pytorch-ignite 0.6.0.dev20250324__py3-none-any.whl → 0.6.0.dev20251103__py3-none-any.whl

ignite/metrics/clustering/silhouette_score.py

@@ -86,7 +86,7 @@ class SilhouetteScore(_ClusteringMetricBase):
 
         .. testoutput::
 
-            0.12607366
+            0.1260736584663391
 
     .. versionadded:: 0.5.2
     """

ignite/metrics/fbeta.py

@@ -1,4 +1,4 @@
-from typing import Callable, Optional, Union
+from typing import Callable, cast, Optional, Union
 
 import torch
 
@@ -15,7 +15,7 @@ def Fbeta(
     precision: Optional[Precision] = None,
     recall: Optional[Recall] = None,
     output_transform: Optional[Callable] = None,
-    device: Union[str, torch.device] = torch.device("cpu"),
+    device: Optional[Union[str, torch.device]] = None,
 ) -> MetricsLambda:
     r"""Calculates F-beta score.
 
@@ -143,17 +143,26 @@ def Fbeta(
     if not (beta > 0):
         raise ValueError(f"Beta should be a positive integer, but given {beta}")
 
-    if precision is not None and output_transform is not None:
-        raise ValueError("If precision argument is provided, output_transform should be None")
+    if precision is not None:
+        if output_transform is not None:
+            raise ValueError("If precision argument is provided, output_transform should be None")
+        if device is not None:
+            raise ValueError("If precision argument is provided, device should be None")
 
-    if recall is not None and output_transform is not None:
-        raise ValueError("If recall argument is provided, output_transform should be None")
+    if recall is not None:
+        if output_transform is not None:
+            raise ValueError("If recall argument is provided, output_transform should be None")
+        if device is not None:
+            raise ValueError("If recall argument is provided, device should be None")
+
+    if precision is None and recall is None and device is None:
+        device = torch.device("cpu")
 
     if precision is None:
         precision = Precision(
             output_transform=(lambda x: x) if output_transform is None else output_transform,
             average=False,
-            device=device,
+            device=cast(Union[str, torch.device], recall._device if recall else device),
         )
     elif precision._average:
         raise ValueError("Input precision metric should have average=False")
@@ -162,7 +171,7 @@ def Fbeta(
         recall = Recall(
             output_transform=(lambda x: x) if output_transform is None else output_transform,
             average=False,
-            device=device,
+            device=cast(Union[str, torch.device], precision._device if precision else device),
         )
     elif recall._average:
         raise ValueError("Input recall metric should have average=False")

ignite/metrics/gan/fid.py

@@ -31,13 +31,13 @@ def fid_score(
     except ImportError:
         raise ModuleNotFoundError("fid_score requires scipy to be installed.")
 
-    mu1, mu2 = mu1.cpu(), mu2.cpu()
-    sigma1, sigma2 = sigma1.cpu(), sigma2.cpu()
+    mu1, mu2 = mu1.detach().cpu(), mu2.detach().cpu()
+    sigma1, sigma2 = sigma1.detach().cpu(), sigma2.detach().cpu()
 
     diff = mu1 - mu2
 
     # Product might be almost singular
-    covmean, _ = scipy.linalg.sqrtm(sigma1.mm(sigma2), disp=False)
+    covmean, _ = scipy.linalg.sqrtm(sigma1.mm(sigma2).numpy(), disp=False)
     # Numerical error might give slight imaginary component
     if np.iscomplexobj(covmean):
         if not np.allclose(np.diagonal(covmean).imag, 0, atol=1e-3):

ignite/metrics/js_divergence.py

@@ -73,7 +73,7 @@ class JSDivergence(KLDivergence):
 
         .. testoutput::
 
-           0.16266516844431558
+           0.1626...
 
     .. versionchanged:: 0.5.1
         ``skip_unrolling`` argument is added.

ignite/metrics/maximum_mean_discrepancy.py

@@ -78,7 +78,7 @@ class MaximumMeanDiscrepancy(Metric):
 
         .. testoutput::
 
-           1.072697639465332
+           1.0726...
 
     .. versionchanged:: 0.5.1
         ``skip_unrolling`` argument is added.

ignite/metrics/metric.py

@@ -361,6 +361,8 @@ class Metric(Serializable, metaclass=ABCMeta):
         device: Union[str, torch.device] = torch.device("cpu"),
         skip_unrolling: bool = False,
     ):
+        if not callable(output_transform):
+            raise TypeError(f"Argument output_transform should be callable, got {type(output_transform)}")
         self._output_transform = output_transform
 
         # Some metrics have a large performance regression when run on XLA devices, so for now, we disallow it.

ignite/metrics/nlp/bleu.py

@@ -2,6 +2,7 @@ import math
 from typing import Any, Callable, Sequence, Tuple, Union
 
 import torch
+from torch import Tensor
 
 from ignite.exceptions import NotComputableError
 from ignite.metrics.metric import Metric, reinit__is_reduced, sync_all_reduce
@@ -71,11 +72,11 @@ class Bleu(Metric):
 
     More details can be found in `Papineni et al. 2002`__.
 
-    __ https://www.aclweb.org/anthology/P02-1040
+    __ https://aclanthology.org/P02-1040/
 
     In addition, a review of smoothing techniques can be found in `Chen et al. 2014`__
 
-    __ https://aclanthology.org/W14-3346.pdf
+    __ https://aclanthology.org/W14-3346/
 
     - ``update`` must receive output of the form ``(y_pred, y)`` or ``{'y_pred': y_pred, 'y': y}``.
     - `y_pred` (list(list(str))) - a list of hypotheses sentences.
@@ -236,12 +237,12 @@ class Bleu(Metric):
     @reinit__is_reduced
     def reset(self) -> None:
         if self.average == "macro":
-            self._sum_of_bleu = torch.tensor(0.0, dtype=torch.double, device=self._device)
+            self._sum_of_bleu = torch.tensor(0.0, dtype=self._double_dtype, device=self._device)
             self._num_sentences = 0
 
         if self.average == "micro":
-            self.p_numerators = torch.zeros(self.ngrams_order + 1)
-            self.p_denominators = torch.zeros(self.ngrams_order + 1)
+            self.p_numerators = torch.zeros(self.ngrams_order + 1, dtype=self._double_dtype)
+            self.p_denominators = torch.zeros(self.ngrams_order + 1, dtype=self._double_dtype)
             self.hyp_length_sum = 0
             self.ref_length_sum = 0
 
