code-loader 1.0.49.dev9__tar.gz → 1.0.49.dev100__tar.gz

{code_loader-1.0.49.dev9 → code_loader-1.0.49.dev100}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: code-loader
-Version: 1.0.49.dev9
+Version: 1.0.49.dev100
 Summary: 
 Home-page: https://github.com/tensorleap/code-loader
 License: MIT

{code_loader-1.0.49.dev9 → code_loader-1.0.49.dev100}/code_loader/contract/datasetclasses.py

@@ -36,9 +36,9 @@ class PreprocessResponse:
     data: Any = None
     sample_ids: Optional[Union[List[str], List[int]]] = None
     state: Optional[DataStateType] = None
-    sample_id_type: Optional[Type] = None
+    sample_id_type: Optional[Union[Type[str], Type[int]]] = None
 
-    def __post_init__(self):
+    def __post_init__(self) -> None:
         if self.length is not None and self.sample_ids is None:
             self.sample_ids = [i for i in range(self.length)]
             self.sample_id_type = int
@@ -47,9 +47,10 @@ class PreprocessResponse:
             if self.sample_id_type is None:
                 self.sample_id_type = str
         else:
-            raise Exception("length is deprecated. Please use sample_ids instead.")
+            raise Exception("length is deprecated.")
 
-    def __len__(self):
+    def __len__(self) -> int:
+        assert self.sample_ids is not None
         return len(self.sample_ids)
 
 
@@ -90,8 +91,8 @@ VisualizerCallableInterface = Union[
     Callable[..., LeapImageWithHeatmap]
 ]
 
-VisualizerCallableReturnType = Union[LeapImage, LeapText, LeapGraph, LeapHorizontalBar,
-LeapImageMask, LeapTextMask, LeapImageWithBBox, LeapImageWithHeatmap]
+LeapData = Union[LeapImage, LeapText, LeapGraph, LeapHorizontalBar, LeapImageMask, LeapTextMask, LeapImageWithBBox,
+LeapImageWithHeatmap]
 
 CustomCallableInterface = Callable[..., Any]
 

{code_loader-1.0.49.dev9 → code_loader-1.0.49.dev100}/code_loader/contract/visualizer_classes.py

@@ -4,8 +4,6 @@ import numpy as np
 import numpy.typing as npt
 from dataclasses import dataclass
 
-import matplotlib.pyplot as plt     # type: ignore
-
 from code_loader.contract.enums import LeapDataType
 from code_loader.contract.responsedataclasses import BoundingBox
 
@@ -51,34 +49,6 @@ class LeapImage:
         validate_type(len(self.data.shape), 3, 'Image must be of shape 3')
         validate_type(self.data.shape[2], [1, 3], 'Image channel must be either 3(rgb) or 1(gray)')
 
-    def plot_visualizer(self) -> None:
-        """
-        Display the image contained in the LeapImage object.
-
-        Returns:
-            None
-
-        Example:
-            image_data = np.random.rand(100, 100, 3).astype(np.float32)
-            leap_image = LeapImage(data=image_data)
-            leap_image.plot_visualizer()
-        """
-        image_data = self.data
-
-        # If the image has one channel, convert it to a 3-channel image for display
-        if image_data.shape[2] == 1:
-            image_data = np.repeat(image_data, 3, axis=2)
-
-        fig, ax = plt.subplots()
-        fig.patch.set_facecolor('black')
-        ax.set_facecolor('black')
-
-        ax.imshow(image_data)
-
-        plt.axis('off')
-        plt.title('Leap Image Visualization', color='white')
-        plt.show()
-
 
 @dataclass
 class LeapImageWithBBox:
@@ -106,60 +76,6 @@ class LeapImageWithBBox:
         validate_type(len(self.data.shape), 3, 'Image must be of shape 3')
         validate_type(self.data.shape[2], [1, 3], 'Image channel must be either 3(rgb) or 1(gray)')
 
