def bootstrap_config(config_id, seed=-1):
"""Method to generate the config (using config id) and set seeds"""
config = get_config(config_id, experiment_id=0)
if seed > 0:
set_seed(seed=seed)
else:
set_seed(seed=config.general.seed)
return config
def bootstrap(config_id):
config_dict = get_config(config_id=config_id)
print(config_dict.log)
set_logger(config_dict)
write_message_logs("Starting Experiment at {}".format(
time.asctime(time.localtime(time.time()))))
write_message_logs("torch version = {}".format(torch.__version__))
write_config_log(config_dict)
set_seed(seed=config_dict.general.seed)
return config_dict
def run(config_id):
print("torch version = {}".format(torch.__version__))
config_dict = get_config(config_id=config_id)
set_seed(seed=config_dict.general.seed)
module_name = "codes.data.loader.loaders"
datatset = importlib.import_module(module_name).RolloutSequenceDataset(
config=config_dict, mode="train")
datatset.load_next_buffer()
for idx in range(1):
a = datatset.__getitem__(idx)[0][0]
show_tensor_as_image((a * 255).numpy().transpose(1, 2, 0))
def start(_config, _run):
config = Dict(_config)
set_seed(seed=config.seed)
run_experiment(config, _run)
def resume(config, experiment):
config = Dict(config)
set_seed(seed=config.general.seed)
run_experiment(config, experiment, resume=True)
def start(config, experiment):
config = Dict(config)
set_seed(seed=config.general.seed)
run_experiment(config, experiment)
def bootstrap(config_dict):
print(config_dict.log)
set_seed(seed=config_dict.general.seed)