@@ -278,8 +279,9 @@ class Bleu(Metric):
         )
         return bleu_score
 
-    def compute(self) -> None:
+    def compute(self) -> Union[None, Tensor, float]:
         if self.average == "macro":
             return self._compute_macro()
         elif self.average == "micro":
             return self._compute_micro()
+        return None

ignite/metrics/nlp/rouge.py

@@ -191,7 +191,7 @@ class RougeN(_BaseRouge):
 
     More details can be found in `Lin 2004`__.
 
-    __ https://www.aclweb.org/anthology/W04-1013.pdf
+    __ https://aclanthology.org/W04-1013
 
     - ``update`` must receive output of the form ``(y_pred, y)`` or ``{'y_pred': y_pred, 'y': y}``.
     - `y_pred` (list(list(str))) must be a sequence of tokens.
@@ -265,7 +265,7 @@ class RougeL(_BaseRouge):
 
     More details can be found in `Lin 2004`__.
 
-    __ https://www.aclweb.org/anthology/W04-1013.pdf
+    __ https://aclanthology.org/W04-1013
 
     - ``update`` must receive output of the form ``(y_pred, y)`` or ``{'y_pred': y_pred, 'y': y}``.
     - `y_pred` (list(list(str))) must be a sequence of tokens.
@@ -331,7 +331,7 @@ class Rouge(Metric):
 
     More details can be found in `Lin 2004`__.
 
-    __ https://www.aclweb.org/anthology/W04-1013.pdf
+    __ https://aclanthology.org/W04-1013
 
     - ``update`` must receive output of the form ``(y_pred, y)`` or ``{'y_pred': y_pred, 'y': y}``.
     - `y_pred` (list(list(str))) must be a sequence of tokens.

ignite/metrics/nlp/utils.py

@@ -63,7 +63,7 @@ def modified_precision(references: Sequence[Sequence[Any]], candidate: Any, n: i
 
     More details can be found in `Papineni et al. 2002`__.
 
-    __ https://www.aclweb.org/anthology/P02-1040.pdf
+    __ https://aclanthology.org/P02-1040
 
     Args:
         references: list of references R

ignite/metrics/precision_recall_curve.py

@@ -97,7 +97,7 @@ class PrecisionRecallCurve(EpochMetric):
         if len(self._predictions) < 1 or len(self._targets) < 1:
             raise NotComputableError("PrecisionRecallCurve must have at least one example before it can be computed.")
 
-        if self._result is None:  # type: ignore
+        if self._result is None:
             _prediction_tensor = torch.cat(self._predictions, dim=0)
             _target_tensor = torch.cat(self._targets, dim=0)
 
@@ -110,11 +110,11 @@ class PrecisionRecallCurve(EpochMetric):
             if idist.get_rank() == 0:
                 # Run compute_fn on zero rank only
                 precision, recall, thresholds = cast(Tuple, self.compute_fn(_prediction_tensor, _target_tensor))
-                precision = torch.tensor(precision, device=_prediction_tensor.device)
-                recall = torch.tensor(recall, device=_prediction_tensor.device)
+                precision = torch.tensor(precision, device=_prediction_tensor.device, dtype=self._double_dtype)
+                recall = torch.tensor(recall, device=_prediction_tensor.device, dtype=self._double_dtype)
                 # thresholds can have negative strides, not compatible with torch tensors
                 # https://discuss.pytorch.org/t/negative-strides-in-tensor-error/134287/2
-                thresholds = torch.tensor(thresholds.copy(), device=_prediction_tensor.device)
+                thresholds = torch.tensor(thresholds.copy(), device=_prediction_tensor.device, dtype=self._double_dtype)
             else:
                 precision, recall, thresholds = None, None, None
 
@@ -126,4 +126,4 @@ class PrecisionRecallCurve(EpochMetric):
 
             self._result = (precision, recall, thresholds)  # type: ignore[assignment]
 
-        return cast(Tuple[torch.Tensor, torch.Tensor, torch.Tensor], self._result)  # type: ignore
+        return cast(Tuple[torch.Tensor, torch.Tensor, torch.Tensor], self._result)

ignite/metrics/regression/_base.py

@@ -30,6 +30,10 @@ def _check_output_types(output: Tuple[torch.Tensor, torch.Tensor]) -> None:
 
 
 def _torch_median(output: torch.Tensor) -> float:
+    # torch.kthvalue used later is not supported on MPS
+    if output.device.type == "mps":
+        output = output.cpu()
+
     output = output.view(-1)
     len_ = len(output)
 

ignite/metrics/regression/fractional_bias.py

@@ -64,7 +64,7 @@ class FractionalBias(_BaseRegression):
 
     @reinit__is_reduced
     def reset(self) -> None:
-        self._sum_of_errors = torch.tensor(0.0, dtype=torch.double, device=self._device)
+        self._sum_of_errors = torch.tensor(0.0, dtype=self._double_dtype, device=self._device)
         self._num_examples = 0
 
     def _update(self, output: Tuple[torch.Tensor, torch.Tensor]) -> None:

ignite/metrics/roc_auc.py

@@ -198,9 +198,9 @@ class RocCurve(EpochMetric):
         if idist.get_rank() == 0:
             # Run compute_fn on zero rank only
             fpr, tpr, thresholds = cast(Tuple, self.compute_fn(_prediction_tensor, _target_tensor))
-            fpr = torch.tensor(fpr, device=_prediction_tensor.device)
-            tpr = torch.tensor(tpr, device=_prediction_tensor.device)
-            thresholds = torch.tensor(thresholds, device=_prediction_tensor.device)
+            fpr = torch.tensor(fpr, dtype=self._double_dtype, device=_prediction_tensor.device)
+            tpr = torch.tensor(tpr, dtype=self._double_dtype, device=_prediction_tensor.device)
+            thresholds = torch.tensor(thresholds, dtype=self._double_dtype, device=_prediction_tensor.device)
         else:
             fpr, tpr, thresholds = None, None, None
 

ignite/metrics/ssim.py

@@ -21,9 +21,9 @@ class SSIM(Metric):
 