-    def plot_visualizer(self) -> None:
-        """
-        Plot an image with overlaid bounding boxes.
-
-        Returns:
-            None
-
-        Example:
-            image_data = np.random.rand(100, 100, 3).astype(np.float32)
-            bbox = BoundingBox(x=0.5, y=0.5, width=0.2, height=0.2, confidence=0.9, label="object")
-            leap_image_with_bbox = LeapImageWithBBox(data=image_data, bounding_boxes=[bbox])
-            leap_image_with_bbox.plot_visualizer()
-        """
-
-        image = self.data
-        bounding_boxes = self.bounding_boxes
-
-        # Create figure and axes
-        fig, ax = plt.subplots(1)
-        fig.patch.set_facecolor('black')
-        ax.set_facecolor('black')
-
-        # Display the image
-        ax.imshow(image)
-        ax.set_title('Leap Image With BBox Visualization', color='white')
-
-        # Draw bounding boxes on the image
-        for bbox in bounding_boxes:
-            x, y, width, height = bbox.x, bbox.y, bbox.width, bbox.height
-            confidence, label = bbox.confidence, bbox.label
-
-            # Convert relative coordinates to absolute coordinates
-            abs_x = x * image.shape[1]
-            abs_y = y * image.shape[0]
-            abs_width = width * image.shape[1]
-            abs_height = height * image.shape[0]
-
-            # Create a rectangle patch
-            rect = plt.Rectangle(
-                (abs_x - abs_width / 2, abs_y - abs_height / 2),
-                abs_width, abs_height,
-                linewidth=3, edgecolor='r', facecolor='none'
-            )
-
-            # Add the rectangle to the axes
-            ax.add_patch(rect)
-
-            # Display label and confidence
-            ax.text(abs_x - abs_width / 2, abs_y - abs_height / 2 - 5,
-                    f"{label} {confidence:.2f}", color='r', fontsize=8,
-                    bbox=dict(facecolor='white', alpha=0.7, edgecolor='none', boxstyle='round,pad=0.3'))
-
-        # Show the image with bounding boxes
-        plt.show()
 
 @dataclass
 class LeapGraph:
@@ -183,40 +99,6 @@ class LeapGraph:
         validate_type(self.data.dtype, np.float32)
         validate_type(len(self.data.shape), 2, 'Graph must be of shape 2')
 
-    def plot_visualizer(self) -> None:
-        """
-        Display the line chart contained in the LeapGraph object.
-
-        Returns:
-            None
-
-        Example:
-            graph_data = np.random.rand(100, 3).astype(np.float32)
-            leap_graph = LeapGraph(data=graph_data)
-            leap_graph.plot_visualizer()
-        """
-        graph_data = self.data
-        num_variables = graph_data.shape[1]
-
-        fig, ax = plt.subplots(figsize=(10, 6))
-
-        # Set the background color to black
-        fig.patch.set_facecolor('black')
-        ax.set_facecolor('black')
-
-        for i in range(num_variables):
-            plt.plot(graph_data[:, i], label=f'Variable {i + 1}')
-
-        ax.set_xlabel('Data Points', color='white')
-        ax.set_ylabel('Values', color='white')
-        ax.set_title('Leap Graph Visualization', color='white')
-        ax.legend()
-        ax.grid(True, color='white')
-
-        # Change the color of the tick labels to white
-        ax.tick_params(colors='white')
-
-        plt.show()
 
 @dataclass
 class LeapText:
@@ -242,41 +124,6 @@ class LeapText:
         for value in self.data:
             validate_type(type(value), str)
 
-    def plot_visualizer(self) -> None:
-        """
-        Display the text contained in the LeapText object.
-
-        Returns:
-            None
-
-        Example:
-            text_data = ['I', 'ate', 'a', 'banana', '', '', '']
-            leap_text = LeapText(data=text_data)
-            leap_text.plot_visualizer()
-        """
-        text_data = self.data
-        # Join the text tokens into a single string, ignoring empty strings
-        display_text = ' '.join([token for token in text_data if token])
-
-        # Create a black image using Matplotlib
-        fig, ax = plt.subplots(figsize=(10, 5))
-        fig.patch.set_facecolor('black')
-        ax.set_facecolor('black')
-
-        # Hide the axes
-        ax.axis('off')
-
-        # Set the text properties
-        font_size = 20
-        font_color = 'white'
-
-        # Add the text to the image
-        ax.text(0.5, 0.5, display_text, color=font_color, fontsize=font_size, ha='center', va='center')
-        ax.set_title('Leap Text Visualization', color='white')
-
-        # Display the image
-        plt.show()
-
 
