kaiko-eva 0.1.6__py3-none-any.whl → 0.1.7__py3-none-any.whl

eva/vision/metrics/segmentation/_utils.py

@@ -6,7 +6,7 @@ import torch
 
 
 def apply_ignore_index(
-    preds: torch.Tensor, target: torch.Tensor, ignore_index: int, num_classes: int
+    preds: torch.Tensor, target: torch.Tensor, ignore_index: int
 ) -> Tuple[torch.Tensor, torch.Tensor]:
     """Applies the ignore index to the predictions and target tensors.
 
@@ -17,7 +17,6 @@ def apply_ignore_index(
         preds: The predictions tensor. Expected to be of shape `(N,C,...)`.
         target: The target tensor. Expected to be of shape `(N,C,...)`.
         ignore_index: The index to ignore.
-        num_classes: The number of classes.
 
     Returns:
         The modified predictions and target tensors of shape `(N,C-1,...)`.

eva/vision/metrics/segmentation/dice.py

@@ -0,0 +1,69 @@
+"""Generalized Dice Score metric for semantic segmentation."""
+
+from typing import Any, Literal
+
+import torch
+from torchmetrics import segmentation
+from torchmetrics.functional.segmentation.dice import _dice_score_update
+from typing_extensions import override
+
+from eva.vision.metrics.segmentation import _utils
+
+
+class DiceScore(segmentation.DiceScore):
+    """Defines the Generalized Dice Score.
+
+    It expands the `torchmetrics` class by including an `ignore_index`
+    functionality and converting tensors to one-hot format.
+    """
+
+    def __init__(
+        self,
+        num_classes: int,
+        include_background: bool = True,
+        average: Literal["micro", "macro", "weighted", "none"] | None = "micro",
+        ignore_index: int | None = None,
+        **kwargs: Any,
+    ) -> None:
+        """Initializes the metric.
+
+        Args:
+            num_classes: The number of classes in the segmentation problem.
+            include_background: Whether to include the background class in the computation
+            average: The method to average the dice score accross the different classes. Options are
+                `"micro"`, `"macro"`, `"weighted"`, `"none"` or `None`.
+            ignore_index: Integer specifying a target class to ignore. If given, this class
+                index does not contribute to the returned score, regardless of reduction method.
+            kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info.
+        """
+        super().__init__(
+            num_classes=num_classes
+            - (ignore_index is not None)
+            + (ignore_index == 0 and not include_background),
+            include_background=include_background,
+            average=average,
+            input_format="one-hot",
+            **kwargs,
+        )
+        self.orig_num_classes = num_classes
+        self.ignore_index = ignore_index
+
+    @override
+    def update(self, preds: torch.Tensor, target: torch.Tensor) -> None:
+        preds = _utils.index_to_one_hot(preds, num_classes=self.orig_num_classes)
+        target = _utils.index_to_one_hot(target, num_classes=self.orig_num_classes)
+        if self.ignore_index is not None:
+            preds, target = _utils.apply_ignore_index(preds, target, self.ignore_index)
+
+        # TODO: Replace _update by super.update() once the following issue is fixed:
+        # https://github.com/Lightning-AI/torchmetrics/issues/2847
+        self._update(preds.long(), target.long())
+        # super().update(preds=preds.long(), target=target.long())
+
+    def _update(self, preds: torch.Tensor, target: torch.Tensor) -> None:
+        numerator, denominator, support = _dice_score_update(
+            preds, target, self.num_classes, self.include_background, self.input_format  # type: ignore
+        )
+        self.numerator.append(numerator)
+        self.denominator.append(denominator)
+        self.support.append(support)

eva/vision/metrics/segmentation/generalized_dice.py

@@ -13,7 +13,7 @@ class GeneralizedDiceScore(segmentation.GeneralizedDiceScore):
     """Defines the Generalized Dice Score.
 
     It expands the `torchmetrics` class by including an `ignore_index`
-    functionality.
+    functionality and converting tensors to one-hot format.
     """
 
     def __init__(
@@ -55,7 +55,5 @@ class GeneralizedDiceScore(segmentation.GeneralizedDiceScore):
         preds = _utils.index_to_one_hot(preds, num_classes=self.orig_num_classes)
         target = _utils.index_to_one_hot(target, num_classes=self.orig_num_classes)
         if self.ignore_index is not None:
-            preds, target = _utils.apply_ignore_index(
-                preds, target, self.ignore_index, self.num_classes
-            )
+            preds, target = _utils.apply_ignore_index(preds, target, self.ignore_index)
         super().update(preds=preds.long(), target=target.long())

eva/vision/metrics/segmentation/mean_iou.py

@@ -13,7 +13,7 @@ class MeanIoU(segmentation.MeanIoU):
     """MeanIoU (mIOU) metric for semantic segmentation.
 
     It expands the `torchmetrics` class by including an `ignore_index`
-    functionality.
+    functionality and converting tensors to one-hot format.
     """
 
     def __init__(
@@ -36,10 +36,8 @@ class MeanIoU(segmentation.MeanIoU):
             kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info.
         """
         super().__init__(
-            num_classes=num_classes
-            - (ignore_index is not None)
-            + (ignore_index == 0 and not include_background),
-            include_background=include_background,
+            include_background=include_background or (ignore_index == 0),
+            num_classes=num_classes - (ignore_index is not None),
             per_class=per_class,
             **kwargs,
         )
@@ -51,7 +49,5 @@ class MeanIoU(segmentation.MeanIoU):
         preds = _utils.index_to_one_hot(preds, num_classes=self.orig_num_classes)
         target = _utils.index_to_one_hot(target, num_classes=self.orig_num_classes)
         if self.ignore_index is not None:
-            preds, target = _utils.apply_ignore_index(
-                preds, target, self.ignore_index, self.num_classes
-            )
+            preds, target = _utils.apply_ignore_index(preds, target, self.ignore_index)
         super().update(preds=preds.long(), target=target.long())