     Args:
         data_range: Range of the image. Typically, ``1.0`` or ``255``.
-        kernel_size: Size of the kernel. Default: (11, 11)
+        kernel_size: Size of the kernel. Default: 11
         sigma: Standard deviation of the gaussian kernel.
-            Argument is used if ``gaussian=True``. Default: (1.5, 1.5)
+            Argument is used if ``gaussian=True``. Default: 1.5
         k1: Parameter of SSIM. Default: 0.01
         k2: Parameter of SSIM. Default: 0.03
         gaussian: ``True`` to use gaussian kernel, ``False`` to use uniform kernel
@@ -36,6 +36,7 @@ class SSIM(Metric):
         skip_unrolling: specifies whether output should be unrolled before being fed to update method. Should be
             true for multi-output model, for example, if ``y_pred`` contains multi-ouput as ``(y_pred_a, y_pred_b)``
             Alternatively, ``output_transform`` can be used to handle this.
+        ndims: Number of dimensions of the input image: 2d or 3d. Accepted values: 2, 3. Default: 2
 
     Examples:
         To use with ``Engine`` and ``process_function``, simply attach the metric instance to the engine.
@@ -68,6 +69,8 @@ class SSIM(Metric):
 
     .. versionchanged:: 0.5.1
         ``skip_unrolling`` argument is added.
+    .. versionchanged:: 0.5.2
+        ``ndims`` argument is added.
     """
 
     _state_dict_all_req_keys = ("_sum_of_ssim", "_num_examples", "_kernel")
@@ -75,28 +78,36 @@ class SSIM(Metric):
     def __init__(
         self,
         data_range: Union[int, float],
-        kernel_size: Union[int, Sequence[int]] = (11, 11),
-        sigma: Union[float, Sequence[float]] = (1.5, 1.5),
+        kernel_size: Union[int, Sequence[int]] = 11,
+        sigma: Union[float, Sequence[float]] = 1.5,
         k1: float = 0.01,
         k2: float = 0.03,
         gaussian: bool = True,
         output_transform: Callable = lambda x: x,
         device: Union[str, torch.device] = torch.device("cpu"),
         skip_unrolling: bool = False,
+        ndims: int = 2,
     ):
+        if ndims not in (2, 3):
+            raise ValueError(f"Expected ndims to be 2 or 3. Got {ndims}.")
+
         if isinstance(kernel_size, int):
-            self.kernel_size: Sequence[int] = [kernel_size, kernel_size]
+            self.kernel_size: Sequence[int] = [kernel_size for _ in range(ndims)]
         elif isinstance(kernel_size, Sequence):
+            if len(kernel_size) != ndims:
+                raise ValueError(f"Expected kernel_size to have length of {ndims}. Got {len(kernel_size)}.")
             self.kernel_size = kernel_size
         else:
-            raise ValueError("Argument kernel_size should be either int or a sequence of int.")
+            raise ValueError(f"Argument kernel_size should be either int or a sequence of int of length {ndims}.")
 
         if isinstance(sigma, float):
-            self.sigma: Sequence[float] = [sigma, sigma]
+            self.sigma: Sequence[float] = [sigma for _ in range(ndims)]
         elif isinstance(sigma, Sequence):
+            if len(sigma) != ndims:
+                raise ValueError(f"Expected sigma to have length of {ndims}. Got {len(sigma)}.")
             self.sigma = sigma
         else:
-            raise ValueError("Argument sigma should be either float or a sequence of float.")
+            raise ValueError(f"Argument sigma should be either float or a sequence of float of length {ndims}.")
 
         if any(x % 2 == 0 or x <= 0 for x in self.kernel_size):
             raise ValueError(f"Expected kernel_size to have odd positive number. Got {kernel_size}.")
@@ -111,7 +122,13 @@ class SSIM(Metric):
         self.c2 = (k2 * data_range) ** 2
         self.pad_h = (self.kernel_size[0] - 1) // 2
         self.pad_w = (self.kernel_size[1] - 1) // 2
-        self._kernel_2d = self._gaussian_or_uniform_kernel(kernel_size=self.kernel_size, sigma=self.sigma)
+        self.pad_d = None
+        self.ndims = ndims
+        if self.ndims == 3:
+            self.pad_d = (self.kernel_size[2] - 1) // 2
+        self._kernel_nd = self._gaussian_or_uniform_kernel(
+            kernel_size=self.kernel_size, sigma=self.sigma, ndims=self.ndims
+        )
         self._kernel: Optional[torch.Tensor] = None
 
     @reinit__is_reduced
@@ -128,23 +145,34 @@ class SSIM(Metric):
         min_, max_ = -2.5, 2.5
         kernel[start_uniform_index:end_uniform_index] = 1 / (max_ - min_)
 
-        return kernel.unsqueeze(dim=0)  # (1, kernel_size)
+        return kernel  # (kernel_size)
 
     def _gaussian(self, kernel_size: int, sigma: float) -> torch.Tensor:
         ksize_half = (kernel_size - 1) * 0.5
         kernel = torch.linspace(-ksize_half, ksize_half, steps=kernel_size, device=self._device)
         gauss = torch.exp(-0.5 * (kernel / sigma).pow(2))
-        return (gauss / gauss.sum()).unsqueeze(dim=0)  # (1, kernel_size)
+        return gauss / gauss.sum()  # (kernel_size)
 
