monai-weekly 1.5.dev2518__tar.gz → 1.5.dev2520__tar.gz

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: monai-weekly
-Version: 1.5.dev2518
+Version: 1.5.dev2520
 Summary: AI Toolkit for Healthcare Imaging
 Home-page: https://monai.io/
 Author: MONAI Consortium
@@ -29,8 +29,7 @@ Classifier: Typing :: Typed
 Requires-Python: >=3.9
 Description-Content-Type: text/markdown; charset=UTF-8
 License-File: LICENSE
-Requires-Dist: torch>=2.3.0; sys_platform != "win32"
-Requires-Dist: torch>=2.4.1; sys_platform == "win32"
+Requires-Dist: torch<2.7.0,>=2.4.1
 Requires-Dist: numpy<3.0,>=1.24
 Provides-Extra: all
 Requires-Dist: nibabel; extra == "all"

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/__init__.py

@@ -136,4 +136,4 @@ except BaseException:
 
     if MONAIEnvVars.debug():
         raise
-__commit_id__ = "b58e883c887e0f99d382807550654c44d94f47bd"
+__commit_id__ = "707b23054025a26271e6ace9e9477801a71d8e0a"

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/_version.py

@@ -8,11 +8,11 @@ import json
 
 version_json = '''
 {
- "date": "2025-05-04T02:35:24+0000",
+ "date": "2025-05-18T02:35:03+0000",
  "dirty": false,
  "error": null,
- "full-revisionid": "09b361dbe4c2be5289c7e2d2f6c10c5166eee273",
- "version": "1.5.dev2518"
+ "full-revisionid": "cc43efa2f524c91686c29a51b04e31e47206dec8",
+ "version": "1.5.dev2520"
 }
 '''  # END VERSION_JSON
 

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/apps/deepgrow/transforms.py

@@ -441,8 +441,8 @@ class SpatialCropForegroundd(MapTransform):
 
         if np.all(np.less(current_size, self.spatial_size)):
             cropper = SpatialCrop(roi_center=center, roi_size=self.spatial_size)
-            box_start = np.array([s.start for s in cropper.slices])
-            box_end = np.array([s.stop for s in cropper.slices])
+            box_start = [s.start for s in cropper.slices]
+            box_end = [s.stop for s in cropper.slices]
         else:
             cropper = SpatialCrop(roi_start=box_start, roi_end=box_end)
 

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/bundle/scripts.py

@@ -31,7 +31,6 @@ from torch.cuda import is_available
 
 from monai._version import get_versions
 from monai.apps.utils import _basename, download_url, extractall, get_logger
-from monai.bundle.config_item import ConfigComponent
 from monai.bundle.config_parser import ConfigParser
 from monai.bundle.utils import DEFAULT_INFERENCE, DEFAULT_METADATA, merge_kv
 from monai.bundle.workflows import BundleWorkflow, ConfigWorkflow
@@ -48,7 +47,6 @@ from monai.networks import (
 from monai.utils import (
     IgniteInfo,
     check_parent_dir,
-    deprecated_arg,
     ensure_tuple,
     get_equivalent_dtype,
     min_version,
@@ -629,9 +627,6 @@ def download(
     _check_monai_version(bundle_dir_, name_)
 
