compute diversity

encode text with clip. inputs: ------- raw_text list (batch_size length) of strings with input text prom

encode text with clip. inputs: ------- raw_text list (batch_size length) of strings with input text prom

evaluate the model for one epoch.

take the folder of ood caption as input.

get the list of generated motion path.

add video to tensorboard

add video to tensorboard

return the beta tensor

return the betas_logscale tensor

the collate function for the evaluation motion dataset

the collate function for the evaluation motion dataset

the collate function for the evaluation motion dataset

the collate function for the combine motion dataset

the collate function for the human motion dataset. # TODO rename varible names in this function.

the collate function for the human motion dataset.

the collate function for the combine motion dataset

the collate function for the smal motion dataset

the collate function for the smpl motion dataset

end the timer and print the time

Calculate the Euclidean distance matrix between two sets of vectors using PyTorch. (X - X_train) * (X - X_train) = -2 * X * X_train + X * X

evaluate the model for one full epoch.

solution from chatgpt about how to find subject.

solution from chatgpt about how to find subject.

input: ------ clip_feature : [bs, 512] the clip feature latent_feature : [bs, 49, 7, 16] the

forward the latent

forward function for the motion mapping model inputs: ------- batch_data a dict of dict of tensor or wor

forward function for the loss calculator inputs: ------- batch_data a dict of dict of tensor or words

forward the latent

forward function for the motion clip model inputs: ------- batch_data: dict batch data

forward function for the loss calculator

inputs: ------- latent latent feature of shape (bs, num_frames / 4, 512)

TODO clean this code

forward function for the loss calculator inputs: ------- batch_data a dict of dict of tensor or words

forward function for the motion mapping model inputs: ------- batch_data a dict of dict of tensor or wor

forward function for the loss calculator inputs: ------- batch_data a dict of dict of tensor or words

forward function for the caption as input. go through smpl decoder. inputs: ------- batch_data a dict of

forward function for the caption as input. go through smpl decoder. inputs: ------- batch_data a dict of

generate motions from the data loader and network. mean while saving the motions inputs: ------- return: -------

generate motions from the data loader and network. mean while saving the motions inputs: ------- return: -------

generate the index of no tail vertices.

get dataloader from args

get dataloader from args

get the lambda at a certain step

recursively load all tensors in a nested dictionary structure onto the gpu.

the main function for visualizing the results

parse args

parse the args for finetuning

parse the args for finetuning

the process function for parallel processing to process the animal motions.

save the faces of smal model with no tail inputs: -------

return the scale tensor