0

我一直在使用 gpu 加速器在 kaggle 笔记本上训练强化学习模型，因为我只有一台 cpu 笔记本电脑可以使用。但是，当我尝试将模型导入本地 jupyter 笔记本时，为了渲染它，我收到以下错误。

AssertionError                            Traceback (most recent call last)
~\AppData\Local\Temp/ipykernel_13752/2941311152.py in <module>
      2 # device = torch.device('cpu')
      3 # model = PPO('CnnPolicy', env, verbose = 1, tensorboard_log = Log_Dir, learning_rate = 0.000001, n_steps = 512)
----> 4 PPO.load('./Training/Saved Models/Mario Models/best_model_100000', env)

c:\users\test\appdata\local\programs\python\python39\lib\site-packages\stable_baselines3\common\base_class.py in load(cls, path, env, device, custom_objects, print_system_info, force_reset, **kwargs)
    728         model.__dict__.update(data)
    729         model.__dict__.update(kwargs)
--> 730         model._setup_model()
    731 
    732         # put state_dicts back in place

c:\users\test\appdata\local\programs\python\python39\lib\site-packages\stable_baselines3\ppo\ppo.py in _setup_model(self)
    156 
    157         # Initialize schedules for policy/value clipping
--> 158         self.clip_range = get_schedule_fn(self.clip_range)
    159         if self.clip_range_vf is not None:
    160             if isinstance(self.clip_range_vf, (float, int)):

c:\users\test\appdata\local\programs\python\python39\lib\site-packages\stable_baselines3\common\utils.py in get_schedule_fn(value_schedule)
     89         value_schedule = constant_fn(float(value_schedule))
     90     else:
---> 91         assert callable(value_schedule)
     92     return value_schedule
     93 

AssertionError:

我尝试查看文档以纠正错误，但转移标准 pytorch 模型的指南不适用于稳定的基线模型。

4

0 回答 0