 
-@deprecated_arg("net_name", since="1.2", removed="1.5", msg_suffix="please use ``model`` instead.")
-@deprecated_arg("net_kwargs", since="1.2", removed="1.5", msg_suffix="please use ``model`` instead.")
-@deprecated_arg("return_state_dict", since="1.2", removed="1.5")
 def load(
     name: str,
     model: torch.nn.Module | None = None,
@@ -650,10 +645,7 @@ def load(
     workflow_name: str | BundleWorkflow | None = None,
     args_file: str | None = None,
     copy_model_args: dict | None = None,
-    return_state_dict: bool = True,
     net_override: dict | None = None,
-    net_name: str | None = None,
-    **net_kwargs: Any,
 ) -> object | tuple[torch.nn.Module, dict, dict] | Any:
     """
     Load model weights or TorchScript module of a bundle.
@@ -699,12 +691,7 @@ def load(
         workflow_name: specified bundle workflow name, should be a string or class, default to "ConfigWorkflow".
         args_file: a JSON or YAML file to provide default values for all the args in "download" function.
         copy_model_args: other arguments for the `monai.networks.copy_model_state` function.
-        return_state_dict: whether to return state dict, if True, return state_dict, else a corresponding network
-            from `_workflow.network_def` will be instantiated and load the achieved weights.
         net_override: id-value pairs to override the parameters in the network of the bundle, default to `None`.
-        net_name: if not `None`, a corresponding network will be instantiated and load the achieved weights.
-            This argument only works when loading weights.
-        net_kwargs: other arguments that are used to instantiate the network class defined by `net_name`.
 
     Returns:
         1. If `load_ts_module` is `False` and `model` is `None`,
@@ -715,13 +702,8 @@ def load(
         3. If `load_ts_module` is `True`, return a triple that include a TorchScript module,
             the corresponding metadata dict, and extra files dict.
             please check `monai.data.load_net_with_metadata` for more details.
-        4. If `return_state_dict` is True, return model weights, only used for compatibility
-            when `model` and `net_name` are all `None`.
 
     """
-    if return_state_dict and (model is not None or net_name is not None):
-        warnings.warn("Incompatible values: model and net_name are all specified, return state dict instead.")
-
     bundle_dir_ = _process_bundle_dir(bundle_dir)
     net_override = {} if net_override is None else net_override
     copy_model_args = {} if copy_model_args is None else copy_model_args
@@ -757,11 +739,8 @@ def load(
         warnings.warn(f"the state dictionary from {full_path} should be a dictionary but got {type(model_dict)}.")
         model_dict = get_state_dict(model_dict)
 
-    if return_state_dict:
-        return model_dict
-
     _workflow = None
-    if model is None and net_name is None:
+    if model is None:
         bundle_config_file = bundle_dir_ / name / "configs" / f"{workflow_type}.json"
         if bundle_config_file.is_file():
             _net_override = {f"network_def#{key}": value for key, value in net_override.items()}
@@ -781,10 +760,6 @@ def load(
                 return model_dict
             else:
                 model = _workflow.network_def
-    elif net_name is not None:
-        net_kwargs["_target_"] = net_name
-        configer = ConfigComponent(config=net_kwargs)
-        model = configer.instantiate()  # type: ignore
 
     model.to(device)  # type: ignore
 

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/bundle/workflows.py

@@ -27,7 +27,7 @@ from monai.bundle.config_parser import ConfigParser
 from monai.bundle.properties import InferProperties, MetaProperties, TrainProperties
 from monai.bundle.utils import DEFAULT_EXP_MGMT_SETTINGS, EXPR_KEY, ID_REF_KEY, ID_SEP_KEY
 from monai.config import PathLike
-from monai.utils import BundleProperty, BundlePropertyConfig, deprecated_arg, ensure_tuple
+from monai.utils import BundleProperty, BundlePropertyConfig, ensure_tuple
 
 __all__ = ["BundleWorkflow", "ConfigWorkflow"]
 
@@ -45,10 +45,6 @@ class BundleWorkflow(ABC):
             or "infer", "inference", "eval", "evaluation" for a inference workflow,
             other unsupported string will raise a ValueError.
             default to `None` for only using meta properties.
-        workflow: specifies the workflow type: "train" or "training" for a training workflow,
-            or "infer", "inference", "eval", "evaluation" for a inference workflow,
-            other unsupported string will raise a ValueError.
-            default to `None` for common workflow.
         properties_path: the path to the JSON file of properties. If `workflow_type` is specified, properties will be
             loaded from the file based on the provided `workflow_type` and meta. If no `workflow_type` is specified,
             properties will default to loading from "meta". If `properties_path` is None, default properties
@@ -65,17 +61,9 @@ class BundleWorkflow(ABC):
     supported_train_type: tuple = ("train", "training")
     supported_infer_type: tuple = ("infer", "inference", "eval", "evaluation")
 