-    def _gaussian_or_uniform_kernel(self, kernel_size: Sequence[int], sigma: Sequence[float]) -> torch.Tensor:
+    def _gaussian_or_uniform_kernel(
+        self, kernel_size: Sequence[int], sigma: Sequence[float], ndims: int
+    ) -> torch.Tensor:
         if self.gaussian:
             kernel_x = self._gaussian(kernel_size[0], sigma[0])
             kernel_y = self._gaussian(kernel_size[1], sigma[1])
+            if ndims == 3:
+                kernel_z = self._gaussian(kernel_size[2], sigma[2])
         else:
             kernel_x = self._uniform(kernel_size[0])
             kernel_y = self._uniform(kernel_size[1])
+            if ndims == 3:
+                kernel_z = self._uniform(kernel_size[2])
 
-        return torch.matmul(kernel_x.t(), kernel_y)  # (kernel_size, 1) * (1, kernel_size)
+        result = (
+            torch.einsum("i,j->ij", kernel_x, kernel_y)
+            if ndims == 2
+            else torch.einsum("i,j,k->ijk", kernel_x, kernel_y, kernel_z)
+        )
+        return result
 
     def _check_type_and_shape(self, y_pred: torch.Tensor, y: torch.Tensor) -> None:
         if y_pred.dtype != y.dtype:
@@ -157,9 +185,11 @@ class SSIM(Metric):
                 f"Expected y_pred and y to have the same shape. Got y_pred: {y_pred.shape} and y: {y.shape}."
             )
 
-        if len(y_pred.shape) != 4 or len(y.shape) != 4:
+        # 2 dimensions are reserved for batch and channel
+        if len(y_pred.shape) - 2 != self.ndims or len(y.shape) - 2 != self.ndims:
             raise ValueError(
-                f"Expected y_pred and y to have BxCxHxW shape. Got y_pred: {y_pred.shape} and y: {y.shape}."
+                "Expected y_pred and y to have BxCxHxW or BxCxDxHxW shape. "
+                f"Got y_pred: {y_pred.shape} and y: {y.shape}."
             )
 
     @reinit__is_reduced
@@ -176,7 +206,7 @@ class SSIM(Metric):
 
         nb_channel = y_pred.size(1)
         if self._kernel is None or self._kernel.shape[0] != nb_channel:
-            self._kernel = self._kernel_2d.expand(nb_channel, 1, -1, -1)
+            self._kernel = self._kernel_nd.expand(nb_channel, 1, *[-1 for _ in range(self.ndims)])
 
         if y_pred.device != self._kernel.device:
             if self._kernel.device == torch.device("cpu"):
@@ -191,14 +221,19 @@ class SSIM(Metric):
                 y_pred = y_pred.to(device=self._kernel.device)
                 y = y.to(device=self._kernel.device)
 
-        y_pred = F.pad(y_pred, [self.pad_w, self.pad_w, self.pad_h, self.pad_h], mode="reflect")
-        y = F.pad(y, [self.pad_w, self.pad_w, self.pad_h, self.pad_h], mode="reflect")
+        padding_shape = [self.pad_w, self.pad_w, self.pad_h, self.pad_h]
+        if self.ndims == 3 and self.pad_d is not None:
+            padding_shape.extend([self.pad_d, self.pad_d])
+
+        y_pred = F.pad(y_pred, padding_shape, mode="reflect")
+        y = F.pad(y, padding_shape, mode="reflect")
 
         if y_pred.dtype != self._kernel.dtype:
             self._kernel = self._kernel.to(dtype=y_pred.dtype)
 
         input_list = [y_pred, y, y_pred * y_pred, y * y, y_pred * y]
-        outputs = F.conv2d(torch.cat(input_list), self._kernel, groups=nb_channel)
+        conv_op = F.conv3d if self.ndims == 3 else F.conv2d
+        outputs = conv_op(torch.cat(input_list), self._kernel, groups=nb_channel)
         batch_size = y_pred.size(0)
         output_list = [outputs[x * batch_size : (x + 1) * batch_size] for x in range(len(input_list))]
 
@@ -224,7 +259,10 @@ class SSIM(Metric):
         if ssim_idx.device.type == "mps" and self._double_dtype == torch.float64:
             double_dtype = torch.float32
 
-        self._sum_of_ssim += torch.mean(ssim_idx, (1, 2, 3), dtype=double_dtype).sum().to(device=self._device)
+        # mean from all dimensions except batch
+        self._sum_of_ssim += (
+            torch.mean(ssim_idx, list(range(1, 2 + self.ndims)), dtype=double_dtype).sum().to(device=self._device)
+        )
 
         self._num_examples += y.shape[0]
 

{pytorch_ignite-0.6.0.dev20250324.dist-info → pytorch_ignite-0.6.0.dev20251103.dist-info}/METADATA

@@ -1,23 +1,17 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.4
 Name: pytorch-ignite
-Version: 0.6.0.dev20250324
+Version: 0.6.0.dev20251103
 Summary: A lightweight library to help with training neural networks in PyTorch.
-Home-page: https://github.com/pytorch/ignite
-Author: PyTorch-Ignite Team
-Author-email: contact@pytorch-ignite.ai
-License: BSD
-Description-Content-Type: text/markdown
+Project-URL: Homepage, https://pytorch-ignite.ai
+Project-URL: Repository, https://github.com/pytorch/ignite
+Author-email: PyTorch-Ignite Team <contact@pytorch-ignite.ai>
+License-Expression: BSD-3-Clause
 License-File: LICENSE
-Requires-Dist: torch<3,>=1.3
+Classifier: Programming Language :: Python :: 3
+Requires-Python: <=3.13,>=3.9
 Requires-Dist: packaging
-Dynamic: author
-Dynamic: author-email
-Dynamic: description
-Dynamic: description-content-type
-Dynamic: home-page
-Dynamic: license
-Dynamic: requires-dist
-Dynamic: summary
+Requires-Dist: torch<3,>=1.10
+Description-Content-Type: text/markdown
 