 @dataclass
 class LeapHorizontalBar:
@@ -309,39 +156,6 @@ class LeapHorizontalBar:
         for label in self.labels:
             validate_type(type(label), str)
 
-    def plot_visualizer(self) -> None:
-        """
-        Display the horizontal bar chart contained in the LeapHorizontalBar object.
-
-        Returns:
-            None
-
-        Example:
-            body_data = np.random.rand(5).astype(np.float32)
-            labels = ['Class A', 'Class B', 'Class C', 'Class D', 'Class E']
-            leap_horizontal_bar = LeapHorizontalBar(body=body_data, labels=labels)
-            leap_horizontal_bar.plot_visualizer()
-        """
-        body_data = self.body
-        labels = self.labels
-
-        fig, ax = plt.subplots()
-
-        fig.patch.set_facecolor('black')
-        ax.set_facecolor('black')
-
-        # Plot horizontal bar chart
-        ax.barh(labels, body_data, color='green')
-
-        # Set the color of the labels and title to white
-        ax.set_xlabel('Scores', color='white')
-        ax.set_title('Leap Horizontal Bar Visualization', color='white')
-
-        # Set the color of the ticks to white
-        ax.tick_params(axis='x', colors='white')
-        ax.tick_params(axis='y', colors='white')
-
-        plt.show()
 
 @dataclass
 class LeapImageMask:
@@ -379,51 +193,6 @@ class LeapImageMask:
         for label in self.labels:
             validate_type(type(label), str)
 
-    def plot_visualizer(self) -> None:
-        """
-        Plots an image with overlaid masks given a LeapImageMask visualizer object.
-
-        Returns:
-            None
-
-
-        Example:
-            image_data = np.random.rand(100, 100, 3).astype(np.float32)
-            mask_data = np.random.randint(0, 2, (100, 100)).astype(np.uint8)
-            labels = ["background", "object"]
-            leap_image_mask = LeapImageMask(image=image_data, mask=mask_data, labels=labels)
-            leap_image_mask.plot_visualizer()
-        """
-
-        image = self.image
-        mask = self.mask
-        labels = self.labels
-
-        # Create a color map for each label
-        colors = plt.cm.jet(np.linspace(0, 1, len(labels)))
-
-        # Make a copy of the image to draw on
-        overlayed_image = image.copy()
-
-        # Iterate through the unique values in the mask (excluding 0)
-        for i, label in enumerate(labels):
-            # Extract binary mask for the current instance
-            instance_mask = (mask == (i + 1))
-
-            # fill the instance mask with a translucent color
-            overlayed_image[instance_mask] = (
-                    overlayed_image[instance_mask] * (1 - 0.5) + np.array(colors[i][:3], dtype=np.uint8) * 0.5)
-
-        # Display the result using matplotlib
-        fig, ax = plt.subplots(1)
-        fig.patch.set_facecolor('black')  # Set the figure background to black
-        ax.set_facecolor('black')  # Set the axis background to black
-
-        ax.imshow(overlayed_image)
-        ax.set_title('Leap Image With Mask Visualization', color='white')
-        plt.axis('off')  # Hide the axis
-        plt.show()
-
 
 @dataclass
 class LeapTextMask:
@@ -461,55 +230,6 @@ class LeapTextMask:
         for label in self.labels:
             validate_type(type(label), str)
 
-    def plot_visualizer(self) -> None:
-        """
-        Plots text with overlaid masks given a LeapTextMask visualizer object.
-
-        Returns:
-            None
-
-        Example:
-            text_data = ['I', 'ate', 'a', 'banana', '', '', '']
-            mask_data = np.array([0, 0, 0, 1, 0, 0, 0]).astype(np.uint8)
-            labels = ["object"]
-            leap_text_mask = LeapTextMask(text=text_data, mask=mask_data, labels=labels)
-        """
-
-        text_data = self.text
-        mask_data = self.mask
-        labels = self.labels
-
-        # Create a color map for each label
-        colors = plt.cm.jet(np.linspace(0, 1, len(labels)))
-
-        # Create a figure and axis
-        fig, ax = plt.subplots()
-
-        # Set background to black
-        fig.patch.set_facecolor('black')
-        ax.set_facecolor('black')
-        ax.set_title('Leap Text Mask Visualization', color='white')
-        ax.axis('off')
-
-        # Set initial position
-        x_pos, y_pos = 0.01, 0.5  # Adjusted initial position for better visibility
-
-        # Display the text with colors
-        for token, mask_value in zip(text_data, mask_data):
-            if mask_value > 0:
-                color = colors[mask_value % len(colors)]
-                bbox = dict(facecolor=color, edgecolor='none',
-                            boxstyle='round,pad=0.3')  # Background color for masked tokens
-            else:
-                bbox = None
-
-            ax.text(x_pos, y_pos, token, fontsize=12, color='white', ha='left', va='center', bbox=bbox)
-
-            # Update the x position for the next token
-            x_pos += len(token) * 0.03 + 0.02  # Adjust the spacing between tokens
-
-        plt.show()
-
 