-    @deprecated_arg(
-        "workflow",
-        since="1.2",
-        removed="1.5",
-        new_name="workflow_type",
-        msg_suffix="please use `workflow_type` instead.",
-    )
     def __init__(
         self,
         workflow_type: str | None = None,
-        workflow: str | None = None,
         properties_path: PathLike | None = None,
         meta_file: str | Sequence[str] | None = None,
         logging_file: str | None = None,
@@ -102,7 +90,6 @@ class BundleWorkflow(ABC):
                         )
                         meta_file = None
 
-        workflow_type = workflow if workflow is not None else workflow_type
         if workflow_type is not None:
             if workflow_type.lower() in self.supported_train_type:
                 workflow_type = "train"
@@ -403,10 +390,6 @@ class ConfigWorkflow(BundleWorkflow):
             or "infer", "inference", "eval", "evaluation" for a inference workflow,
             other unsupported string will raise a ValueError.
             default to `None` for common workflow.
-        workflow: specifies the workflow type: "train" or "training" for a training workflow,
-            or "infer", "inference", "eval", "evaluation" for a inference workflow,
-            other unsupported string will raise a ValueError.
-            default to `None` for common workflow.
         properties_path: the path to the JSON file of properties. If `workflow_type` is specified, properties will be
             loaded from the file based on the provided `workflow_type` and meta. If no `workflow_type` is specified,
             properties will default to loading from "train". If `properties_path` is None, default properties
@@ -419,13 +402,6 @@ class ConfigWorkflow(BundleWorkflow):
 
     """
 
-    @deprecated_arg(
-        "workflow",
-        since="1.2",
-        removed="1.5",
-        new_name="workflow_type",
-        msg_suffix="please use `workflow_type` instead.",
-    )
     def __init__(
         self,
         config_file: str | Sequence[str],
@@ -436,11 +412,9 @@ class ConfigWorkflow(BundleWorkflow):
         final_id: str = "finalize",
         tracking: str | dict | None = None,
         workflow_type: str | None = "train",
-        workflow: str | None = None,
         properties_path: PathLike | None = None,
         **override: Any,
     ) -> None:
-        workflow_type = workflow if workflow is not None else workflow_type
         if config_file is not None:
             _config_files = ensure_tuple(config_file)
             config_root_path = Path(_config_files[0]).parent

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/metrics/__init__.py

@@ -19,7 +19,7 @@ from .f_beta_score import FBetaScore
 from .fid import FIDMetric, compute_frechet_distance
 from .froc import compute_fp_tp_probs, compute_fp_tp_probs_nd, compute_froc_curve_data, compute_froc_score
 from .generalized_dice import GeneralizedDiceScore, compute_generalized_dice
-from .hausdorff_distance import HausdorffDistanceMetric, compute_hausdorff_distance, compute_percent_hausdorff_distance
+from .hausdorff_distance import HausdorffDistanceMetric, compute_hausdorff_distance
 from .loss_metric import LossMetric
 from .meandice import DiceHelper, DiceMetric, compute_dice
 from .meaniou import MeanIoU, compute_iou

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/metrics/generalized_dice.py

@@ -14,7 +14,7 @@ from __future__ import annotations
 import torch
 
 from monai.metrics.utils import do_metric_reduction, ignore_background
-from monai.utils import MetricReduction, Weight, deprecated_arg, deprecated_arg_default, look_up_option
+from monai.utils import MetricReduction, Weight, deprecated_arg, look_up_option
 
 from .metric import CumulativeIterationMetric
 
@@ -37,6 +37,8 @@ class GeneralizedDiceScore(CumulativeIterationMetric):
         reduction: Define mode of reduction to the metrics. Available reduction modes:
             {``"none"``, ``"mean"``, ``"sum"``, ``"mean_batch"``, ``"sum_batch"``,
             ``"mean_channel"``, ``"sum_channel"``}, default to ``"mean"``. if "none", will not do reduction.
+            Default value is changed from `MetricReduction.MEAN_BATCH` to `MetricReduction.MEAN` in v1.5.0.
+            Old versions computed `mean` when `mean_batch` was provided due to bug in reduction.
         weight_type: {``"square"``, ``"simple"``, ``"uniform"``}. Type of function to transform
             ground truth volume into a weight factor. Defaults to ``"square"``.
 