 <div align="center">
 
@@ -418,7 +412,7 @@ Few pointers to get you started:
 - [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/pytorch/ignite/blob/master/examples/notebooks/FastaiLRFinder_MNIST.ipynb) [Basic example of LR finder on
   MNIST](https://github.com/pytorch/ignite/blob/master/examples/notebooks/FastaiLRFinder_MNIST.ipynb)
 - [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/pytorch/ignite/blob/master/examples/notebooks/Cifar100_bench_amp.ipynb) [Benchmark mixed precision training on Cifar100:
-  torch.cuda.amp vs nvidia/apex](https://github.com/pytorch/ignite/blob/master/examples/notebooks/Cifar100_bench_amp.ipynb)
+  torch.amp vs nvidia/apex](https://github.com/pytorch/ignite/blob/master/examples/notebooks/Cifar100_bench_amp.ipynb)
 - [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/pytorch/ignite/blob/master/examples/notebooks/MNIST_on_TPU.ipynb) [MNIST training on a single
   TPU](https://github.com/pytorch/ignite/blob/master/examples/notebooks/MNIST_on_TPU.ipynb)
 - [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/drive/1E9zJrptnLJ_PKhmaP5Vhb6DTVRvyrKHx) [CIFAR10 Training on multiple TPUs](https://github.com/pytorch/ignite/tree/master/examples/cifar10)

{pytorch_ignite-0.6.0.dev20250324.dist-info → pytorch_ignite-0.6.0.dev20251103.dist-info}/RECORD

