def main(): ''' main ''' # pdb.set_trace() # Debug step trainer = Trainer() params = trainer.params decode_fn = get_decode_fn(params.decode, params.max_decode_len) trainer.load_data(params.dataset, params.train, params.dev, params.test) trainer.setup_evalutator() if params.load and params.load != '0': if params.load == 'smart': start_epoch = trainer.smart_load_model(params.model) + 1 else: start_epoch = trainer.load_model(params.load) + 1 trainer.logger.info('continue training from epoch %d', start_epoch) trainer.setup_training() trainer.load_training(params.model) else: # start from scratch start_epoch = 0 trainer.build_model() if params.init: if os.path.isfile(params.init): trainer.load_state_dict(params.init) else: trainer.dump_state_dict(params.init) trainer.setup_training() # pdb.set_trace() # Debug step trainer.run(start_epoch, decode_fn=decode_fn)
def main(): """ main """ trainer = Trainer() params = trainer.params decode_fn = get_decode_fn(params.decode, params.max_decode_len, params.decode_beam_size) trainer.load_data(params.dataset, params.train, params.dev, params.test) trainer.setup_evalutator() if params.load and params.load != "0": if params.load == "smart": start_epoch = trainer.smart_load_model(params.model) + 1 else: start_epoch = trainer.load_model(params.load) + 1 trainer.logger.info("continue training from epoch %d", start_epoch) trainer.setup_training() trainer.load_training(params.model) else: # start from scratch start_epoch = 0 trainer.build_model() if params.init: if os.path.isfile(params.init): trainer.load_state_dict(params.init) else: trainer.dump_state_dict(params.init) trainer.setup_training() trainer.run(start_epoch, decode_fn=decode_fn)
def main(): """ main """ trainer = Trainer() params = trainer.params decode_fn = get_decode_fn(params.decode, params.max_decode_len, params.decode_beam_size) trainer.load_data(params.dataset, params.train, params.dev, params.test) trainer.setup_evalutator() assert params.load trainer.reload_and_test(params.model, params.load, params.bs, decode_fn)