@@ -44,21 +46,10 @@ class GeneralizedDiceScore(CumulativeIterationMetric):
         ValueError: When the `reduction` is not one of MetricReduction enum.
     """
 
-    @deprecated_arg_default(
-        "reduction",
-        old_default=MetricReduction.MEAN_BATCH,
-        new_default=MetricReduction.MEAN,
-        since="1.4.0",
-        replaced="1.5.0",
-        msg_suffix=(
-            "Old versions computed `mean` when `mean_batch` was provided due to bug in reduction， "
-            "If you want to retain the old behavior (calculating the mean), please explicitly set the parameter to 'mean'."
-        ),
-    )
     def __init__(
         self,
         include_background: bool = True,
-        reduction: MetricReduction | str = MetricReduction.MEAN_BATCH,
+        reduction: MetricReduction | str = MetricReduction.MEAN,
         weight_type: Weight | str = Weight.SQUARE,
     ) -> None:
         super().__init__()

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/metrics/hausdorff_distance.py

@@ -17,18 +17,12 @@ from typing import Any
 import numpy as np
 import torch
 
-from monai.metrics.utils import (
-    do_metric_reduction,
-    get_edge_surface_distance,
-    get_surface_distance,
-    ignore_background,
-    prepare_spacing,
-)
-from monai.utils import MetricReduction, convert_data_type, deprecated
+from monai.metrics.utils import do_metric_reduction, get_edge_surface_distance, ignore_background, prepare_spacing
+from monai.utils import MetricReduction, convert_data_type
 
 from .metric import CumulativeIterationMetric
 
-__all__ = ["HausdorffDistanceMetric", "compute_hausdorff_distance", "compute_percent_hausdorff_distance"]
+__all__ = ["HausdorffDistanceMetric", "compute_hausdorff_distance"]
 
 
 class HausdorffDistanceMetric(CumulativeIterationMetric):
@@ -216,31 +210,3 @@ def _compute_percentile_hausdorff_distance(
     if 0 <= percentile <= 100:
         return torch.quantile(surface_distance, percentile / 100)
     raise ValueError(f"percentile should be a value between 0 and 100, get {percentile}.")
-
-
-@deprecated(since="1.3.0", removed="1.5.0")
-def compute_percent_hausdorff_distance(
-    edges_pred: np.ndarray,
-    edges_gt: np.ndarray,
-    distance_metric: str = "euclidean",
-    percentile: float | None = None,
-    spacing: int | float | np.ndarray | Sequence[int | float] | None = None,
-) -> float:
-    """
-    This function is used to compute the directed Hausdorff distance.
-    """
-
-    surface_distance: np.ndarray = get_surface_distance(  # type: ignore
-        edges_pred, edges_gt, distance_metric=distance_metric, spacing=spacing
-    )
-
-    # for both pred and gt do not have foreground
-    if surface_distance.shape == (0,):
-        return np.nan
-
-    if not percentile:
-        return surface_distance.max()  # type: ignore[no-any-return]
-
-    if 0 <= percentile <= 100:
-        return np.percentile(surface_distance, percentile)  # type: ignore[no-any-return]
-    raise ValueError(f"percentile should be a value between 0 and 100, get {percentile}.")

