gt4sd.frameworks.torch package¶
Generic utils for pytorch.
Submodules:
Summary¶
Reference¶
- claim_device_name()[source]¶
Claim a device name.
- Return type
str- Returns
device name, if on GPU is available returns CPU.
- device_claim(device=None)[source]¶
Satidfy a device claim.
- Parameters
device (
Union[device,str,None]) – device where the inference is running either as a dedicated class or a string. If not provided is inferred.- Returns
the claimed device or a default one.
- Return type
torch.device
- get_device_from_tensor(tensor)[source]¶
Get the device from a tensor.
- Parameters
tensor (
Tensor) – a tensor.- Return type
device- Returns
the device.
- get_gpu_device_names()[source]¶
Get GPU device names as a list.
- Return type
List[str]- Returns
names of available GPU devices.
- map_tensor_dict(tensor_dict, device)[source]¶
Maps a dictionary of tensors to a specific device.
- Parameters
tensor_dict (
Dict[str,Tensor]) – A dictionary of tensors.device (
device) – The device to map the tensors to.
- Return type
Dict[str,Tensor]- Returns
A dictionary of tensors mapped to the device.