|
from dizoo.mujoco.config.hopper_sac_data_generation_config import main_config, create_config |
|
from ding.entry import collect_demo_data, eval |
|
import torch |
|
import copy |
|
|
|
|
|
def eval_ckpt(args): |
|
config = copy.deepcopy([main_config, create_config]) |
|
eval(config, seed=args.seed, load_path=main_config.policy.learn.learner.hook.load_ckpt_before_run) |
|
|
|
|
|
def generate(args): |
|
config = copy.deepcopy([main_config, create_config]) |
|
state_dict = torch.load(main_config.policy.learn.learner.load_path, map_location='cpu') |
|
collect_demo_data( |
|
config, |
|
collect_count=main_config.policy.other.replay_buffer.replay_buffer_size, |
|
seed=args.seed, |
|
expert_data_path=main_config.policy.collect.save_path, |
|
state_dict=state_dict |
|
) |
|
|
|
|
|
if __name__ == "__main__": |
|
import argparse |
|
|
|
parser = argparse.ArgumentParser() |
|
parser.add_argument('--seed', '-s', type=int, default=0) |
|
args = parser.parse_args() |
|
|
|
eval_ckpt(args) |
|
generate(args) |
|
|