eva/vision/metrics/segmentation/monai_dice.py

@@ -0,0 +1,57 @@
+"""Wrapper for dice score metric from MONAI."""
+
+from monai.metrics.meandice import DiceMetric
+from typing_extensions import override
+
+from eva.vision.metrics import wrappers
+from eva.vision.metrics.segmentation import _utils
+
+
+class MonaiDiceScore(wrappers.MonaiMetricWrapper):
+    """Wrapper to make MONAI's `DiceMetric` compatible with `torchmetrics`."""
+
+    def __init__(
+        self,
+        num_classes: int,
+        include_background: bool = True,
+        reduction: str = "mean",
+        ignore_index: int | None = None,
+        **kwargs,
+    ):
+        """Initializes metric.
+
+        Args:
+            num_classes: The number of classes in the dataset.
+            include_background: Whether to include the background class in the computation.
+            reduction: The method to reduce the dice score. Options are `"mean"`, `"sum"`, `"none"`.
+            ignore_index: Integer specifying a target class to ignore. If given, this class
+                index does not contribute to the returned score.
+            kwargs: Additional keyword arguments for instantiating monai's `DiceMetric` class.
+        """
+        super().__init__(
+            DiceMetric(
+                include_background=include_background or (ignore_index == 0),
+                reduction=reduction,
+                num_classes=num_classes - (ignore_index is not None),
+                **kwargs,
+            )
+        )
+
+        self.reduction = reduction
+        self.orig_num_classes = num_classes
+        self.ignore_index = ignore_index
+
+    @override
+    def update(self, preds, target):
+        preds = _utils.index_to_one_hot(preds, num_classes=self.orig_num_classes)
+        target = _utils.index_to_one_hot(target, num_classes=self.orig_num_classes)
+        if self.ignore_index is not None:
+            preds, target = _utils.apply_ignore_index(preds, target, self.ignore_index)
+        return super().update(preds, target)
+
+    @override
+    def compute(self):
+        result = super().compute()
+        if self.reduction == "none" and len(result) > 1:
+            result = result.nanmean(dim=0)
+        return result

eva/vision/metrics/wrappers/__init__.py

@@ -0,0 +1,5 @@
+"""Metrics wrappers API."""
+
+from eva.vision.metrics.wrappers.monai import MonaiMetricWrapper
+
+__all__ = ["MonaiMetricWrapper"]

eva/vision/metrics/wrappers/monai.py

@@ -0,0 +1,32 @@
+"""Monai metrics wrappers."""
+
+import torch
+import torchmetrics
+from monai.metrics.metric import CumulativeIterationMetric
+from typing_extensions import override
+
+
+class MonaiMetricWrapper(torchmetrics.Metric):
+    """Wrapper class to make MONAI metrics compatible with `torchmetrics`."""
+
+    def __init__(self, monai_metric: CumulativeIterationMetric):
+        """Initializes the monai metric wrapper.
+
+        Args:
+            monai_metric: The MONAI metric to wrap.
+        """
+        super().__init__()
+        self._monai_metric = monai_metric
+
+    @override
+    def update(self, preds: torch.Tensor, target: torch.Tensor) -> None:
+        self._monai_metric(preds, target)
+
+    @override
+    def compute(self) -> torch.Tensor:
+        return self._monai_metric.aggregate()
+
+    @override
+    def reset(self) -> None:
+        super().reset()
+        self._monai_metric.reset()

eva/vision/models/modules/semantic_segmentation.py

@@ -12,7 +12,7 @@ from typing_extensions import override
 from eva.core.metrics import structs as metrics_lib
 from eva.core.models.modules import module
 from eva.core.models.modules.typings import INPUT_BATCH, INPUT_TENSOR_BATCH
-from eva.core.models.modules.utils import batch_postprocess, grad
+from eva.core.models.modules.utils import batch_postprocess, grad, submodule_state_dict
 from eva.core.utils import parser
 from eva.vision.models.networks import decoders
 from eva.vision.models.networks.decoders.segmentation.typings import DecoderInputs
@@ -31,6 +31,7 @@ class SemanticSegmentationModule(module.ModelModule):
         lr_scheduler: LRSchedulerCallable = lr_scheduler.ConstantLR,
         metrics: metrics_lib.MetricsSchema | None = None,
         postprocess: batch_postprocess.BatchPostProcess | None = None,
+        save_decoder_only: bool = True,
     ) -> None:
         """Initializes the neural net head module.
 
@@ -49,6 +50,8 @@ class SemanticSegmentationModule(module.ModelModule):
             postprocess: A list of helper functions to apply after the
                 loss and before the metrics calculation to the model
                 predictions and targets.
+            save_decoder_only: Whether to save only the decoder during checkpointing. If False,
+                will also save the encoder (not recommended when frozen).
         """
         super().__init__(metrics=metrics, postprocess=postprocess)
 
@@ -58,6 +61,7 @@ class SemanticSegmentationModule(module.ModelModule):
         self.lr_multiplier_encoder = lr_multiplier_encoder
         self.optimizer = optimizer
         self.lr_scheduler = lr_scheduler
+        self.save_decoder_only = save_decoder_only
 
     @override
     def configure_model(self) -> None:
@@ -83,6 +87,20 @@ class SemanticSegmentationModule(module.ModelModule):
         lr_scheduler = self.lr_scheduler(optimizer)
         return {"optimizer": optimizer, "lr_scheduler": lr_scheduler}
 
+    @override
+    def on_save_checkpoint(self, checkpoint: Dict[str, Any]) -> None:
+        if self.save_decoder_only:
+            checkpoint["state_dict"] = submodule_state_dict(checkpoint["state_dict"], "decoder")
+        super().on_save_checkpoint(checkpoint)
+
+    @override
+    def on_load_checkpoint(self, checkpoint: Dict[str, Any]) -> None:
+        if self.save_decoder_only and self.encoder is not None:
+            checkpoint["state_dict"].update(
+                {f"encoder.{k}": v for k, v in self.encoder.state_dict().items()}  # type: ignore
+            )
+        super().on_load_checkpoint(checkpoint)
+
     @override
     def forward(
         self,

eva/vision/models/networks/backbones/__init__.py

@@ -1,6 +1,6 @@
 """Vision Model Backbones API."""
 
