gym.Env TypeError: 'type' object is not subscriptable

[Zeigar]

🐛 Bug Description

To Reproduce

Steps to reproduce the behavior:

1. from qlib.rl.trainer import Checkpoint, train

Expected Behavior

Screenshot

Environment

---

TypeError Traceback (most recent call last)
Cell In[20], line 4
2 from pathlib import Path
3 from typing import cast
----> 4 from qlib.rl.trainer import Checkpoint, train
7 NSTEPS = 10
9 trainer_kwargs = {
10 "max_iters": 10,
11 "finite_env_type": "dummy",
(...)
18 ],
19 }

File ~.conda\envs\TradeMaster\lib\site-packages\qlib\rl\trainer_init_.py:6
1 # Copyright (c) Microsoft Corporation.
2 # Licensed under the MIT License.
4 """Train, test, inference utilities."""
----> 6 from .api import backtest, train
7 from .callbacks import Checkpoint, EarlyStopping, MetricsWriter
8 from .trainer import Trainer

File ~.conda\envs\TradeMaster\lib\site-packages\qlib\rl\trainer\api.py:13
11 from qlib.rl.reward import Reward
12 from qlib.rl.simulator import InitialStateType, Simulator
---> 13 from qlib.rl.utils import FiniteEnvType, LogWriter
15 from .trainer import Trainer
16 from .vessel import TrainingVessel

File ~.conda\envs\TradeMaster\lib\site-packages\qlib\rl\utils_init_.py:5
1 # Copyright (c) Microsoft Corporation.
2 # Licensed under the MIT License.
4 from .data_queue import DataQueue
----> 5 from .env_wrapper import EnvWrapper, EnvWrapperStatus
6 from .finite_env import FiniteEnvType, vectorize_env
7 from .log import ConsoleWriter, CsvWriter, LogBuffer, LogCollector, LogLevel, LogWriter

File ~.conda\envs\TradeMaster\lib\site-packages\qlib\rl\utils\env_wrapper.py:52
47 action_history: list
48 reward_history: list
51 class EnvWrapper(
---> 52 gym.Env[ObsType, PolicyActType],
53 Generic[InitialStateType, StateType, ActType, ObsType, PolicyActType],
54 ):
55 """Qlib-based RL environment, subclassing gym.Env.
56 A wrapper of components, including simulator, state-interpreter, action-interpreter, reward.
57
(...)
90 Can be none when no trajectory is available.
91 """
93 simulator: Simulator[InitialStateType, StateType, ActType]

TypeError: 'type' object is not subscriptable

Note: User could run cd scripts && python collect_info.py all under project directory to get system information
and paste them here directly.

• Qlib version:
• Python version:
• OS (Windows, Linux, MacOS):
• Commit number (optional, please provide it if you are using the dev version):

Additional Notes

[Cadenza-Li]

Hi,you can try downgrade tianshou to 4.10, it works for me.
https://github.com/microsoft/qlib/blob/main/setup.py

  # In version 0.4.11 of tianshou, the code:
  # logits, hidden = self.actor(batch.obs, state=state, info=batch.info)
  # was changed in PR787,
  # which causes pytest errors(AttributeError: 'dict' object has no attribute 'info') in CI,
  # so we restricted the version of tianshou.
  # References:
  # https://github.com/thu-ml/tianshou/releases
  "tianshou<=0.4.10",
  "gym>=0.24",  # If you do not put gym at the end, gym will degrade causing pytest results to fail.