l5kit.dataset.utils module

l5kit.dataset.utils.convert_str_to_fixed_length_tensor(string: str, max_length: int = 70) → torch.Tensor

Converts a string into a fixed length tensor of type torch.uint8.

Parameters

• string (str) – String to convert

• max_length (int) – Fixed length, default kMaxStrLength

Returns

A fixed length tensor of type torch.uint8 that encodes the string

Return type

torch.Tensor

l5kit.dataset.utils.move_to_device(data: Dict[str, torch.Tensor], device: torch.device) → Dict[str, torch.Tensor]

Move the data dict to a given torch device

Parameters

• data – the torch dict

• device – the device where to move each value of the dict

Returns

the torch dict on the new device

l5kit.dataset.utils.move_to_numpy(data: Dict[str, torch.Tensor]) → Dict[str, numpy.ndarray]

Move a torch dict into numpy (on cpu)

Parameters

data – the torch dict

Returns

the numpy dict