@@ -1,4 +1,4 @@
-ignite/__init__.py,sha256=u53gJrQ1tKolHq-S8V3qTfCJ0Wnlh2uEx_j19_tjg1I,194
+ignite/__init__.py,sha256=tyL6Ulkx95oYbGTYtecqCVA424B4eABtJNaYIY5ujds,194
 ignite/_utils.py,sha256=XDPpUDJ8ykLXWMV2AYTqGSj8XCfApsyzsQ3Vij_OB4M,182
 ignite/exceptions.py,sha256=5ZWCVLPC9rgoW8t84D-VeEleqz5O7XpAGPpCdU8rKd0,150
 ignite/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -52,36 +52,36 @@ ignite/distributed/utils.py,sha256=D97JwWgL9RKP8rTfDRf1zMmfRUeJizr7XfLZ8LAScOI,2
 ignite/distributed/comp_models/__init__.py,sha256=S2WHl463U7BvpcUe9-JaGtuCi3G1cMHFW5QFBQ6fv20,1357
 ignite/distributed/comp_models/base.py,sha256=pTIylP1h2g6NWopBEponfXC6UefqS1l2lEdzTUTNXFc,14185
 ignite/distributed/comp_models/horovod.py,sha256=Mn5xG92bzBT1Dd-0IxEaN342kihUwJNkLCqsxP06ijc,9076
-ignite/distributed/comp_models/native.py,sha256=2kiHxJjjoS1GHs8GRuHcQQqnoz-yBhxGwks9Hv6SFI8,28087
+ignite/distributed/comp_models/native.py,sha256=oxJeK-Dc_dtLpG02wz29y5BvjJv_a3vEXPlRVRLWEIU,28098
 ignite/distributed/comp_models/xla.py,sha256=kVa5HrXaWt8TCLObxUMVpSsUnDv3SPOQ0dP2SlqpmLg,6281
-ignite/engine/__init__.py,sha256=ASvoTDzjl0ix7ZqHq_N3cDCNUy33gD_UUfs8-u9zni4,36126
+ignite/engine/__init__.py,sha256=MRFj6yywKhVkov4ccPkrw4dX1O8PfqceiJkngrcFb7A,36094
 ignite/engine/deterministic.py,sha256=lIacEolZPmfPgVWM_T0Eqg-2G08Wpi_hc1fifzFq0p8,11632
-ignite/engine/engine.py,sha256=zu7hpdKbmIUnGGW1uWVqxCHBseQkiNdeKAvT57A5zK8,59815
+ignite/engine/engine.py,sha256=xeUx7QkjC4p6iLMCTzFDU7Pq7PXoaoFaRf8TcpIuSog,60915
 ignite/engine/events.py,sha256=ydfG3HPMo3HKcycFSG_GrZ199Tuttcjmd85eQaV_5c0,21807
 ignite/engine/utils.py,sha256=QG5mkdg4OipspqgpNQcJuoHTYdr2Sx5LS16kfjOHDdI,1073
-ignite/handlers/__init__.py,sha256=t2UbfFcFNjR_zh3_WHoa0N3n-n_oD1jibbJw0JC3LsE,2641
-ignite/handlers/base_logger.py,sha256=3FXA0-4T2c0SXvJ1d2GE4yHN_B-Z4KTMdE4YlERZ-9U,11793
-ignite/handlers/checkpoint.py,sha256=xg0bEShzdAg7SKx8CLAPiWAy7pHi2Jm-EAQKb0NJpyo,44882
-ignite/handlers/clearml_logger.py,sha256=jZZ3t1m7LlbKZFuuMmfbf5Usaz8rf8aCnB0V5MOQCjg,37473
+ignite/handlers/__init__.py,sha256=Qq85YTtHPcii6UAfMOoCPg9RwigH96iqxOJKIlRfDqw,2728
+ignite/handlers/base_logger.py,sha256=wPiGn9iCh5ung1GaRUf_qAlqe63h1NpUUQ0XK709p2k,13011
+ignite/handlers/checkpoint.py,sha256=1k_RhDW5rjkJB4oz2jNWxjpuGgLvbNCt7_-1Pqz9Lxg,46266
+ignite/handlers/clearml_logger.py,sha256=12a9eue6hnFh5CrdSFz_EpGF0-XKRMlBXpR2NWWw8DY,37949
 ignite/handlers/early_stopping.py,sha256=UA6TiKho5CbD085R-16H8w3r0BYPQcWQjhEXg8aITSw,4139
 ignite/handlers/ema_handler.py,sha256=SmUyyWIFPZW3yMvjD_sSk5m_LfnMFl9R-uQdbXNFfY0,11854
-ignite/handlers/fbresearch_logger.py,sha256=onsUIHv6lYWcGf3VNeTWDCL_s1igK_PXmLGaTksjyk4,11120
-ignite/handlers/lr_finder.py,sha256=LdyBDQEg193mgTWJZHtZ8jP-L3giJSqSE3ffyo-wDoo,22117
-ignite/handlers/mlflow_logger.py,sha256=mMxPnBqJRcolYMtbBE8U32qneOdiwZOzOa7tEveon1Q,12311
-ignite/handlers/neptune_logger.py,sha256=LictGhcBeUNU5HrtmX3xi1qliNY1xMWPmGIgFC8Wzdo,27001
+ignite/handlers/fbresearch_logger.py,sha256=MfQeiBIXBYLEwZoDIld2oCceMeTAsz8rc5cd7fLtpJs,11133
+ignite/handlers/lr_finder.py,sha256=AAQLUcRLrfkrVOQufmRNZqAOTw1MpmAAo8YLk3JkdQs,22145
+ignite/handlers/mlflow_logger.py,sha256=M5Mggrnr2wMsms8wbEaHqNtTk5L1zNs1MlPWD0ZCpDQ,13894
+ignite/handlers/neptune_logger.py,sha256=SrehLclS8ccyuxO_0HYPvt5SN8EB8g9KWFfqQMQsGAw,27298
 ignite/handlers/param_scheduler.py,sha256=c730LIS6obDNNH2jitc2BRDK6AO36FfD3e1x336Oen4,68261
-ignite/handlers/polyaxon_logger.py,sha256=USJuycaEggP3J45fReP8-mwQeoWr-lKuGgNXJweRB0I,12056
+ignite/handlers/polyaxon_logger.py,sha256=5b7Zxhksne8Ufg_SBTG-rlf_9CPSjkBQOJR4-ynoZnQ,12354
 ignite/handlers/state_param_scheduler.py,sha256=xBOF07_JVexafmC-k4ifL_nN31IF8ThbebGWIxlbLs8,20745
 ignite/handlers/stores.py,sha256=8XM_Qqsitfu0WtOOE-K2FMtv51vD90r3GgQlCzRABYc,2616
-ignite/handlers/tensorboard_logger.py,sha256=BWcAo2dn7HCl0zVORHCKaSvtbLDBQbhkbiDEu86MRiU,26381
+ignite/handlers/tensorboard_logger.py,sha256=q3YxXkbIFayBggI_kcHyl-upttVVjjnqFOLgyjj2cRo,27967
 ignite/handlers/terminate_on_nan.py,sha256=RFSKd3Oqn9Me2xLCos4lSE-hnY7fYWWjE9blioeMlIs,2103
 ignite/handlers/time_limit.py,sha256=heTuS-ReBbOUCm1NcNJGhzxI080Hanc4hOLB2Y4GeZk,1567
 ignite/handlers/time_profilers.py,sha256=GZCoOpiFSc2yVgHQjpS1To8Yjb6G6HwydsiWMjwMQfA,30301
