Flatten nested tuples and print tensors with their shape and other informations.

Run the basic layers match stage of the conversion process.

Run the init stage of the conversion process.

Run the shape and layers match stage of the conversion process.

Whether to show only the tag when printing the module. This is useful to distinguish between Chain subclasses that override their forward fro

Verify that the source and target models have the same number of basic layers.

Verify that the source and target models do not have missing basic layers.

Adaptive instance normalization. Adjust the reference features to have the similar color and illuminations as those in the degradate features

Patch refiners ScaledDotProductAttention so that it uses sliced attention It reduces peak memory usage.

Monkey patches the default python logger to have timed logs

🤖🧠 ImaginAIry. Pythonic generation of images via AI

Align and crop face with landmarks. The output_size and transform_size are based on width. The height is adjusted based on shrink_ratio_h/shr

Performs all_gather operation on the provided tensors. Graph remains connected for backward grad computation.

Analyze image and return some suggestions of crops (coordinates). This implementation / algorithm is really slow for large images.

Calculate the aspect ratio of a given width and height. Args: width (int): The width dimension. height (int): The height dimension.

Assigns detections to tracked object (both represented as bounding boxes) Returns: 3 lists of matches, unmatched_detections and unmatched

Returns the names of available CLIP models.

Create a 1D, 2D, or 3D average pooling module.

Create a 1D, 2D, or 3D average pooling module.

Return a dictionary of basic attributes of the module. Basic attributes are public attributes made of basic types (int, float, str, bool) or

Decode locations from predictions using priors to undo the encoding we did for offset regression at train time. Args: b_loc (tensor):

Decode landm from predictions using priors to undo the encoding we did for offset regression at train time. Args: pre (tensor): landm

Arguments: frames: a list of PIL.Image, or torch.Tensor(shape=[n, h, w, c], type=np.float32, BGR format).

Return the names of available phrase-lists.

Check that the torch version is compatible with ImaginAIry. https://github.com/brycedrennan/imaginAIry/issues/329

Chooses the best weights file from a list of filenames Prefers safetensors format and fp16 dtype

Colorizes black and white images or re-colors existing images. Optionally adds a caption. Args: img (Image): The image to be coloriz

Compare the outputs of the source and target models. - `source_args`: The arguments to pass to the source model it can be either a t

Convert applicable model parameters to fp16.

Create a map of node numbers to paths in a nested dictionary. Include all nodes, not just leaves. Example: data = {"a": {"b": ["c",

Function: ---------- Convert Transform Matrix 'trans' into 'cv2_trans' which could be directly used by cv2.warpAffine():

Decode locations from predictions using priors to undo the encoding we did for offset regression at train time. Args: loc (tensor): lo

Decode landm from predictions using priors to undo the encoding we did for offset regression at train time. Args: pre (tensor): landm