File size: 17,476 Bytes
9c6594c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 |
# Copyright The 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.
from collections.abc import Sequence
from copy import deepcopy
from typing import Any, Optional, Union
import torch
from torch import Tensor
from torch.nn import ModuleList
from torchmetrics.collections import MetricCollection
from torchmetrics.metric import Metric
from torchmetrics.utilities.imports import _MATPLOTLIB_AVAILABLE
from torchmetrics.utilities.plot import _AX_TYPE, _PLOT_OUT_TYPE, plot_single_or_multi_val
from torchmetrics.utilities.prints import rank_zero_warn
if not _MATPLOTLIB_AVAILABLE:
__doctest_skip__ = ["MetricTracker.plot"]
class MetricTracker(ModuleList):
"""A wrapper class that can help keeping track of a metric or metric collection over time.
The wrapper implements the standard ``.update()``, ``.compute()``, ``.reset()`` methods that just
calls corresponding method of the currently tracked metric. However, the following additional methods are
provided:
-``MetricTracker.n_steps``: number of metrics being tracked
-``MetricTracker.increment()``: initialize a new metric for being tracked
-``MetricTracker.compute_all()``: get the metric value for all steps
-``MetricTracker.best_metric()``: returns the best value
Out of the box, this wrapper class fully supports that the base metric being tracked is a single `Metric`, a
`MetricCollection` or another `MetricWrapper` wrapped around a metric. However, multiple layers of nesting, such
as using a `Metric` inside a `MetricWrapper` inside a `MetricCollection` is not fully supported, especially the
`.best_metric` method that cannot auto compute the best metric and index for such nested structures.
Args:
metric: instance of a ``torchmetrics.Metric`` or ``torchmetrics.MetricCollection``
to keep track of at each timestep.
maximize: either single bool or list of bool indicating if higher metric values are
better (``True``) or lower is better (``False``).
Example (single metric):
>>> from torch import randint
>>> from torchmetrics.wrappers import MetricTracker
>>> from torchmetrics.classification import MulticlassAccuracy
>>> tracker = MetricTracker(MulticlassAccuracy(num_classes=10, average='micro'))
>>> for epoch in range(5):
... tracker.increment()
... for batch_idx in range(5):
... tracker.update(randint(10, (100,)), randint(10, (100,)))
... print(f"current acc={tracker.compute()}")
current acc=0.1120000034570694
current acc=0.08799999952316284
current acc=0.12600000202655792
current acc=0.07999999821186066
current acc=0.10199999809265137
>>> best_acc, which_epoch = tracker.best_metric(return_step=True)
>>> best_acc # doctest: +ELLIPSIS
0.1260...
>>> which_epoch
2
>>> tracker.compute_all()
tensor([0.1120, 0.0880, 0.1260, 0.0800, 0.1020])
Example (multiple metrics using MetricCollection):
>>> from torch import randn
>>> from torchmetrics.wrappers import MetricTracker
>>> from torchmetrics import MetricCollection
>>> from torchmetrics.regression import MeanSquaredError, ExplainedVariance
>>> tracker = MetricTracker(MetricCollection([MeanSquaredError(), ExplainedVariance()]), maximize=[False, True])
>>> for epoch in range(5):
... tracker.increment()
... for batch_idx in range(5):
... tracker.update(randn(100), randn(100))
... print(f"current stats={tracker.compute()}") # doctest: +NORMALIZE_WHITESPACE
current stats={'MeanSquaredError': tensor(2.3292), 'ExplainedVariance': tensor(-0.9516)}
current stats={'MeanSquaredError': tensor(2.1370), 'ExplainedVariance': tensor(-1.0775)}
current stats={'MeanSquaredError': tensor(2.1695), 'ExplainedVariance': tensor(-0.9945)}
current stats={'MeanSquaredError': tensor(2.1072), 'ExplainedVariance': tensor(-1.1878)}
current stats={'MeanSquaredError': tensor(2.0562), 'ExplainedVariance': tensor(-1.0754)}
>>> from pprint import pprint
>>> best_res, which_epoch = tracker.best_metric(return_step=True)
>>> pprint(best_res) # doctest: +ELLIPSIS
{'ExplainedVariance': -0.951...,
'MeanSquaredError': 2.056...}
>>> which_epoch
{'MeanSquaredError': 4, 'ExplainedVariance': 0}
>>> pprint(tracker.compute_all())
{'ExplainedVariance': tensor([-0.9516, -1.0775, -0.9945, -1.1878, -1.0754]),
'MeanSquaredError': tensor([2.3292, 2.1370, 2.1695, 2.1072, 2.0562])}
"""
maximize: Union[bool, list[bool]]
_base_metric: Union[Metric, MetricCollection]
def __init__(
self, metric: Union[Metric, MetricCollection], maximize: Optional[Union[bool, list[bool]]] = True
) -> None:
super().__init__()
if not isinstance(metric, (Metric, MetricCollection)):
raise TypeError(
f"Metric arg need to be an instance of a torchmetrics `Metric` or `MetricCollection` but got {metric}"
)
self._base_metric = metric
if maximize is None:
if isinstance(metric, Metric):
if getattr(metric, "higher_is_better", None) is None:
raise AttributeError(
f"The metric '{metric.__class__.__name__}' does not have a 'higher_is_better' attribute."
