Issues with retraining from saved agent

[AvisP]

System information

• Grid2op version: 1.8.1
• l2rpn-baselines version: 0.6.0.post1
• System: mac osx
• stable-baseline 3 version 1.7.0
• Baseline concerned: PPO_SB3

Bug description

When I am trying to resume training from a saved agent, I am getting some errors. The saved agent however is working properly with evaluate function

How to reproduce

import re
import copy
import grid2op
from grid2op.Reward import LinesCapacityReward  # or any other rewards
from grid2op.Chronics import MultifolderWithCache  # highly recommended
from lightsim2grid import LightSimBackend  # highly recommended for training !
from l2rpn_baselines.PPO_SB3 import train, evaluate

env_name = "l2rpn_case14_sandbox"
obs_attr_to_keep = ["day_of_week", "hour_of_day", "minute_of_hour", "prod_p", "prod_v", "load_p", "load_q",
                    "actual_dispatch", "target_dispatch", "topo_vect", "time_before_cooldown_line",
                    "time_before_cooldown_sub", "rho", "timestep_overflow", "line_status",
                    "storage_power", "storage_charge"]
act_attr_to_keep = ["redispatch"]

env = grid2op.make(env_name,
                    reward_class=LinesCapacityReward,
                    backend=LightSimBackend(),
                    chronics_class=MultifolderWithCache)
env.chronics_handler.real_data.set_filter(lambda x: re.match(".*00$", x) is not None)
env.chronics_handler.real_data.reset()


train(env,
        iterations=1000,  # any number of iterations you want
        logs_dir="./logs/PPO_SB3_test",  # where the tensorboard logs will be put
        save_path="./saved_model/PPO_SB3_test",  # where the NN weights will be saved
        name="Reload_test",  # name of the baseline
        net_arch=[200, 200, 200],  # architecture of the NN
        obs_attr_to_keep=copy.deepcopy(obs_attr_to_keep),
        act_attr_to_keep=copy.deepcopy(act_attr_to_keep),
        normalize_obs=True,
        )

evaluate(env,
            nb_episode=3,
            load_path="./saved_model/PPO_SB3_test/",  # should be the same as what has been called in the train function !
            name="Reload_test",  # should be the same as what has been called in the train function !
            logs_path = "./logs/PPO_SB3/",
            nb_process=1,
            verbose=True,
            )

train(env,
        iterations=1000,  # any number of iterations you want
        logs_dir="./logs/PPO_SB3_test",  # where the tensorboard logs will be put
        load_path="./saved_model/PPO_SB3_test/Reload_test",
        save_path="./saved_model/PPO_SB3_test",  # where the NN weights will be saved
        name="Reload_test.zip",  # name of the baseline
        obs_attr_to_keep=copy.deepcopy(obs_attr_to_keep),
        act_attr_to_keep=copy.deepcopy(act_attr_to_keep),
        normalize_obs=True,
        )

Current output

I am getting the following error message

/Users/paula/Desktop/Projects/venvs/L2PRN/lib/python3.10/site-packages/grid2op/gym_compat/box_gym_obsspace.py:765: UserWarning: The normalization of attribute "[False False False False False False]" cannot be performed entirely as there are some non finite value, or `high == `low` for some components.
  warnings.warn(f"The normalization of attribute \"{both_finite}\" cannot be performed entirely as "
/Users/paula/Desktop/Projects/venvs/L2PRN/lib/python3.10/site-packages/grid2op/gym_compat/box_gym_obsspace.py:765: UserWarning: The normalization of attribute "[False False False False False False False False False False False]" cannot be performed entirely as there are some non finite value, or `high == `low` for some components.
  warnings.warn(f"The normalization of attribute \"{both_finite}\" cannot be performed entirely as "
/Users/paula/Desktop/Projects/venvs/L2PRN/lib/python3.10/site-packages/grid2op/gym_compat/box_gym_obsspace.py:765: UserWarning: The normalization of attribute "[False False False False False False False False False False False False
 False False False False False False False False]" cannot be performed entirely as there are some non finite value, or `high == `low` for some components.
  warnings.warn(f"The normalization of attribute \"{both_finite}\" cannot be performed entirely as "
Traceback (most recent call last):
  File "/Users/paula/Desktop/Projects/RL Practice/L2RPN Aspen/Demo Notebooks/PPO_SB3_train_reload.py", line 45, in <module>
    train(env,
  File "/............../L2PRN/lib/python3.10/site-packages/l2rpn_baselines/PPO_SB3/train.py", line 305, in train
    agent.nn_model.learn(total_timesteps=iterations,
  File "/............../L2PRN/lib/python3.10/site-packages/stable_baselines3/ppo/ppo.py", line 307, in learn
    return super().learn(
  File "/............../L2PRN/lib/python3.10/site-packages/stable_baselines3/common/on_policy_algorithm.py", line 236, in learn
    total_timesteps, callback = self._setup_learn(
  File "/............../L2PRN/lib/python3.10/site-packages/stable_baselines3/common/base_class.py", line 408, in _setup_learn
    self._last_obs = self.env.reset()  # pytype: disable=annotation-type-mismatch
AttributeError: 'NoneType' object has no attribute 'reset'