dgs.utils.torchtools

Tools for handling recurring torch tasks. Mostly taken from the torchreid package

Module Functions

close_all_layers(model)	Closes / Freezes all layers in this model, e.g., for evaluation.
close_specified_layers(model, close_layers)	Close / Freeze the specified layers in the given model for training while keeping all other layers unchanged.
configure_torch_module(orig_cls, *_args, ...)	Decorator to decorate a class, which has to be a child of torch.nn.Module and the BaseModule! The decorator will then call BaseModule.configure_torch_model on themselves after initializing the original class.
get_model_from_module(module)	Given either a torch module or an instance of BaseModule, return a torch module.
init_instance_params(instance)	Given a module instance, initialize a single instance.
init_model_params(module)	Given a torch module, initialize the model parameters using some default weights.
load_checkpoint(fpath[, device])	Load a given checkpoint.
load_pretrained_weights(model, weight_path)	Loads pretrianed weights to model.
open_all_layers(model)	Opens all layers in this model for training.
open_specified_layers(model, open_layers[, ...])	Opens the specified layers in the given model for training while keeping all other layers unchanged or frozen.
resume_from_checkpoint(fpath, model[, ...])	Resumes training from a checkpoint.
save_checkpoint(state, save_dir, *[, ...])	Save a given checkpoint.
set_bn_to_eval(module)	Sets BatchNorm layers to eval mode.
torch_memory_analysis(f[, file_name, max_events])	A decorator for torch memory analysis using torch.cuda.memory._record_memory_history().
torch_memory_analysis_win(f[, file_name, ...])	A decorator for torch memory analysis using torch.cuda.memory._record_memory_history_legacy() that works on Windows machines.