|
@ -89,21 +89,8 @@ def main(): |
|
|
imageAndVideoCallback = ImageRecorderCallback(eval_env, render_freq, n_eval_episodes=1, evaluation_method=evaluate_policy, log_dir=log_dir, deterministic=True, verbose=0) |
|
|
imageAndVideoCallback = ImageRecorderCallback(eval_env, render_freq, n_eval_episodes=1, evaluation_method=evaluate_policy, log_dir=log_dir, deterministic=True, verbose=0) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
model.learn(steps,callback=[imageAndVideoCallback, InfoCallback()]) |
|
|
|
|
|
|
|
|
|
|
|
#vec_env = model.get_env() |
|
|
|
|
|
#obs = vec_env.reset() |
|
|
|
|
|
#terminated = truncated = False |
|
|
|
|
|
#while not terminated and not truncated: |
|
|
|
|
|
# action_masks = None |
|
|
|
|
|
# action, _states = model.predict(obs, action_masks=action_masks) |
|
|
|
|
|
# print(action) |
|
|
|
|
|
# obs, reward, terminated, truncated, info = env.step(action) |
|
|
|
|
|
# # action, _states = model.predict(obs, deterministic=True) |
|
|
|
|
|
# # obs, rewards, dones, info = vec_env.step(action) |
|
|
|
|
|
# vec_env.render("human") |
|
|
|
|
|
# time.sleep(0.2) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
model.learn(steps,callback=[imageAndVideoCallback, InfoCallback(), evalCallback]) |
|
|
|
|
|
model.save(f"{log_dir}/{expname(args)}") |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
if __name__ == '__main__': |
|
|
if __name__ == '__main__': |
|
|