'AgentPPO' object has no attribute 'traj_list'

[HarryLukKF]

There is an error when I try to excute the train function in Demo_MultiCrypto_Trading.ipynb

AttributeError: 'AgentPPO' object has no attribute 'traj_list'

Please kindly advise how should I fix this error? Many thanks in advance

[altosssss]

same here!

[AlexLiddiard]

Also encountered this issue when running Demo_MultiCrypto_Trading.ipynb I spent quite a long time debugging it and finally got it working by changing these things:

1. elegantrl/agents/AgentBase.py line 43, insert:

   self.traj_list = [
   [list() for _ in range(4 if self.if_off_policy else 5)]
   for _ in range(self.env_num)
   ]

2. elegantrl/agents/AgentPPO.py line 88, change to: return self.convert_trajectory(traj_list, last_done), step_i

3. elegantrl/train/evaluator.py line 60, comment it out

4. finrl-meta/agents/elegantrl_models.py line 112, change to: environment.cash

5. finrl-meta/agents/elegantrl_models.py line 114 change to: environment.price_array[environment.time] * environment.stocks

[altosssss]

Also encountered this issue when running Demo_MultiCrypto_Trading.ipynb I spent quite a long time debugging it and finally got it working by changing these things:

1. elegantrl/agents/AgentBase.py line 43, insert:

self.traj_list = [
  [list() for _ in range(4 if self.if_off_policy else 5)]
  for _ in range(self.env_num)
]

2. elegantrl/agents/AgentPPO.py line 88, change to: return self.convert_trajectory(traj_list, last_done), step_i
3. elegantrl/train/evaluator.py line 60, comment it out
4. finrl-meta/agents/elegantrl_models.py line 112, change to: environment.cash
5. finrl-meta/agents/elegantrl_models.py line 114 change to: environment.price_array[environment.time] * environment.stocks

Thanks Alex. Wondering if you tried run it on Colab?

[bassemfg]

Also encountered this issue when running Demo_MultiCrypto_Trading.ipynb I spent quite a long time debugging it and finally got it working by changing these things:

1. elegantrl/agents/AgentBase.py line 43, insert:

self.traj_list = [
    [list() for _ in range(4 if self.if_off_policy else 5)]
    for _ in range(self.env_num)
]

2. elegantrl/agents/AgentPPO.py line 88, change to: return self.convert_trajectory(traj_list, last_done), step_i
3. elegantrl/train/evaluator.py line 60, comment it out
4. finrl-meta/agents/elegantrl_models.py line 112, change to: environment.cash
5. finrl-meta/agents/elegantrl_models.py line 114 change to: environment.price_array[environment.time] * environment.stocks

Thanks Alex. Wondering if you tried run it on Colab?

Yes, I am getting the same error both on ubuntu and google Colab

[mhdmyz]

Same problem here! Solution mentioned by @AlexLiddiard solved the issue. But why it hasn't been fixed yet on elegantrl?!

[bassemfg]

Also, if when I try other agents like ddpg or td3 or sac, I get another error in the same line saying that init() expects 4 arguments but got 5. I spent a couple of hours trying to debug it but no luck. Anyone has any ideas?

[shixun404]

I modified the question related to 'traj_list'. The file ' demo_A2C_PPO.py ' and ' demo_PPO_H.py ' could run correctly now!

[shixun404]

By the way, demo_DDPG_TD3_SAC.py and demo_DDPG_H.py run correctly as well~