Exemple #1
0
from replay_buffer import ReplayBuffer
import json

with open("param.json", "r") as f:
    param = json.load(f)

config = param

memory = ReplayBuffer((8, ), (1, ), config["expert_buffer_size"],
                      config["device"])
memory.load_memory(config["buffer_path"])

memory_expert = ReplayBuffer((8, ), (1, ), 100000, config["device"])

for idx in range(memory.idx):
    memory_expert.add_expert(memory.obses[idx], memory.actions[0],
                             memory.rewards[idx], memory.next_obses[idx],
                             memory.not_dones[idx],
                             memory.not_dones_no_max[idx])
    print(memory_expert.idx)

memory_expert.save_memory("all_action_expert")