init embeddings uniformly

Load result file and return a result api object. :param resFile (str) : file name of result file :return: res (obj) : resul

returns mask of the tokens

:param d_model: dimension of the model :param length: length of positions :return: length*d_model position matrix

:param d_model: dimension of the model :param height: height of the positions :param width: width of the positions :r

create logdirs, check dataset, seed pseudo-random generators

Gets object ID from out, maps the object ID to a class instance segmentation mask from the CV model and adds the binary mask predicted by th

Processes the outputs of the CV model to return a sings matching instance mask prediction corresponding to the object instance ID.

process a single trajectory, return it as a dict

assign the largest output vocab to all datasets, compute embedding sizes

Resize pos_embed weights. Resize pos_embed using bicubic interpolate method. Args: pos_embed (torch.Tensor): Position embe

training loop

This function disables printing when not in master process

This function disables printing when not in master process

cast values to torch tensors, put them to the correct device and pad sequences

train a network using an lmdb dataset

generate upper triangular matrix filled with ones

Upload file to s3 bucket.

wrap datasets with torch loaders