" Please provide the `maximize` argument explicitly."
)
self.maximize = metric.higher_is_better # type: ignore[assignment] # this is false alarm
elif isinstance(metric, MetricCollection):
self.maximize = []
for name, m in metric.items():
if getattr(m, "higher_is_better", None) is None:
raise AttributeError(
f"The metric '{name}' in the MetricCollection does not have a 'higher_is_better' attribute."
" Please provide the `maximize` argument explicitly."
)
self.maximize.append(m.higher_is_better) # type: ignore[arg-type] # this is false alarm
else:
rank_zero_warn(
"The default value for `maximize` will be changed from `True` to `None` in v1.7.0 of TorchMetrics,"
"will automatically infer the value based on the `higher_is_better` attribute of the metric"
" (if such attribute exists) or raise an error if it does not. If you are explicitly setting the"
" `maximize` argument to either `True` or `False` already, you can ignore this warning.",
FutureWarning,
)
if not isinstance(maximize, (bool, list)):
raise ValueError("Argument `maximize` should either be a single bool or list of bool")
if isinstance(maximize, list) and not all(isinstance(m, bool) for m in maximize):
raise ValueError("Argument `maximize` is list but not type of bool.")
if isinstance(maximize, list) and isinstance(metric, MetricCollection) and len(maximize) != len(metric):
raise ValueError("The len of argument `maximize` should match the length of the metric collection")
if isinstance(metric, Metric) and not isinstance(maximize, bool):
raise ValueError("Argument `maximize` should be a single bool when `metric` is a single Metric")
self.maximize = maximize
self._increment_called = False
@property
def n_steps(self) -> int:
"""Returns the number of times the tracker has been incremented."""
return len(self) - 1 # subtract the base metric
def increment(self) -> None:
"""Create a new instance of the input metric that will be updated next."""
self._increment_called = True
self.append(deepcopy(self._base_metric))
def forward(self, *args: Any, **kwargs: Any) -> None:
"""Call forward of the current metric being tracked."""
self._check_for_increment("forward")
if not isinstance(self[-1], (Metric, MetricCollection)):
raise TypeError(f"Expected the last item to be a Metric or MetricCollection, but got {type(self[-1])}.")
return self[-1](*args, **kwargs)
def update(self, *args: Any, **kwargs: Any) -> None:
"""Update the current metric being tracked."""
self._check_for_increment("update")
if not isinstance(self[-1], (Metric, MetricCollection)):
raise TypeError(f"Expected the last item to be a Metric or MetricCollection, but got {type(self[-1])}.")
self[-1].update(*args, **kwargs)
def compute(self) -> Any:
"""Call compute of the current metric being tracked."""
self._check_for_increment("compute")
if not isinstance(self[-1], (Metric, MetricCollection)):
raise TypeError(f"Expected the last item to be a Metric or MetricCollection, but got {type(self[-1])}.")
return self[-1].compute()
def compute_all(self) -> Any:
"""Compute the metric value for all tracked metrics.
Return:
By default will try stacking the results from all increments into a single tensor if the tracked base
object is a single metric. If a metric collection is provided a dict of stacked tensors will be returned.
If the stacking process fails a list of the computed results will be returned.
Raises:
ValueError:
If `self.increment` have not been called before this method is called.
"""
self._check_for_increment("compute_all")
# The i!=0 accounts for the self._base_metric should be ignored
res = [metric.compute() for i, metric in enumerate(self) if i != 0]
try:
if isinstance(res[0], dict):
keys = res[0].keys()
return {k: torch.stack([r[k] for r in res], dim=0) for k in keys}
if isinstance(res[0], list):
return torch.stack([torch.stack(r, dim=0) for r in res], 0)
return torch.stack(res, dim=0)
except TypeError: # fallback solution to just return as it is if we cannot successfully stack
return res
def reset(self) -> None:
"""Reset the current metric being tracked."""
if not isinstance(self[-1], (Metric, MetricCollection)):
raise TypeError(f"Expected the last item to be a Metric or MetricCollection, but got {type(self[-1])}.")