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/metrics/utils.py

@@ -30,7 +30,6 @@ from monai.utils import (
     convert_to_numpy,
     convert_to_tensor,
     deprecated_arg,
-    deprecated_arg_default,
     ensure_tuple_rep,
     look_up_option,
     optional_import,
@@ -131,9 +130,6 @@ def do_metric_reduction(
     return f, not_nans
 
 
-@deprecated_arg_default(
-    name="always_return_as_numpy", since="1.3.0", replaced="1.5.0", old_default=True, new_default=False
-)
 @deprecated_arg(
     name="always_return_as_numpy",
     since="1.5.0",
@@ -146,7 +142,7 @@ def get_mask_edges(
     label_idx: int = 1,
     crop: bool = True,
     spacing: Sequence | None = None,
-    always_return_as_numpy: bool = True,
+    always_return_as_numpy: bool = False,
 ) -> tuple[NdarrayTensor, NdarrayTensor]:
     """
     Compute edges from binary segmentation masks. This
@@ -175,6 +171,7 @@ def get_mask_edges(
             otherwise `scipy`'s binary erosion is used to calculate the edges.
         always_return_as_numpy: whether to a numpy array regardless of the input type.
             If False, return the same type as inputs.
+            The default value is changed from `True` to `False` in v1.5.0.
     """
     # move in the funciton to avoid using all the GPUs
     cucim_binary_erosion, has_cucim_binary_erosion = optional_import("cucim.skimage.morphology", name="binary_erosion")

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/networks/nets/autoencoderkl.py

@@ -153,9 +153,9 @@ class Encoder(nn.Module):
         channels: sequence of block output channels.
         out_channels: number of channels in the bottom layer (latent space) of the autoencoder.
         num_res_blocks: number of residual blocks (see _ResBlock) per level.
-        norm_num_groups: number of groups for the GroupNorm layers, num_channels must be divisible by this number.
+        norm_num_groups: number of groups for the GroupNorm layers, channels must be divisible by this number.
         norm_eps: epsilon for the normalization.
-        attention_levels: indicate which level from num_channels contain an attention block.
+        attention_levels: indicate which level from channels contain an attention block.
         with_nonlocal_attn: if True use non-local attention block.
         include_fc: whether to include the final linear layer. Default to True.
         use_combined_linear: whether to use a single linear layer for qkv projection, default to False.
@@ -299,9 +299,9 @@ class Decoder(nn.Module):
         in_channels: number of channels in the bottom layer (latent space) of the autoencoder.
         out_channels: number of output channels.
         num_res_blocks: number of residual blocks (see _ResBlock) per level.
-        norm_num_groups: number of groups for the GroupNorm layers, num_channels must be divisible by this number.
+        norm_num_groups: number of groups for the GroupNorm layers, channels must be divisible by this number.
         norm_eps: epsilon for the normalization.
-        attention_levels: indicate which level from num_channels contain an attention block.
+        attention_levels: indicate which level from channels contain an attention block.
         with_nonlocal_attn: if True use non-local attention block.
         use_convtranspose: if True, use ConvTranspose to upsample feature maps in decoder.
         include_fc: whether to include the final linear layer. Default to True.
@@ -483,7 +483,7 @@ class AutoencoderKL(nn.Module):
         channels: number of output channels for each block.
         attention_levels: sequence of levels to add attention.
         latent_channels: latent embedding dimension.
-        norm_num_groups: number of groups for the GroupNorm layers, num_channels must be divisible by this number.
+        norm_num_groups: number of groups for the GroupNorm layers, channels must be divisible by this number.
         norm_eps: epsilon for the normalization.
         with_encoder_nonlocal_attn: if True use non-local attention block in the encoder.
         with_decoder_nonlocal_attn: if True use non-local attention block in the decoder.
@@ -518,10 +518,10 @@ class AutoencoderKL(nn.Module):
 
         # All number of channels should be multiple of num_groups
         if any((out_channel % norm_num_groups) != 0 for out_channel in channels):
-            raise ValueError("AutoencoderKL expects all num_channels being multiple of norm_num_groups")
+            raise ValueError("AutoencoderKL expects all channels being multiple of norm_num_groups")
 
         if len(channels) != len(attention_levels):
-            raise ValueError("AutoencoderKL expects num_channels being same size of attention_levels")
+            raise ValueError("AutoencoderKL expects channels being same size of attention_levels")
 
         if isinstance(num_res_blocks, int):
             num_res_blocks = ensure_tuple_rep(num_res_blocks, len(channels))
@@ -529,7 +529,7 @@ class AutoencoderKL(nn.Module):
         if len(num_res_blocks) != len(channels):
             raise ValueError(
                 "`num_res_blocks` should be a single integer or a tuple of integers with the same length as "
-                "`num_channels`."
+                "`channels`."
             )
 
         self.encoder: nn.Module = Encoder(

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/networks/nets/swin_unetr.py