 ignite/handlers/timing.py,sha256=nHeBHvPwYdPRMAx-jk_8MjZit4a7rmsmIWkUrajAG-s,4705
-ignite/handlers/tqdm_logger.py,sha256=Yu8GBtTrKlO8H6_b9VS9obWJ_TIJ7p62OI_VwkGYdN4,13040
+ignite/handlers/tqdm_logger.py,sha256=5N70XA9rRm2x6sWYAJB1U5Y_bky2fa3qhec8SVgB3hY,13049
 ignite/handlers/utils.py,sha256=X4LRqo1kqGsbmX0pEuZKYR6K4C8sZudAqxCLriiXtCg,872
-ignite/handlers/visdom_logger.py,sha256=zDMjyBcVVHYExQFdxbrEXhoy37lpYuEXm_b7XL2JCEk,21551
-ignite/handlers/wandb_logger.py,sha256=0TQoOzjjexr7nE_eO3Aoo4PA5vSZJBnXK3uEuXRrtaA,14392
+ignite/handlers/visdom_logger.py,sha256=sg75ohEkDT7gYfEbLn5464GO-s0MLUEWxdFw_zSVSYw,21830
+ignite/handlers/wandb_logger.py,sha256=gGvbFNjc6gCfVFfOXcnz3-P4sqqP-P9at1UwUV_mwMg,14701
 ignite/metrics/__init__.py,sha256=m-8F8J17r-aEwsO6Ww-8AqDRN59WFfYBwCDKwqGDSmI,3627
 ignite/metrics/accumulation.py,sha256=xWdsm9u6JfsfODX_GUKzQc_omrdFDJ4yELBR-xXgc4s,12448
 ignite/metrics/accuracy.py,sha256=rI1TG-7WdJxcqGCMxGErXBWLmTNP1yczJgjjRyby0No,10168
@@ -92,50 +92,50 @@ ignite/metrics/confusion_matrix.py,sha256=dZDuK3vxrrbiQh6VfyV5aWFpuTJWsfnZ30Mxt6
 ignite/metrics/cosine_similarity.py,sha256=myq1iGFBBUgEhyOg_ZxkOqUQpS6FYAc3PAcnObc3Dp4,4429
 ignite/metrics/entropy.py,sha256=gJZkR5Sl1ZdIzJ9pFkydf1186bZU8OnkOLvOtKz6Wrs,4511
 ignite/metrics/epoch_metric.py,sha256=H4PVsDtcqk53l47Ehc3kliKT4QtyZUf600ut-8rRP8M,7050
-ignite/metrics/fbeta.py,sha256=Ioq_cscx5N8oF77B7QYkPMzGFlp0KtFtYXHgfJ1twO8,6376
+ignite/metrics/fbeta.py,sha256=2oDsRM7XXJ8LPVrn7iwLdRy75RLJELijmshtMQO3mJM,6870
 ignite/metrics/frequency.py,sha256=NW01rPgWnW1chVOSNAPCcPBu2CvjyXkoyFDAmjOK9A4,4037
 ignite/metrics/gpu_info.py,sha256=kcDIifr9js_P-32LddizEggvvL6eqFLYCHYeFDR4GL0,4301
 ignite/metrics/hsic.py,sha256=am-gor2mXY3H3u2vVNQGPJtkx_5W5JNZeukl2uYqajE,7099
-ignite/metrics/js_divergence.py,sha256=cng6KpR7I7znEEUcr434kJQ2uswgo5wbT5pAAcyVqdI,4838
+ignite/metrics/js_divergence.py,sha256=HAgj12JwL9bT33cCSAX7g4EKSfqFNNehkgwZfJuncfw,4828
 ignite/metrics/kl_divergence.py,sha256=FdC5BT-nd8nmYqT95Xozw-hW0hZC6dtTklkpJdwWJ6o,5152
 ignite/metrics/loss.py,sha256=mB-zYptymtcyIys0OlbVgUOAqL2WHT2dCPMFda-Klpo,4818
-ignite/metrics/maximum_mean_discrepancy.py,sha256=FVBt-IJNrZo_zV0vwpsjaShDhl4Z5nJWVLRjTV-GpoE,6452
+ignite/metrics/maximum_mean_discrepancy.py,sha256=AcrlYW6seQn3ZQKcnPIrLzYK2Ho0riGjuRsJmTNtCms,6444
 ignite/metrics/mean_absolute_error.py,sha256=gfbzoXNdyj9GCEzSxHXn0359TNNjnKBYshSnCBQk7i4,3695
 ignite/metrics/mean_average_precision.py,sha256=cXP9pYidQnAazGXBrhC80WoI4eK4lb3avNO5d70TLd4,19136
 ignite/metrics/mean_pairwise_distance.py,sha256=Ys6Rns6s-USS_tyP6Pa3bWZSI7f_hP5-lZM64UGJGjo,4104
 ignite/metrics/mean_squared_error.py,sha256=QdxXMYzxltfciMMRxxK5JhdlKXsdHe370EzwvIbwSmA,3679
-ignite/metrics/metric.py,sha256=IZucnzQL-HZVWa-Ppta8V1mw67XC-UP8cIA3DJwaP6o,34953
+ignite/metrics/metric.py,sha256=3dv3vy-YTgzC5aIZgLCW6pdiPzF5GqgKpqeSreEifA8,35103
 ignite/metrics/metric_group.py,sha256=UE7WrMbpKlO9_DPqxQdlmFAWveWoT1knKwRlHDl9YIU,2544
 ignite/metrics/metrics_lambda.py,sha256=NwKZ1J-KzFFbSw7YUaNJozdfKZLVqrkjQvFKT6ixnkg,7309
 ignite/metrics/multilabel_confusion_matrix.py,sha256=1pjLNPGTDJWAkN_BHdBPekcish6Ra0uRUeEbdj3Dm6Y,7377
 ignite/metrics/mutual_information.py,sha256=lu1ucVfkx01tGQfELyXzS9woCPOMVImFHfrbIXCvPe8,4692
 ignite/metrics/precision.py,sha256=xe8_e13cPMaC1Mfw-RTlmkag6pdcHCIbi70ASI1IahY,18622
-ignite/metrics/precision_recall_curve.py,sha256=2Gqv5B_Q5xP-mVlX1bmM5XZNnUTRdGG-MJZZvVcbQxc,6182
+ignite/metrics/precision_recall_curve.py,sha256=rcmG2W7dDuA_8fyekHNk4ronecewolMprW4rxUB8xsc,6228
 ignite/metrics/psnr.py,sha256=G994inwIczTWC5JfwECr0LSAtgquRGCs0283GylPR8c,5558
 ignite/metrics/recall.py,sha256=MaywS5E8ioaHZvTPGhQaYPQV-xDmptYuv8kDRe_-BEY,9867
-ignite/metrics/roc_auc.py,sha256=jcp6KpLFPr7FrSq6ePvx5lzweny0l0kFwzTS4PEBJ6M,9129
+ignite/metrics/roc_auc.py,sha256=NW_8GKX9W2tSLXn_d9G2A69gkbG62HWOc_YdyzBYO2s,9207
 ignite/metrics/root_mean_squared_error.py,sha256=yiOn5AQeg-RL2wM1MAng5Q98FHJc21chXU65tITT0Wo,2903
 ignite/metrics/running_average.py,sha256=vcC_LtsrJxEMea05TmBFzFqCK6nZd8hHavsfIlf2C6c,11333
-ignite/metrics/ssim.py,sha256=VhzEnpbpG2eQtkQKgmsVIf0IAavRckz52nIKEHcIGIM,10279
+ignite/metrics/ssim.py,sha256=_uJJdoHP4E4_sitcvFr9wTcoocK3iTxtSh_pA5J7Ss8,11766