-from eva.vision.models.networks.backbones import pathology, timm, universal
+from eva.vision.models.networks.backbones import pathology, timm, torchhub, universal
 from eva.vision.models.networks.backbones.registry import BackboneModelRegistry, register_model
 
-__all__ = ["pathology", "timm", "universal", "BackboneModelRegistry", "register_model"]
+__all__ = ["pathology", "timm", "torchhub", "universal", "BackboneModelRegistry", "register_model"]

eva/vision/models/networks/backbones/torchhub/__init__.py

@@ -0,0 +1,5 @@
+"""torch.hub backbones API."""
+
+from eva.vision.models.networks.backbones.torchhub.backbones import torch_hub_model
+
+__all__ = ["torch_hub_model"]

eva/vision/models/networks/backbones/torchhub/backbones.py

@@ -0,0 +1,61 @@
+"""torch.hub backbones."""
+
+import functools
+from typing import Tuple
+
+import torch
+from loguru import logger
+from torch import nn
+
+from eva.core.models import wrappers
+from eva.vision.models.networks.backbones.registry import BackboneModelRegistry
+
+HUB_REPOS = ["facebookresearch/dinov2:main", "kaiko-ai/towards_large_pathology_fms"]
+"""List of torch.hub repositories for which to add the models to the registry."""
+
+
+def torch_hub_model(
+    model_name: str,
+    repo_or_dir: str,
+    checkpoint_path: str | None = None,
+    pretrained: bool = False,
+    out_indices: int | Tuple[int, ...] | None = None,
+    **kwargs,
+) -> nn.Module:
+    """Initializes any ViT model from torch.hub with weights from a specified checkpoint.
+
+    Args:
+        model_name: The name of the model to load.
+        repo_or_dir: The torch.hub repository or local directory to load the model from.
+        checkpoint_path: The path to the checkpoint file.
+        pretrained: If set to `True`, load pretrained model weights if available.
+        out_indices: Whether and which multi-level patch embeddings to return.
+        **kwargs: Additional arguments to pass to the model
+
+    Returns:
+        The VIT model instance.
+    """
+    logger.info(
+        f"Loading torch.hub model {model_name} from {repo_or_dir}"
+        + (f"using checkpoint {checkpoint_path}" if checkpoint_path else "")
+    )
+
+    return wrappers.TorchHubModel(
+        model_name=model_name,
+        repo_or_dir=repo_or_dir,
+        pretrained=pretrained,
+        checkpoint_path=checkpoint_path or "",
+        out_indices=out_indices,
+        model_kwargs=kwargs,
+    )
+
+
+BackboneModelRegistry._registry.update(
+    {
+        f"torchhub/{repo}:{model_name}": functools.partial(
+            torch_hub_model, model_name=model_name, repo_or_dir=repo
+        )
+        for repo in HUB_REPOS
+        for model_name in torch.hub.list(repo, verbose=False)
+    }
+)

eva/vision/models/networks/decoders/segmentation/decoder2d.py

@@ -52,7 +52,7 @@ class Decoder2D(base.Decoder):
         """
         if isinstance(features, torch.Tensor):
             features = [features]
-        if not isinstance(features, list) or features[0].ndim != 4:
+        if not isinstance(features, (list, tuple)) or features[0].ndim != 4:
             raise ValueError(
                 "Input features should be a list of four (4) dimensional inputs of "
                 "shape (batch_size, hidden_size, n_patches_height, n_patches_width)."

eva/vision/models/wrappers/__init__.py

@@ -3,4 +3,4 @@
 from eva.vision.models.wrappers.from_registry import ModelFromRegistry
 from eva.vision.models.wrappers.from_timm import TimmModel
 
-__all__ = ["TimmModel", "ModelFromRegistry"]
+__all__ = ["ModelFromRegistry", "TimmModel"]

{kaiko_eva-0.1.6.dist-info → kaiko_eva-0.1.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: kaiko-eva
-Version: 0.1.6
+Version: 0.1.7
 Summary: Evaluation Framework for oncology foundation models.
 Keywords: machine-learning,evaluation-framework,oncology,foundation-models
 Author-Email: Ioannis Gatopoulos <ioannis@kaiko.ai>, =?utf-8?q?Nicolas_K=C3=A4nzig?= <nicolas@kaiko.ai>, Roman Moser <roman@kaiko.ai>
@@ -227,7 +227,7 @@ Requires-Dist: onnxruntime>=1.15.1
 Requires-Dist: onnx>=1.16.0
 Requires-Dist: toolz>=0.12.1
 Requires-Dist: rich>=13.7.1
-Requires-Dist: torchmetrics>=1.4.2
+Requires-Dist: torchmetrics>=1.6.0
 Provides-Extra: vision
 Requires-Dist: h5py>=3.10.0; extra == "vision"
 Requires-Dist: nibabel>=4.0.1; extra == "vision"