@@ -25,7 +25,6 @@ from monai.networks.blocks import MLPBlock as Mlp
 from monai.networks.blocks import PatchEmbed, UnetOutBlock, UnetrBasicBlock, UnetrUpBlock
 from monai.networks.layers import DropPath, trunc_normal_
 from monai.utils import ensure_tuple_rep, look_up_option, optional_import
-from monai.utils.deprecate_utils import deprecated_arg
 
 rearrange, _ = optional_import("einops", name="rearrange")
 
@@ -50,16 +49,8 @@ class SwinUNETR(nn.Module):
     <https://arxiv.org/abs/2201.01266>"
     """
 
-    @deprecated_arg(
-        name="img_size",
-        since="1.3",
-        removed="1.5",
-        msg_suffix="The img_size argument is not required anymore and "
-        "checks on the input size are run during forward().",
-    )
     def __init__(
         self,
-        img_size: Sequence[int] | int,
         in_channels: int,
         out_channels: int,
         patch_size: int = 2,
@@ -83,10 +74,6 @@ class SwinUNETR(nn.Module):
     ) -> None:
         """
         Args:
-            img_size: spatial dimension of input image.
-                This argument is only used for checking that the input image size is divisible by the patch size.
-                The tensor passed to forward() can have a dynamic shape as long as its spatial dimensions are divisible by 2**5.
-                It will be removed in an upcoming version.
             in_channels: dimension of input channels.
             out_channels: dimension of output channels.
             patch_size: size of the patch token.
@@ -113,13 +100,13 @@ class SwinUNETR(nn.Module):
         Examples::
 
             # for 3D single channel input with size (96,96,96), 4-channel output and feature size of 48.
-            >>> net = SwinUNETR(img_size=(96,96,96), in_channels=1, out_channels=4, feature_size=48)
+            >>> net = SwinUNETR(in_channels=1, out_channels=4, feature_size=48)
 
             # for 3D 4-channel input with size (128,128,128), 3-channel output and (2,4,2,2) layers in each stage.
-            >>> net = SwinUNETR(img_size=(128,128,128), in_channels=4, out_channels=3, depths=(2,4,2,2))
+            >>> net = SwinUNETR(in_channels=4, out_channels=3, depths=(2,4,2,2))
 
             # for 2D single channel input with size (96,96), 2-channel output and gradient checkpointing.
-            >>> net = SwinUNETR(img_size=(96,96), in_channels=3, out_channels=2, use_checkpoint=True, spatial_dims=2)
+            >>> net = SwinUNETR(in_channels=3, out_channels=2, use_checkpoint=True, spatial_dims=2)
 
         """
 
@@ -130,12 +117,9 @@ class SwinUNETR(nn.Module):
 
         self.patch_size = patch_size
 
-        img_size = ensure_tuple_rep(img_size, spatial_dims)
         patch_sizes = ensure_tuple_rep(self.patch_size, spatial_dims)
         window_size = ensure_tuple_rep(window_size, spatial_dims)
 
-        self._check_input_size(img_size)
-
         if not (0 <= drop_rate <= 1):
             raise ValueError("dropout rate should be between 0 and 1.")
 