self[-1].reset()
def reset_all(self) -> None:
"""Reset all metrics being tracked."""
for metric in self:
if not isinstance(metric, (Metric, MetricCollection)):
raise TypeError(f"Expected all metrics to be Metric or MetricCollection, but got {type(metric)}.")
metric.reset()
def best_metric(
self, return_step: bool = False
) -> Union[
None,
float,
Tensor,
tuple[Union[int, float, Tensor], Union[int, float, Tensor]],
tuple[None, None],
dict[str, Union[float, None]],
tuple[dict[str, Union[float, None]], dict[str, Union[int, None]]],
]:
"""Return the highest metric out of all tracked.
Args:
return_step: If ``True`` will also return the step with the highest metric value.
Returns:
Either a single value or a tuple, depends on the value of ``return_step`` and the object being tracked.
- If a single metric is being tracked and ``return_step=False`` then a single tensor will be returned
- If a single metric is being tracked and ``return_step=True`` then a 2-element tuple will be returned,
where the first value is optimal value and second value is the corresponding optimal step
- If a metric collection is being tracked and ``return_step=False`` then a single dict will be returned,
where keys correspond to the different values of the collection and the values are the optimal metric
value
- If a metric collection is being bracked and ``return_step=True`` then a 2-element tuple will be returned
where each is a dict, with keys corresponding to the different values of th collection and the values
of the first dict being the optimal values and the values of the second dict being the optimal step
In addition the value in all cases may be ``None`` if the underlying metric does have a proper defined way
of being optimal or in the case where a nested structure of metrics are being tracked.
"""
res = self.compute_all()
if isinstance(res, list):
rank_zero_warn(
"Encountered nested structure. You are probably using a metric collection inside a metric collection,"
" or a metric wrapper inside a metric collection, which is not supported by `.best_metric()` method."
" Returning `None` instead."
)
if return_step:
return None, None
return None
if isinstance(self._base_metric, Metric):
fn = torch.max if self.maximize else torch.min
try:
value, idx = fn(res, 0)
if return_step:
return value.item(), idx.item()
return value.item()
except (ValueError, RuntimeError) as error:
rank_zero_warn(
f"Encountered the following error when trying to get the best metric: {error}"
"this is probably due to the 'best' not being defined for this metric."
"Returning `None` instead.",
UserWarning,
)
if return_step:
return None, None
return None
else: # this is a metric collection
maximize = self.maximize if isinstance(self.maximize, list) else len(res) * [self.maximize]
value, idx = {}, {} # type: ignore[assignment]
for i, (k, v) in enumerate(res.items()):
try:
fn = torch.max if maximize[i] else torch.min
out = fn(v, 0)
value[k], idx[k] = out[0].item(), out[1].item()
except (ValueError, RuntimeError) as error: # noqa: PERF203 # todo
rank_zero_warn(
f"Encountered the following error when trying to get the best metric for metric {k}:"
f"{error} this is probably due to the 'best' not being defined for this metric."
"Returning `None` instead.",
UserWarning,
)
value[k], idx[k] = None, None # type: ignore[assignment]
if return_step:
return value, idx
return value
def _check_for_increment(self, method: str) -> None:
"""Check that a metric that can be updated/used for computations has been initialized."""
if not self._increment_called:
raise ValueError(f"`{method}` cannot be called before `.increment()` has been called.")
def plot(
self, val: Optional[Union[Tensor, Sequence[Tensor]]] = None, ax: Optional[_AX_TYPE] = None
) -> _PLOT_OUT_TYPE:
"""Plot a single or multiple values from the metric.
Args:
val: Either a single result from calling `metric.forward` or `metric.compute` or a list of these results.
If no value is provided, will automatically call `metric.compute` and plot that result.
ax: An matplotlib axis object. If provided will add plot to that axis
Returns:
Figure and Axes object
Raises:
ModuleNotFoundError:
If `matplotlib` is not installed
.. plot::
:scale: 75
>>> # Example plotting a single value
>>> import torch
>>> from torchmetrics.wrappers import MetricTracker
>>> from torchmetrics.classification import BinaryAccuracy
>>> tracker = MetricTracker(BinaryAccuracy())
>>> for epoch in range(5):
... tracker.increment()
... for batch_idx in range(5):
... tracker.update(torch.randint(2, (10,)), torch.randint(2, (10,)))
>>> fig_, ax_ = tracker.plot() # plot all epochs
"""
val = val if val is not None else self.compute_all()
fig, ax = plot_single_or_multi_val(
val,
ax=ax,
name=self.__class__.__name__,
)
return fig, ax
|