{kaiko_eva-0.1.6.dist-info → kaiko_eva-0.1.7.dist-info}/RECORD

@@ -17,20 +17,24 @@ eva/core/cli/logo.py,sha256=x6-vGWI0s9gza-xxQrBDi2wneb2wFU_mQGHgpAiq2MQ,786
 eva/core/cli/setup.py,sha256=kR-7l4X5Hu8kSLoQZGYGIeLXtn9S_EU52dauDy6fm0w,2663
 eva/core/data/__init__.py,sha256=yG3BeOWhp1EjVYMFqx8M_TBWFDyfIwwksQGQmMdSPaI,340
 eva/core/data/dataloaders/__init__.py,sha256=fbNClVZ8J3QoGi4qiPq635ig1j9GdI7six3RhfwDbjY,110
-eva/core/data/dataloaders/dataloader.py,sha256=n2NNlWKurEC1TVis347l6YmPgNOGTaoxwIyoClyPt7s,2383
+eva/core/data/dataloaders/dataloader.py,sha256=Ek_OqlOuQSiipFjBJ39DFyWvW3CdqAB2bufOcEs0ChU,2525
 eva/core/data/datamodules/__init__.py,sha256=qZchYbgxo9lxYnGoqdk0C6MfS2IbF0WItO0kCdP9Mqc,229
 eva/core/data/datamodules/call.py,sha256=jjj9w3UXYuQB-qyCcw1EZpRJW10OC1I3dvgvsuQWLck,940
-eva/core/data/datamodules/datamodule.py,sha256=dclC2YJAXUGEUpV9ZRWQS43-ksFIPgVeFudsyrj9kdc,3878
-eva/core/data/datamodules/schemas.py,sha256=EXnUPNd9Pj3RjnxJIzAcC2qp6TtBSvPDx28fV_ovWAA,1869
-eva/core/data/datasets/__init__.py,sha256=YfbHEVy9AXdvPAobZ8btV-mVEaWAVDr7feK8xydLW2w,391
-eva/core/data/datasets/base.py,sha256=NLZlxznB4SCYNf070OhfNJztaOpqwQWemwpGkFv_CA0,2005
+eva/core/data/datamodules/datamodule.py,sha256=_pK59oXDe53oDkmv6eoJUvfl44WlFkrbC8KXSRMs_20,5533
+eva/core/data/datamodules/schemas.py,sha256=rzcf3uow6T6slVSwxEGDVmpi3QUvkiDoT_gCF3aMAEE,2262
+eva/core/data/datasets/__init__.py,sha256=jWPxT3gjQjwS6HqVZAb7KhMgzgklPgHeH51iPxDh_Tg,493
+eva/core/data/datasets/base.py,sha256=BLzlRFuByhrGmI7NFwn7-Tw0vpSYSRhl2Y65iX4KaMw,2526
 eva/core/data/datasets/classification/__init__.py,sha256=wJ2jD9YODftt-dMcMf0TbCjJt47qXYBKkD4-XXajvRQ,340
 eva/core/data/datasets/classification/embeddings.py,sha256=bgBVQyGxlxVCvGjmwNB52E360QwzrhGZQ44rPNFR4k8,1110
 eva/core/data/datasets/classification/multi_embeddings.py,sha256=j_o0MH2gwn_y3rNFXEUzNg6WErlG3Rq_vn5Og1Yk7J0,4603
 eva/core/data/datasets/dataset.py,sha256=tA6Wd_7vqOE9GsukSWrgN9zaZKtKCHaE58SqIfWxWdg,124
 eva/core/data/datasets/embeddings.py,sha256=zNEO8KxqiOopcN_lTjwtEAm9xbnYDSjOE8X2-iZQIhU,5545
-eva/core/data/samplers/__init__.py,sha256=WikBo1DemCx6o2vFfNwSwODlmCT2zWUXtCNwiWCVAFE,100
-eva/core/data/samplers/sampler.py,sha256=vrrXERWC67fjmTk_uwD7s9-8-rdhvnx7OlSipHE6sdY,119
+eva/core/data/datasets/typings.py,sha256=KSmckjsU64pGV-8uSLkD1HmvPKYlyypngiRx9yy4RDs,383
+eva/core/data/samplers/__init__.py,sha256=rRrKtg4l6YoziD3M0MkctQvX1NdRxaQa5sm6RHH_jXc,315
+eva/core/data/samplers/classification/__init__.py,sha256=gvv7BH4lG9JlkMaTOnaL0f4k1ghiVBgrH64bh1-rreQ,147
+eva/core/data/samplers/classification/balanced.py,sha256=YE6InKu12Jnu7AObi_gjKLzeHAFlQsbJVrggeA8X4DU,3517
+eva/core/data/samplers/random.py,sha256=znl0Z9a-X-3attP-EH9jwwo83n40FXW_JzOLNZAml_c,1252
+eva/core/data/samplers/sampler.py,sha256=0DOLUzFoweqEubuO1A4bZBRU0AWFoWGWrO3pawRT-eI,877
 eva/core/data/splitting/__init__.py,sha256=VQJ8lfakbv6y2kAk3VDtITAvh7kcZo3H1JwJBc5jT08,198
 eva/core/data/splitting/random.py,sha256=r6iy7j34seRTlyB79_Xy7m6lsKRi8ZM9X5Ln1b-SBjg,1453
 eva/core/data/splitting/stratified.py,sha256=dliRHgEyZTOzpJrR8FzaRaAnjUxu_VsZUuy_4MPQjlY,2265
@@ -63,35 +67,37 @@ eva/core/metrics/defaults/classification/multiclass.py,sha256=8Aesy_rKtp4KxfXJtD
 eva/core/metrics/structs/__init__.py,sha256=cvn7E4k5vJmpwJj_zezmtZa_Nl_RddDM1G-MO8TP0po,422
 eva/core/metrics/structs/collection.py,sha256=bNfCekHN8pzD49-YTqVxrmxFtiQfNxnv-RwkxCL6rbc,149