 ignite/metrics/top_k_categorical_accuracy.py,sha256=pqsArVTSxnwt49S3lZFVqOkCXbzx-WPxfQnhtQ390RM,4706
 ignite/metrics/clustering/__init__.py,sha256=QljKwToBY-0fHblKbj1GsmP7rE5tlzHkrtw98MYEX44,233
 ignite/metrics/clustering/_base.py,sha256=lpQwtR54oTUrif7vQ7EE3ch8PJ91ECnzLov8z34gf5E,1526
-ignite/metrics/clustering/calinski_harabasz_score.py,sha256=i9DbAuFOFIgi7UVnHiiD_YHKnGgdItOyWqM-XrqLgwk,4654
+ignite/metrics/clustering/calinski_harabasz_score.py,sha256=jePNE7u72jh8RYL8Sew9rDn3BX6ydYq5Z2FPst4pqB0,4663
 ignite/metrics/clustering/davies_bouldin_score.py,sha256=VGC0jA3_gh9s4v3bm7Cw-5IV1ZUbqssYmU3s-rmnl_8,4646
-ignite/metrics/clustering/silhouette_score.py,sha256=MewWftWKR17OmkeBHLbzG_3RJs7XvSnfjal2D_3U62c,5151
+ignite/metrics/clustering/silhouette_score.py,sha256=Q9mMcyoR9woHwjxwrAPecFPhKA9bkptoKhhe5-mBfLA,5159
 ignite/metrics/gan/__init__.py,sha256=mBZQNI5uBd72iMyJs6GpbSBLEMm1-Lu1KtgmDAoH_4I,149
-ignite/metrics/gan/fid.py,sha256=QrpTNLLqw1mHPUU5_DfWpIapWH4AjlTXzFdF1IdT8So,10014
+ignite/metrics/gan/fid.py,sha256=rqITDukGd7CgQAMY8GRVPSLVrkF3MjjFR8bxE6M1kpg,10058
 ignite/metrics/gan/inception_score.py,sha256=78_qrECWb_KsbLbo1lvDnvFJ9FsWPsbUi1aKWyvp8kg,5601
 ignite/metrics/gan/utils.py,sha256=3nihbBrcM9MRcu6r0p3x5SgZQ5V4aag20ZppM7j_HiM,3993
 ignite/metrics/nlp/__init__.py,sha256=TiDKRhw7lhZeoL2Cn4s306cKIuBbXl2fizN1ZepMhwI,168
-ignite/metrics/nlp/bleu.py,sha256=O88d0-6gEm4ZztSWGkq6f2PPu3Icd8eqXUB6UJKYHmk,11424
-ignite/metrics/nlp/rouge.py,sha256=ybdmmne0Td3oWR5KX0jNSTTShsse5p_TyAFR6DerWOc,15364
-ignite/metrics/nlp/utils.py,sha256=o6zWzT8lugNAQVxJq-SEDFI35ve5-P-1TwyVu9wZCpM,2353
+ignite/metrics/nlp/bleu.py,sha256=NyQZ3CQB1xUnH_KWer5QtxkM_S_aiO3ok86UMxHaQ_w,11539
+ignite/metrics/nlp/rouge.py,sha256=pcIBCFBybJczYnPxuoLibwzNXYOMxf_JtyFiJkgo10A,15328
+ignite/metrics/nlp/utils.py,sha256=CA0MRMk9l97QockFYYhU6k0-hLhP3GwW36ONZ7TRqmc,2341
 ignite/metrics/regression/__init__.py,sha256=I594yB38ypWi9IDi9rrdshdXeBnSRcST09tnLRjN0yk,1472
-ignite/metrics/regression/_base.py,sha256=K0Xs3ZmtodhPB4GaAkFBwEgb6gbylDbCngBqsK_lbrs,2242
+ignite/metrics/regression/_base.py,sha256=5V6GkkaBYRuW9J3yDXucyTZp1XJ2uIG7F4w2XcBsd3w,2365
 ignite/metrics/regression/canberra_metric.py,sha256=HqQe-0lfwMMO5e_8hBIaAPS6PyKrIEtBKfRBNJV941Q,3077
 ignite/metrics/regression/fractional_absolute_error.py,sha256=ANQFQoadcg17ksTj_k0dY1M9E2OO8eboQCzjpRS-FNE,3259
-ignite/metrics/regression/fractional_bias.py,sha256=JKxhEaX9vINfwA1UJNIRNw3l3bonagHvj-ts8-iTDrs,3172
+ignite/metrics/regression/fractional_bias.py,sha256=IafPS6cJxhDL_OdJe2SsCtOkiwE-DVt7B2RLkY_SHpM,3178
 ignite/metrics/regression/geometric_mean_absolute_error.py,sha256=4KWSqONfKK1au4oLZDwWLJw3ENJE7rDGVNYRQYLxj1E,3195
 ignite/metrics/regression/geometric_mean_relative_absolute_error.py,sha256=vzvnt2sSqBHFaKRu0NqwzGHKwXpamhzv4YqJ4RN8CFA,4265
 ignite/metrics/regression/kendall_correlation.py,sha256=XVeqnhru0CQSXRz5wezbfWtdIqw9T20xGk_QOf8CztM,5280
@@ -153,8 +153,7 @@ ignite/metrics/regression/spearman_correlation.py,sha256=IzmN4WIe7C4cTUU3BOkBmaw
 ignite/metrics/regression/wave_hedges_distance.py,sha256=Ji_NRUgnZ3lJgi5fyNFLRjbHO648z4dBmqVDQU9ImKA,2792
 ignite/metrics/vision/__init__.py,sha256=lPBAEq1idc6Q17poFm1SjttE27irHF1-uNeiwrxnLrU,159
 ignite/metrics/vision/object_detection_average_precision_recall.py,sha256=PwdXVeGAF0sLIxUrvnnE7ZojpFNkZB5O6bYoopqc3M4,25024
-pytorch_ignite-0.6.0.dev20250324.dist-info/LICENSE,sha256=SwJvaRmy1ql-k9_nL4WnER4_ODTMF9fWoP9HXkoicgw,1527
-pytorch_ignite-0.6.0.dev20250324.dist-info/METADATA,sha256=EO1dJtV8huhPpGCp0c7WqHSoSVVDgWQZXy26uMD07WI,27997
-pytorch_ignite-0.6.0.dev20250324.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-pytorch_ignite-0.6.0.dev20250324.dist-info/top_level.txt,sha256=P2CnXR6kxvOX7ZMdd-9kVUTwLNz98t0sdjKeyvFBkR4,7
-pytorch_ignite-0.6.0.dev20250324.dist-info/RECORD,,
+pytorch_ignite-0.6.0.dev20251103.dist-info/METADATA,sha256=L4NDgHM4p5mtFYV7Fbr_OBleFmh2cfvlESJWn0QAyNo,27979
+pytorch_ignite-0.6.0.dev20251103.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+pytorch_ignite-0.6.0.dev20251103.dist-info/licenses/LICENSE,sha256=SwJvaRmy1ql-k9_nL4WnER4_ODTMF9fWoP9HXkoicgw,1527
+pytorch_ignite-0.6.0.dev20251103.dist-info/RECORD,,