@@ -1109,7 +1093,7 @@ def filter_swinunetr(key, value):
         from monai.networks.utils import copy_model_state
         from monai.networks.nets.swin_unetr import SwinUNETR, filter_swinunetr
 
-        model = SwinUNETR(img_size=(96, 96, 96), in_channels=1, out_channels=3, feature_size=48)
+        model = SwinUNETR(in_channels=1, out_channels=3, feature_size=48)
         resource = (
             "https://github.com/Project-MONAI/MONAI-extra-test-data/releases/download/0.8.1/ssl_pretrained_weights.pth"
         )

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/networks/schedulers/ddpm.py

@@ -238,7 +238,7 @@ class DDPMScheduler(Scheduler):
         pred_prev_sample = pred_original_sample_coeff * pred_original_sample + current_sample_coeff * sample
 
         # 6. Add noise
-        variance = 0
+        variance: torch.Tensor = torch.tensor(0)
         if timestep > 0:
             noise = torch.randn(
                 model_output.size(),

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/networks/schedulers/scheduler.py

@@ -105,9 +105,11 @@ def _cosine_beta(num_train_timesteps: int, s: float = 8e-3):
     x = torch.linspace(0, num_train_timesteps, num_train_timesteps + 1)
     alphas_cumprod = torch.cos(((x / num_train_timesteps) + s) / (1 + s) * torch.pi * 0.5) ** 2
     alphas_cumprod /= alphas_cumprod[0].item()
-    alphas = torch.clip(alphas_cumprod[1:] / alphas_cumprod[:-1], 0.0001, 0.9999)
-    betas = 1.0 - alphas
-    return betas, alphas, alphas_cumprod[:-1]
+    betas = 1.0 - (alphas_cumprod[1:] / alphas_cumprod[:-1])
+    betas = torch.clip(betas, 0.0, 0.999)
+    alphas = 1.0 - betas
+    alphas_cumprod = torch.cumprod(alphas, dim=0)
+    return betas, alphas, alphas_cumprod
 
 
 class Scheduler(nn.Module):

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/transforms/croppad/array.py

@@ -51,7 +51,6 @@ from monai.utils import (
     TransformBackends,
     convert_data_type,
     convert_to_tensor,
-    deprecated_arg_default,
     ensure_tuple,
     ensure_tuple_rep,
     fall_back_tuple,
@@ -809,13 +808,12 @@ class CropForeground(Crop):
 
     """
 
-    @deprecated_arg_default("allow_smaller", old_default=True, new_default=False, since="1.2", replaced="1.5")
     def __init__(
         self,
         select_fn: Callable = is_positive,
         channel_indices: IndexSelection | None = None,
         margin: Sequence[int] | int = 0,
-        allow_smaller: bool = True,
+        allow_smaller: bool = False,
         return_coords: bool = False,
         k_divisible: Sequence[int] | int = 1,
         mode: str = PytorchPadMode.CONSTANT,
@@ -830,7 +828,8 @@ class CropForeground(Crop):
             margin: add margin value to spatial dims of the bounding box, if only 1 value provided, use it for all dims.
             allow_smaller: when computing box size with `margin`, whether to allow the image edges to be smaller than the
                 final box edges. If `False`, part of a padded output box might be outside of the original image, if `True`,
-                the image edges will be used as the box edges. Default to `True`.
+                the image edges will be used as the box edges. Default to `False`.
+                The default value is changed from `True` to `False` in v1.5.0.
             return_coords: whether return the coordinates of spatial bounding box for foreground.
             k_divisible: make each spatial dimension to be divisible by k, default to 1.
                 if `k_divisible` is an int, the same `k` be applied to all the input spatial dimensions.

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/transforms/croppad/dictionary.py

@@ -50,7 +50,7 @@ from monai.transforms.inverse import InvertibleTransform
 from monai.transforms.traits import LazyTrait, MultiSampleTrait
 from monai.transforms.transform import LazyTransform, MapTransform, Randomizable
 from monai.transforms.utils import is_positive
-from monai.utils import MAX_SEED, Method, PytorchPadMode, deprecated_arg_default, ensure_tuple_rep
+from monai.utils import MAX_SEED, Method, PytorchPadMode, ensure_tuple_rep
 
