are the epoch and iteration counts at the output.

Warning

If your environment needs to specify a backend of matplotlib explicitly, please call matplotlib.use before calling manager.run_iteration. For example:

import matplotlib
matplotlib.use('Agg')

manager.extend(
    extensions.PlotReport(['main/loss', 'validation/main/loss'],
                          'epoch', filename='loss.png'))
with manager.run_iteration():
    pass

Then, once one of instances of this extension is called, matplotlib.use will have no effect.

For the details, please see here: https://matplotlib.org/faq/usage_faq.html#what-is-a-backend

Parameters

y_keys (iterable of strs) – Keys of values regarded as y. If this is None, nothing is output to the graph.
x_key (str) – Keys of values regarded as x. The default value is ‘iteration’.
trigger – Trigger that decides when to aggregate the result and output the values. This is distinct from the trigger of this extension itself. If it is a tuple in the form <int>, 'epoch' or <int>, 'iteration', it is passed to IntervalTrigger.
postprocess – Callback to postprocess the result dictionaries. Figure object, Axes object, and all plot data are passed to this callback in this order. This callback can modify the figure.
filename (str) – Name of the figure file under the output directory. It can be a format string. For historical reasons file_name is also accepted as an alias of this argument.
marker (str) – The marker used to plot the graph. Default is 'x'. If None is given, it draws with no markers.
grid (bool) – If True, set the axis grid on. The default value is True.
writer (writer object, optional) – must be callable. object to dump the log to. If specified, it needs to have a correct savefun defined. The writer can override the save location in the pytorch_pfn_extras.training.ExtensionsManager object

__init__(y_keys, x_key='iteration', trigger=(1, 'epoch'), postprocess=None, filename=None, marker='x', grid=True, **kwargs)¶

Methods

`__init__`(y_keys[, x_key, trigger, …])
`available`()
`finalize`()	Finalizes the extension.
`initialize`(manager)	Initializes up the manager state.
`load_state_dict`(to_load)
`on_error`(manager, exc, tb)	Handles the error raised during training before finalization.
`state_dict`()	Serializes the extension state.

Attributes

`default_name`	Default name of the extension.
`name`
`priority`
`trigger`

pytorch_pfn_extras.training.extensions.ParameterStatistics pytorch_pfn_extras.training.extensions.PrintReport