 @dataclass
 class LeapImageWithHeatmap:
@@ -547,46 +267,6 @@ class LeapImageWithHeatmap:
             raise LeapValidationError(
                 'Number of heatmaps and labels must be equal')
 
-    def plot_visualizer(self) -> None:
-        """
-        Display the image with overlaid heatmaps contained in the LeapImageWithHeatmap object.
-
-        Returns:
-            None
-
-        Example:
-            image_data = np.random.rand(100, 100, 3).astype(np.float32)
-            heatmaps = np.random.rand(3, 100, 100).astype(np.float32)
-            labels = ["heatmap1", "heatmap2", "heatmap3"]
-            leap_image_with_heatmap = LeapImageWithHeatmap(image=image_data, heatmaps=heatmaps, labels=labels)
-            leap_image_with_heatmap.plot_visualizer()
-        """
-        image = self.image
-        heatmaps = self.heatmaps
-        labels = self.labels
-
-        # Plot the base image
-        fig, ax = plt.subplots()
-        fig.patch.set_facecolor('black')  # Set the figure background to black
-        ax.set_facecolor('black')  # Set the axis background to black
-        ax.imshow(image, cmap='gray')
-
-        # Overlay each heatmap with some transparency
-        for i in range(len(labels)):
-            heatmap = heatmaps[i]
-            ax.imshow(heatmap, cmap='jet', alpha=0.5)  # Adjust alpha for transparency
-            ax.set_title(f'Heatmap: {labels[i]}', color='white')
-
-            # Display a colorbar for the heatmap
-            cbar = plt.colorbar(ax.imshow(heatmap, cmap='jet', alpha=0.5))
-            cbar.set_label(labels[i], color='white')
-            cbar.ax.yaxis.set_tick_params(color='white')  # Set color for the colorbar ticks
-            plt.setp(plt.getp(cbar.ax.axes, 'yticklabels'), color='white')  # Set color for the colorbar labels
-
-        plt.axis('off')
-        plt.title('Leap Image With Heatmaps Visualization', color='white')
-        plt.show()
-
 
 map_leap_data_type_to_visualizer_class = {
     LeapDataType.Image.value: LeapImage,

{code_loader-1.0.49.dev9 → code_loader-1.0.49.dev100}/code_loader/inner_leap_binder/leapbinder.py

@@ -1,4 +1,5 @@
 import inspect
+import sys
 from typing import Callable, List, Optional, Dict, Any, Type, Union
 
 import numpy as np
@@ -8,7 +9,7 @@ from code_loader.contract.datasetclasses import SectionCallableInterface, InputH
     GroundTruthHandler, MetadataHandler, DatasetIntegrationSetup, VisualizerHandler, PreprocessResponse, \
     PreprocessHandler, VisualizerCallableInterface, CustomLossHandler, CustomCallableInterface, PredictionTypeHandler, \
     MetadataSectionCallableInterface, UnlabeledDataPreprocessHandler, CustomLayerHandler, MetricHandler, \
-    CustomCallableInterfaceMultiArgs, ConfusionMatrixCallableInterfaceMultiArgs, VisualizerCallableReturnType, \
+    CustomCallableInterfaceMultiArgs, ConfusionMatrixCallableInterfaceMultiArgs, LeapData, \
     CustomMultipleReturnCallableInterfaceMultiArgs, DatasetBaseHandler, custom_latent_space_attribute, RawInputsForHeatmap
 from code_loader.contract.enums import LeapDataType, DataStateEnum, DataStateType, MetricDirection
 from code_loader.contract.responsedataclasses import DatasetTestResultPayload
@@ -37,6 +38,8 @@ class LeapBinder:
         self._encoder_names: List[str] = list()
         self._extend_with_default_visualizers()
 
