get_env_attr not working in SubprocVectorEnv? about tianshou HOT 2 OPEN

AttributeError: Can't pickle local object 'config_valve..' did you use lambda expression for config_value? Could you provide how you set config_value?

from tianshou.

AttributeError: Can't pickle local object 'config_valve..' did you use lambda expression for config_value? Could you provide how you set config_value?

What does the config_value refer to? If you mean the paramter env_fns in SubprocVectorEnv, I was actually using a lambda expression, as follows:
`

train_env_ = [lambda port=i: get_env(
    problem_config(
        problem=args.problem,
        prefix=args.prefix + '_train_port%d' % port,
        state_dir='state_dir',
        result_dir='result_dir',
        error_dir='error_dir',
        msh_dir='msh_dir',
        max_len=args.max_len,
        penalty=args.penalty,
        phase='train',
        local=args.prefix + '_train_port%d' % port,
        port=port,
    )
) for i in range(2036, 2036 + args.training_num)]

`

from tianshou.