 eva/core/metrics/structs/metric.py,sha256=zdnE0ZVTSYAMl7rW_OL6e1XiZDvLTirYqV0lgJCleXY,109
-eva/core/metrics/structs/module.py,sha256=qAyk9uSGTFdvSg6ukl2c-OC-FdaCCsUf3Lh8UbUD-r8,3619
+eva/core/metrics/structs/module.py,sha256=pHpIAt5HQDoYWvyFXxYTZleTKMW1iaTCgwAktygjzDw,4681
 eva/core/metrics/structs/schemas.py,sha256=ZaSrx0j_NfIwT7joMUD1LyrKdAXTLaeSzWYTHDsc6h0,1641
 eva/core/metrics/structs/typings.py,sha256=qJd-FiD2IhJgBeo8FyP0vpVUIH4RKb1k6zYvHtjUA04,388
-eva/core/models/__init__.py,sha256=bQSpfQJKuDMWosjcMhP7t5jdOSV6OyxdxTOIW9w1woE,369
+eva/core/models/__init__.py,sha256=T6Fo886LxMj-Y58_ylzkPkFSnFR2aISiMIbuO_weC4s,430
 eva/core/models/modules/__init__.py,sha256=QJWJ42BceXZBzDGgk5FHBcCaRrB9egTFKVF6gDsBYfM,255
-eva/core/models/modules/head.py,sha256=iHrEOjYfshFI6OdXxJJTZyfCoUs2fimitINNcB6ENsc,4321
+eva/core/models/modules/head.py,sha256=Wza8IFAXFl_DwVnNqYKproI06iS-oIuUlGjRE6jAKXw,5185
 eva/core/models/modules/inference.py,sha256=ih-0Rr2oNf2N6maiXPOW7XH5KVwUT1_MOxnJKOhJ1uQ,978
 eva/core/models/modules/module.py,sha256=LtjYxTZb7UY0owonmt_yQ5EySw3sX-xD9HLN2io8EK4,6697
 eva/core/models/modules/typings.py,sha256=yFMJCE4Nrfd8VEXU1zk8p6Sz5M7UslwitYPVC2OPLSY,776
-eva/core/models/modules/utils/__init__.py,sha256=pnbxlEhT87JimWNr-NSNCv7VNR-IyDi_A9qRWmvlzwQ,227
+eva/core/models/modules/utils/__init__.py,sha256=ScLCHwQfzlg_UsHVi5sf_SavUkh9secwtRn_umC_qA8,325
 eva/core/models/modules/utils/batch_postprocess.py,sha256=RwnDcjJy3uvVirpgx_80Q2CUYKfJKipVwjyX7AF2CKw,3088
+eva/core/models/modules/utils/checkpoint.py,sha256=Zp42rtmjgUC4VUMwFyG5aar-E0Hc5i7qUsxkV7AVKkE,700
 eva/core/models/modules/utils/grad.py,sha256=bl8qb8g4Nhg1KAGfbEV_9HTKkoT0azRwfs9KGX9swGs,706
 eva/core/models/networks/__init__.py,sha256=yqx6UmG1Eg3vb1O_tnK_axnJWabEl9ULkDWiPN440Xc,85
 eva/core/models/networks/mlp.py,sha256=thk-x4pviE3fCaMW9k3I2Oe5_DxfC-CqUrtolvVdXug,2418
 eva/core/models/transforms/__init__.py,sha256=oYL3gNUUKZFViTu6GT1jVE2Kv1xFYPuyiYp-sErtVVg,257
 eva/core/models/transforms/extract_cls_features.py,sha256=tFRd4H-eGFIGCfZt6wuZGibDmAoNXKSsn15bBw0IDdc,1482
 eva/core/models/transforms/extract_patch_features.py,sha256=k50jTLPWxbfvciH9QZSzTAGqWwDSVpXAteme_Qg2d6E,2202
-eva/core/models/wrappers/__init__.py,sha256=P-ipr4NtKqPU6ubAjKLGxFf1Qt2yDSNtgS2Xz5sBahQ,364
+eva/core/models/wrappers/__init__.py,sha256=jaiANQdbO-IPgH8U-Y0ftFsuuCAM5i5KuYRHauKw5k8,450
 eva/core/models/wrappers/_utils.py,sha256=HXUyGcILaa8GK31ViIHCKRU4f9kbjAPYQmhvN2N7jSc,957
 eva/core/models/wrappers/base.py,sha256=xKMUSXk93wI67p_wmh7jujK-bxvIefO1noYaAJN_5Ak,1359
 eva/core/models/wrappers/from_function.py,sha256=_vKBwtfilCNCnOaJTofE6l5bM2K3qJ8GyBT-0CM5FXY,1831
+eva/core/models/wrappers/from_torchhub.py,sha256=OAImGKRG4pfDXHsoriykC_iiO8QvK3nAWnQCE0mIGuk,3285
 eva/core/models/wrappers/huggingface.py,sha256=5CoNieivdjwvoawo7dZtWfYZkW-Mey1j0EjazuxDaqU,1302
 eva/core/models/wrappers/onnx.py,sha256=-iV-IlczTvTTEQuJycZeSVWdSl2kVJXc1eeRLgQQZ7Q,1834
 eva/core/trainers/__init__.py,sha256=jhsKJF7HAae7EOiG3gKIAHH_h3dZlTE2JRcCHJmOzJc,208
 eva/core/trainers/_logging.py,sha256=gi4FqPy2GuVmh0WZY6mYwF7zMPvnoFA050B0XdCP6PU,2571
 eva/core/trainers/_recorder.py,sha256=y6i5hfXftWjeV3eQHmMjUOkWumnZ2QNv_u275LLmvPA,7702
 eva/core/trainers/_utils.py,sha256=M3h8lVhUmkeSiEXpX9hRdMvThGFCnTP15gv-hd1CZkc,321
-eva/core/trainers/functional.py,sha256=NPxFCtU5KgquVowjeXAf_xj4-Thj7ZxN9F3sHRDrDUA,4388
-eva/core/trainers/trainer.py,sha256=Vw_KhTyh-3YV5qo_XHxz9oy-v2PxrgoOWMeYi8-41R0,3949