+        self.batch_size_to_validate: Optional[int] = None
+
     def _extend_with_default_visualizers(self) -> None:
         self.set_visualizer(function=default_image_visualizer, name=DefaultVisualizer.Image.value,
                             visualizer_type=LeapDataType.Image)
@@ -103,7 +106,7 @@ class LeapBinder:
         if visualizer_type.value not in map_leap_data_type_to_visualizer_class:
             raise Exception(
                 f'The visualizer_type is invalid. current visualizer_type: {visualizer_type}, '  # type: ignore[attr-defined]
-                f'should be one of : {", ".join([arg.__name__ for arg in VisualizerCallableReturnType.__args__])}')
+                f'should be one of : {", ".join([arg.__name__ for arg in LeapData.__args__])}')
 
         func_annotations = function.__annotations__
         if "return" not in func_annotations:
@@ -113,10 +116,10 @@ class LeapBinder:
                   f"https://docs.python.org/3/library/typing.html")
         else:
             return_type = func_annotations["return"]
-            if return_type not in VisualizerCallableReturnType.__args__:  # type: ignore[attr-defined]
+            if return_type not in LeapData.__args__:  # type: ignore[attr-defined]
                 raise Exception(
                     f'The return type of function {function.__name__} is invalid. current return type: {return_type}, '  # type: ignore[attr-defined]
-                    f'should be one of : {", ".join([arg.__name__ for arg in VisualizerCallableReturnType.__args__])}')
+                    f'should be one of : {", ".join([arg.__name__ for arg in LeapData.__args__])}')
 
             expected_return_type = map_leap_data_type_to_visualizer_class[visualizer_type.value]
             if not issubclass(return_type, expected_return_type):
@@ -430,6 +433,7 @@ class LeapBinder:
     def check_handler(
             preprocess_response: PreprocessResponse, test_result: List[DatasetTestResultPayload],
             dataset_base_handler: Union[DatasetBaseHandler, MetadataHandler]) -> List[DatasetTestResultPayload]:
+        assert preprocess_response.sample_ids is not None
         raw_result = dataset_base_handler.function(preprocess_response.sample_ids[0], preprocess_response)
         handler_type = 'metadata' if isinstance(dataset_base_handler, MetadataHandler) else None
         if isinstance(dataset_base_handler, MetadataHandler) and isinstance(raw_result, dict):
@@ -471,4 +475,19 @@ class LeapBinder:
         self.check_handlers(preprocess_result)
         print("Successful!")
 
+    def set_batch_size_to_validate(self, batch_size: int):
+        self.batch_size_to_validate = batch_size
+
+    @staticmethod
+    def init():
+        available_functions = inspect.getmembers(sys.modules[__name__], inspect.isfunction)
+        for func_name, func in available_functions:
+            if 'tensorleap_custom_metric' in str(func):
+                try:
+                    func()
+                except:
+                    pass
+
+
+
 