 __all__ = [
     "Padd",
@@ -716,7 +716,6 @@ class CropForegroundd(Cropd):
     for more information.
     """
 
-    @deprecated_arg_default("allow_smaller", old_default=True, new_default=False, since="1.2", replaced="1.5")
     def __init__(
         self,
         keys: KeysCollection,
@@ -724,7 +723,7 @@ class CropForegroundd(Cropd):
         select_fn: Callable = is_positive,
         channel_indices: IndexSelection | None = None,
         margin: Sequence[int] | int = 0,
-        allow_smaller: bool = True,
+        allow_smaller: bool = False,
         k_divisible: Sequence[int] | int = 1,
         mode: SequenceStr = PytorchPadMode.CONSTANT,
         start_coord_key: str | None = "foreground_start_coord",
@@ -744,7 +743,8 @@ class CropForegroundd(Cropd):
             margin: add margin value to spatial dims of the bounding box, if only 1 value provided, use it for all dims.
             allow_smaller: when computing box size with `margin`, whether to allow the image edges to be smaller than the
                 final box edges. If `False`, part of a padded output box might be outside of the original image, if `True`,
-                the image edges will be used as the box edges. Default to `True`.
+                the image edges will be used as the box edges. Default to `False`.
+                The default value is changed from `True` to `False` in v1.5.0.
             k_divisible: make each spatial dimension to be divisible by k, default to 1.
                 if `k_divisible` is an int, the same `k` be applied to all the input spatial dimensions.
             mode: available modes for numpy array:{``"constant"``, ``"edge"``, ``"linear_ramp"``, ``"maximum"``,

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/transforms/utils.py

@@ -58,7 +58,6 @@ from monai.utils import (
     SplineMode,
     TraceKeys,
     TraceStatusKeys,
-    deprecated_arg_default,
     ensure_tuple,
     ensure_tuple_rep,
     ensure_tuple_size,
@@ -1067,13 +1066,12 @@ def _create_translate(
     return array_func(affine)  # type: ignore
 
 
-@deprecated_arg_default("allow_smaller", old_default=True, new_default=False, since="1.2", replaced="1.5")
 def generate_spatial_bounding_box(
     img: NdarrayOrTensor,
     select_fn: Callable = is_positive,
     channel_indices: IndexSelection | None = None,
     margin: Sequence[int] | int = 0,
-    allow_smaller: bool = True,
+    allow_smaller: bool = False,
 ) -> tuple[list[int], list[int]]:
     """
     Generate the spatial bounding box of foreground in the image with start-end positions (inclusive).
@@ -1093,8 +1091,9 @@ def generate_spatial_bounding_box(
             of image. if None, select foreground on the whole image.
         margin: add margin value to spatial dims of the bounding box, if only 1 value provided, use it for all dims.
         allow_smaller: when computing box size with `margin`, whether to allow the image edges to be smaller than the
-                final box edges. If `True`, the bounding boxes edges are aligned with the input image edges, if `False`,
-                the bounding boxes edges are aligned with the final box edges. Default to `True`.
+            final box edges. If `True`, the bounding boxes edges are aligned with the input image edges, if `False`,
+            the bounding boxes edges are aligned with the final box edges. Default to `False`.
+            The default value is changed from `True` to `False` in v1.5.0.
 
     """
     check_non_lazy_pending_ops(img, name="generate_spatial_bounding_box")

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/utils/misc.py

@@ -546,7 +546,7 @@ class MONAIEnvVars:
 
     @staticmethod
     def algo_hash() -> str | None:
-        return os.environ.get("MONAI_ALGO_HASH", "4c18daf")
+        return os.environ.get("MONAI_ALGO_HASH", "21ed8e5")
 
     @staticmethod
     def trace_transform() -> str | None:

{monai_weekly-1.5.dev2518 → monai_weekly-1.5.dev2520}/monai/visualize/occlusion_sensitivity.py

@@ -80,7 +80,7 @@ class OcclusionSensitivity:
         n_batch: int = 16,
         verbose: bool = True,
         mode: str | float | Callable = "gaussian",
-        overlap: float = 0.25,
+        overlap: float = 0.6,
         activate: bool | Callable = True,
     ) -> None:
         """