+eva/core/trainers/functional.py,sha256=7OK2BNfX4_amHsyucr1ZNQRG3RgVKoagzd1zNN4nU3U,4472
+eva/core/trainers/trainer.py,sha256=HJNSfTG0k4j2ShqZzuUUSxnSu8NrwJ4karhvAto2Zn0,4229
 eva/core/utils/__init__.py,sha256=cndVBvtYxEW7hykH39GCNVI86zkXNn8Lw2A0sUJHS04,237
 eva/core/utils/clone.py,sha256=qcThZOuAs1cs0uV3BL5eKeM2VIBjuRPBe1t-NiUFM5Y,569
 eva/core/utils/io/__init__.py,sha256=Py03AmoxhmTHkro6CzNps27uXKkXPzdA18mG97xHhWI,172
@@ -123,17 +129,18 @@ eva/vision/data/datasets/classification/panda.py,sha256=BU_gDoX3ZSDUugwaO2n0XSZh
 eva/vision/data/datasets/classification/patch_camelyon.py,sha256=fElKteZKx4M6AjylnhhgNH1jewHegWc1K8h4FFKp0gE,7171
 eva/vision/data/datasets/classification/wsi.py,sha256=x3mQ8iwyiSdfQOjJuV7_cd8-LRjjhY9tjtzuD8O87Lg,4099
 eva/vision/data/datasets/segmentation/__init__.py,sha256=hGNr7BM_StxvmlOKWWfHp615qgsrB6BB3qMOiYhE0Og,791
+eva/vision/data/datasets/segmentation/_total_segmentator.py,sha256=DTaQaAisY7j1h0-zYk1_81Sr4b3D9PTMieYX0PMPtIc,3127
 eva/vision/data/datasets/segmentation/_utils.py,sha256=ps1qpuEkPgvwUw6H-KKaLaYqDBGmN7dNGk3bnS1l6sI,1261
 eva/vision/data/datasets/segmentation/base.py,sha256=11IMODMB7KJ8Bs5p7MyOsBXCyPFJXfYcDLAIMitUwEk,3023
 eva/vision/data/datasets/segmentation/bcss.py,sha256=NHjHd1tgIfIw6TxsZTGb63iMEwXFbWX_JAwRT5WVsj4,8274
-eva/vision/data/datasets/segmentation/consep.py,sha256=dCD8VsZSvI3-RbHHHAwGWfsNwOJCPkzHpVtzrcuAEVo,5871
+eva/vision/data/datasets/segmentation/consep.py,sha256=Pw3LvVIK2scj_ys7rVNRb9B8snP8HlDIAbaI3v6ObQk,6056
 eva/vision/data/datasets/segmentation/embeddings.py,sha256=0KaadzPxN6OrKNnFu3YsGBFkG6XqqvkOZYUhERPwL4A,1220
-eva/vision/data/datasets/segmentation/lits.py,sha256=_R5AGFX8jVPwK3UKaYQfIRLBpM5ZmDg6KRziisUDYps,7175
+eva/vision/data/datasets/segmentation/lits.py,sha256=cBRU5lkiTMAi_ZwyDQUN3ODyXUlLtuMWFLPDajcZnOo,7194
 eva/vision/data/datasets/segmentation/lits_balanced.py,sha256=s5kPfqB41Vkcm5Jh34mLAO0NweMSIlV2fMXJsRjJsF8,3384
 eva/vision/data/datasets/segmentation/monusac.py,sha256=OTWHAD1b48WeT6phVf466w_nJUOGdBCGKWiWw68PAdw,8423
-eva/vision/data/datasets/segmentation/total_segmentator_2d.py,sha256=oyb38pAV8GN5Ph1-NSVooFhNP1TfOuXSUjUIAf6rdiY,14376
+eva/vision/data/datasets/segmentation/total_segmentator_2d.py,sha256=A6A_lXmGDfV_9Mcp9KSgN6K8Q0T8XXjv6lT4I7iLUcw,16833
 eva/vision/data/datasets/structs.py,sha256=RaTDW-B36PumcR5gymhCiX-r8GiKqIFcjqoEEjjFyUE,389
-eva/vision/data/datasets/vision.py,sha256=hKKFMb65UJQzOyYm8FTGkOGBOinMRu7R8sOFMbCmQX4,1100
+eva/vision/data/datasets/vision.py,sha256=RHcBBNTd5u1OB6l2iA5V8pv8kjZsTehi9At7J-FVqr4,657
 eva/vision/data/datasets/wsi.py,sha256=-rypkcd6CPBM_oPuLszUx9q4zSPzeO1H6JKqvOtLlHw,8282
 eva/vision/data/transforms/__init__.py,sha256=WeFii6JwB0CiOOGLR3tkgAoKgRdmOf2lm0Dadixn8OI,260
 eva/vision/data/transforms/common/__init__.py,sha256=6tvxUgb8wfhgvqejMVulwqssHTJLF7f4_vpf44kxgxY,234
@@ -160,22 +167,26 @@ eva/vision/data/wsi/patching/samplers/foreground_grid.py,sha256=EhXkr5EFz2-RXEis
 eva/vision/data/wsi/patching/samplers/grid.py,sha256=dImrMSyCL3E_j5KRqpVJUWTe-mrJpfttg1Z9rbm3j0k,1363
 eva/vision/data/wsi/patching/samplers/random.py,sha256=0clmwCZ47bnTaSFke7jtjsmrFoY1ID2LjoiaE52dC3o,1228
 eva/vision/losses/__init__.py,sha256=htafabZgVcqbJjPURwsmGJ7AT6hIXc1-9SEuuaGU9SA,121
-eva/vision/losses/dice.py,sha256=qIMxtQlBbzES3hJ7x7pq0bd0GvIFpfGNL3KnMi0vRds,3669
-eva/vision/metrics/__init__.py,sha256=NtbcCrAUhVMMxSygTnbvNceJZBUzUD2tZp4nDmFjG3w,360