code_loader-1.0.49.dev100/code_loader/inner_leap_binder/leapbinder_decorators.py

@@ -0,0 +1,56 @@
+from typing import Optional, Union
+
+import numpy as np
+
+from code_loader.contract.datasetclasses import CustomCallableInterfaceMultiArgs, \
+    CustomMultipleReturnCallableInterfaceMultiArgs, ConfusionMatrixCallableInterfaceMultiArgs
+from code_loader.contract.enums import MetricDirection
+from code_loader import leap_binder
+
+
+def tensorleap_custom_metric(name: str, direction: Optional[MetricDirection] = MetricDirection.Downward):
+    def decorating_function(
+            user_function: Union[CustomCallableInterfaceMultiArgs,
+                                 CustomMultipleReturnCallableInterfaceMultiArgs,
+                                 ConfusionMatrixCallableInterfaceMultiArgs]
+    ):
+
+        leap_binder.add_custom_metric(user_function, name, direction)
+
+        def _validate_custom_metric_input_args(*args, **kwargs):
+            for i, arg in enumerate(args):
+                assert isinstance(arg, np.ndarray), (f'tensorleap_custom_metric validation failed: '
+                                                     f'Argument #{i} should be a numpy array. Got {type(arg)}.')
+                if leap_binder.batch_size_to_validate:
+                    assert arg.shape[0] == leap_binder.batch_size_to_validate, \
+                        (f'tensorleap_custom_metric validation failed: Argument #{i} '
+                         f'first dim should be as the batch size. Got {arg.shape[0]} '
+                         f'instead of {leap_binder.batch_size_to_validate}')
+
+            for _arg_name, arg in kwargs.items():
+                assert isinstance(arg, np.ndarray), (f'tensorleap_custom_metric validation failed: '
+                                                     f'Argument {_arg_name} should be a numpy array. Got {type(arg)}.')
+                if leap_binder.batch_size_to_validate:
+                    assert arg.shape[0] == leap_binder.batch_size_to_validate, \
+                        (f'tensorleap_custom_metric validation failed: Argument {_arg_name} '
+                         f'first dim should be as the batch size. Got {arg.shape[0]} '
+                         f'instead of {leap_binder.batch_size_to_validate}')
+
+        def _validate_custom_metric_result(result):
+            assert isinstance(result, np.ndarray), (f'tensorleap_custom_metric validation failed: '
+                                                    f'The return type should be a numpy array. Got {type(result)}.')
+            assert len(result.shape) == 1, (f'tensorleap_custom_metric validation failed: '
+                                            f'The return shape should be 1D. Got {len(result.shape)}D.')
+            if leap_binder.batch_size_to_validate:
+                assert result.shape[0] == leap_binder.batch_size_to_validate, \
+                    f'tensorleap_custom_metric validation failed: The return len should be as the batch size.'
+
+        def inner(*args, **kwargs):
+            _validate_custom_metric_input_args(*args, **kwargs)
+            result = user_function(*args, **kwargs)
+            _validate_custom_metric_result(result)
+            return result
+
+        return inner
+
+    return decorating_function

{code_loader-1.0.49.dev9 → code_loader-1.0.49.dev100}/code_loader/leaploader.py

@@ -12,7 +12,7 @@ import numpy as np
 import numpy.typing as npt
 
 from code_loader.contract.datasetclasses import DatasetSample, DatasetBaseHandler, GroundTruthHandler, \
-    PreprocessResponse, VisualizerHandler, VisualizerCallableReturnType, CustomLossHandler, \
+    PreprocessResponse, VisualizerHandler, LeapData, CustomLossHandler, \
     PredictionTypeHandler, MetadataHandler, CustomLayerHandler, MetricHandler
 from code_loader.contract.enums import DataStateEnum, TestingSectionEnum, DataStateType, DatasetMetadataType
 from code_loader.contract.exceptions import DatasetScriptException
@@ -201,7 +201,7 @@ class LeapLoader:
         return all_dataset_base_handlers
 
     def run_visualizer(self, visualizer_name: str, input_tensors_by_arg_name: Dict[str, npt.NDArray[np.float32]],
-                       ) -> VisualizerCallableReturnType:
+                       ) -> LeapData:
         return self.visualizer_by_name()[visualizer_name].function(**input_tensors_by_arg_name)
 
     def run_heatmap_visualizer(self, visualizer_name: str, input_tensors_by_arg_name: Dict[str, npt.NDArray[np.float32]]

{code_loader-1.0.49.dev9 → code_loader-1.0.49.dev100}/code_loader/utils.py

@@ -10,7 +10,7 @@ from code_loader.contract.datasetclasses import SectionCallableInterface, Prepro
 
 
 def to_numpy_return_wrapper(encoder_function: SectionCallableInterface) -> SectionCallableInterface:
-    def numpy_encoder_function(idx: int, samples: PreprocessResponse) -> npt.NDArray[np.float32]:
+    def numpy_encoder_function(idx: Union[int, str], samples: PreprocessResponse) -> npt.NDArray[np.float32]:
         result = encoder_function(idx, samples)
         numpy_result: npt.NDArray[np.float32] = np.array(result)
         return numpy_result

{code_loader-1.0.49.dev9 → code_loader-1.0.49.dev100}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "code-loader"
-version = "1.0.49.dev9"
+version = "1.0.49.dev100"
 description = ""
 authors = ["dorhar <doron.harnoy@tensorleap.ai>"]
 license = "MIT"