chop.actions

chop.actions.train

chop.actions.train.train(model, model_info, data_module, dataset_info, task, optimizer, learning_rate, weight_decay, scheduler_args, plt_trainer_args, auto_requeue, save_path, visualizer, load_name, load_type)

chop.actions.test

chop.actions.test.test(model, model_info, data_module, dataset_info, task, optimizer, learning_rate, weight_decay, plt_trainer_args, auto_requeue, save_path, visualizer, load_name, load_type)

chop.actions.search

chop.actions.search.search.parse_search_config(search_config)

Parse search config from a dict or a toml file and do sanity check.

— The search config must consist of two parts: strategy and search_space.

chop.actions.search.search.search(model: Module, model_info, task: str, dataset_info, data_module, search_config: dict | PathLike, save_path: PathLike, accelerator: str, load_name: PathLike = None, load_type: str = None, visualizer=None)

Parameters:

model – the model to be searched