+eva/vision/losses/dice.py,sha256=8CCtEFB_zPok3WGLMHJX__K5IEm0HmL5WipQo8ZhQwc,3556
+eva/vision/metrics/__init__.py,sha256=zXOc1Idgfk86CGE5yBHn3B22iD5tRyfl4H-kTSB2dCQ,528
 eva/vision/metrics/defaults/__init__.py,sha256=ncQ9uH5q5SpfalyPX6dINPRLk34HLw6z9u8ny_HHbFQ,174
 eva/vision/metrics/defaults/segmentation/__init__.py,sha256=ve6dwyfhJGYBYKS6l6OySCBs32JnEBFnvhAyNvj-Uqo,191
-eva/vision/metrics/defaults/segmentation/multiclass.py,sha256=Qk4-OC0oujg99MEwRIjif-fS2mbAiWSFTZVRq4cY_Vo,1758
+eva/vision/metrics/defaults/segmentation/multiclass.py,sha256=MUBp-PIyiJB2VVV_NintRrP7Ha2lJ75_3xvqSdeDYwE,2855
 eva/vision/metrics/segmentation/BUILD,sha256=Nf7BYWWe1USoFEIsIiEVZ8sa05J5FPkMJ-UIMDLrU8o,17
-eva/vision/metrics/segmentation/__init__.py,sha256=lfuyjuo2XjS7_dvaL-8e_LhLsyK2mc_mNGwyGqcb5X4,234
-eva/vision/metrics/segmentation/_utils.py,sha256=ebxTqymtxZ0iwMiH2snQHV_NVfDhZUSBlBS9AShWu_8,2464
-eva/vision/metrics/segmentation/generalized_dice.py,sha256=FqFzo7YWBwSlihmlgQg-O_ld1ZBQma0YTXk5XZ7faZM,2443
-eva/vision/metrics/segmentation/mean_iou.py,sha256=xR3wQOHT77SNKTRRPdDaWpJ88qgk9PIBT5n2lnKTUfM,2161
+eva/vision/metrics/segmentation/__init__.py,sha256=7iz3fFNd-iBuNyxdeSfsgp6D7oZtmPsbyA0ZKRzzRCw,402
+eva/vision/metrics/segmentation/_utils.py,sha256=_ubv2sP1-f_dLKy8Y4wLkj5ed56fAFLURfv1shQWVcs,2402
+eva/vision/metrics/segmentation/dice.py,sha256=H_U6XSZcieX0xb6aptxxW1s-Jshs8Lp4P1SAwjdwntM,2905
+eva/vision/metrics/segmentation/generalized_dice.py,sha256=T57An-lBVefnlv6dIWVRNghFxy0e0K470xwla0TbCSk,2436
+eva/vision/metrics/segmentation/mean_iou.py,sha256=2PjqTa_VAtnW4nxHzT93uBKgnml7INU-wt_jR68RM54,2104
+eva/vision/metrics/segmentation/monai_dice.py,sha256=febnvA2gtTyydLZMwjQBS1zq2NjZcsXf0EcV0eRn8Aw,2117
+eva/vision/metrics/wrappers/__init__.py,sha256=V4z3hradMa6CQgTkk1bc2cbZzCgcoIYw7-hufMK3D_4,128
+eva/vision/metrics/wrappers/monai.py,sha256=FNa1yHN2U3vO6BGqS0BFm8uJAL6DCzSE4XOFCV4aBjg,885
 eva/vision/models/__init__.py,sha256=a-P6JL73A3miHQnqgqUz07XtVmQB_o4DqPImk5rEATo,275
 eva/vision/models/modules/__init__.py,sha256=vaM_V6OF2s0lYjralP8dzv8mAtv_xIMZItfXgz0NZg8,156
-eva/vision/models/modules/semantic_segmentation.py,sha256=i1hYgWnVLf7RTzt_ZrO76bQYNOKmyjt_Hl8AFQqFhAk,6400
+eva/vision/models/modules/semantic_segmentation.py,sha256=PSeqm5h6YgbzQ0jA9lUexGYUE3ehfWx-LH1NgZ7cGhw,7300
 eva/vision/models/networks/__init__.py,sha256=j43IurizNlAyKPH2jwDHaeq49L2QvwbHWqUaptA1mG4,100
 eva/vision/models/networks/abmil.py,sha256=N1eH4fn1nXmgXurSQyQIxxonv7nsqeeuPWaQSHeltfs,6796
-eva/vision/models/networks/backbones/__init__.py,sha256=LsMx92eEoCQ5aNVFp7mHjrD-9ZeNawMiK6zZSYzl_PU,296
+eva/vision/models/networks/backbones/__init__.py,sha256=CvK0sHKufUq4chwX-p2cFVBZFReMuwmeHFTG5LUA6CM,318
 eva/vision/models/networks/backbones/_utils.py,sha256=V7xeod4mElEuuO1TRW0xJE051cUyS1Saraw3-KcK1Mw,1667
 eva/vision/models/networks/backbones/pathology/__init__.py,sha256=goR59h8bfzd-Wa3rxPPdaSlAOH_df8SHBkTSKi08TS8,1147
 eva/vision/models/networks/backbones/pathology/bioptimus.py,sha256=wUSKjYgxcRV3FRHGaPwF1uRAQcGO0rHNHGmK1QDJXk4,991
@@ -189,18 +200,20 @@ eva/vision/models/networks/backbones/pathology/paige.py,sha256=MjOLgdEKk8tdAIpCi
 eva/vision/models/networks/backbones/registry.py,sha256=anjILtEHHB6Ltwiw22h1bsgWtIjh_l5_fkPh87K7-d0,1631
 eva/vision/models/networks/backbones/timm/__init__.py,sha256=cZH3av9gIZcvEVD0rwKsI-MEq7zPqaW4dQ0E05CksvQ,128
 eva/vision/models/networks/backbones/timm/backbones.py,sha256=fCTiwqU6NhQ-ccAMzmpPDddXkFzRAB3mw4lcQ9um_PU,1646
+eva/vision/models/networks/backbones/torchhub/__init__.py,sha256=zBLJBvkwKJ1jD7M3Wt5BE6Cx-R8G2YRoyPG7p2V-3nQ,147
+eva/vision/models/networks/backbones/torchhub/backbones.py,sha256=hgCCoP8AdRSsli0w9a_PRNB-UR36-SLLhBIW0BFrkdE,1911
 eva/vision/models/networks/backbones/universal/__init__.py,sha256=MAlkALSJ2_w6spSbB7NmKlL0Jsk1YKEycatdI0xO0_I,252
 eva/vision/models/networks/backbones/universal/vit.py,sha256=kpUCoXpefR34hRNlQDFK9lGr4oqS8Mn5vTLKWZ-gaOs,1820
 eva/vision/models/networks/decoders/__init__.py,sha256=RXFWmoYw2i6E9VOUCJmU8c72icHannVuo-cUKy6fnLM,200
 eva/vision/models/networks/decoders/segmentation/__init__.py,sha256=N6jrhXHj0P7i7RptZbZ-JFehT2BM7meFyNIK0owAkaE,517
 eva/vision/models/networks/decoders/segmentation/base.py,sha256=b2TIJKiJR9vejVRpNyedMJLPTrpHhAEXvco8atb9TPU,411
-eva/vision/models/networks/decoders/segmentation/decoder2d.py,sha256=0jZrgFSdH5nlMYlbBmDb1E4kIQ3cG-LNOsiij51_NSA,4447
+eva/vision/models/networks/decoders/segmentation/decoder2d.py,sha256=A7vz0LJ_YweftpKeEBJm0Y3N7hbVLDSIkAajaQv1UgE,4456
 eva/vision/models/networks/decoders/segmentation/linear.py,sha256=-i9RVaKM1UsB3AXDDKdMmHiD7y2sr5HfF-WvkB47Fhw,4743
 eva/vision/models/networks/decoders/segmentation/semantic/__init__.py,sha256=Ubs8GXyQpEHs26JUeUuiVP3jfn47eiBZM_UVbu749XU,398
 eva/vision/models/networks/decoders/segmentation/semantic/common.py,sha256=fPTb0T-2FiOU-jT81ynASKaW7fJiRk6vQjuPkzHOluc,2530
 eva/vision/models/networks/decoders/segmentation/semantic/with_image.py,sha256=I5PyGKKo8DcXYcw4xlCFzuavRJNRrzGT-szpDidMPXI,3516
 eva/vision/models/networks/decoders/segmentation/typings.py,sha256=8zAqIJLlQdCjsx-Dl4lnF4BB1VxTg_AyIquBVwpZlHg,537
-eva/vision/models/wrappers/__init__.py,sha256=8MT8qFM4nUXGpK1_i3rp70ODkOjn2KhhRo2I17qZCPM,210
+eva/vision/models/wrappers/__init__.py,sha256=ogmr-eeVuGaOCcsuxSp6PGyauP2QqWTb8dGTtbC7lRU,210
 eva/vision/models/wrappers/from_registry.py,sha256=gdnxyg9drqlxfTNuS3aLbWGbZIwX1VNl0uudfjzVsXM,1614
 eva/vision/models/wrappers/from_timm.py,sha256=Z38Nb1i6OPKkgvFZOvGx-O3AZQuscf1zRVyrEBXQdJg,2320
 eva/vision/utils/__init__.py,sha256=vaUovprE743SmyFH8l6uk4pYSWpI4zxn7lN0EwePTJI,96
@@ -212,8 +225,8 @@ eva/vision/utils/io/image.py,sha256=IdOkr5MYqhYHz8U9drZ7wULTM3YHwCWSjZlu_Qdl4GQ,
 eva/vision/utils/io/mat.py,sha256=qpGifyjmpE0Xhv567Si7-zxKrgkgE0sywP70cHiLFGU,808
 eva/vision/utils/io/nifti.py,sha256=4YoKjKuoNdE0qY7tYB_WlnSsYAx2oBzZRZXczc_8HAU,2555
 eva/vision/utils/io/text.py,sha256=qYgfo_ZaDZWfG02NkVVYzo5QFySqdCCz5uLA9d-zXtI,701
-kaiko_eva-0.1.6.dist-info/METADATA,sha256=3pDsvMybfMJ6LpX3dRxqRFXvI6JC8x5EmKH2sYccV_4,24869
-kaiko_eva-0.1.6.dist-info/WHEEL,sha256=thaaA2w1JzcGC48WYufAs8nrYZjJm8LqNfnXFOFyCC4,90
-kaiko_eva-0.1.6.dist-info/entry_points.txt,sha256=6CSLu9bmQYJSXEg8gbOzRhxH0AGs75BB-vPm3VvfcNE,88
-kaiko_eva-0.1.6.dist-info/licenses/LICENSE,sha256=e6AEzr7j_R-PYr2qLO-JwLn8y70jbVD3U2mxbRmwcI4,11338
-kaiko_eva-0.1.6.dist-info/RECORD,,
+kaiko_eva-0.1.7.dist-info/METADATA,sha256=ToVlQrhgzB06Ptgus3EhUrMNaEBUFGm7YHTuZdy6JMM,24869
+kaiko_eva-0.1.7.dist-info/WHEEL,sha256=thaaA2w1JzcGC48WYufAs8nrYZjJm8LqNfnXFOFyCC4,90
+kaiko_eva-0.1.7.dist-info/entry_points.txt,sha256=6CSLu9bmQYJSXEg8gbOzRhxH0AGs75BB-vPm3VvfcNE,88
+kaiko_eva-0.1.7.dist-info/licenses/LICENSE,sha256=e6AEzr7j_R-PYr2qLO-JwLn8y70jbVD3U2mxbRmwcI4,11338
+kaiko_eva-0.1.7.dist-info/RECORD,,