Open louise-sawyer opened 2 years ago
Thank you for helping us to find these bugs. I am fixing these bugs.
if_use_per
. We will remove this attribute.wandb.init()
or remove the code use wandb
.init() takes 4 positional arguments but 5 were given
. The layer_num
cause this bug. We will fix it.Interested to know whether this bug/issue also happen for this file: https://github.com/AI4Finance-Foundation/ElegantRL/blob/master/demo_IsaacGym.py
Interested to know whether this bug/issue also happen for this file: https://github.com/AI4Finance-Foundation/ElegantRL/blob/master/demo_IsaacGym.py
It doesn't also happen. Because when I run the file, it will occur an error in config.py like can't pickle isaacgym._bindings.linux-x86_64.gym_37.Gym objects at line 206 code "env = deepcopy(env)".
In train/run.py, line 35, "if_use_per=args.if_use_per", Argument has no attribute "if_use_per".
In train/evaluator.py, coder should import wandb and write "wandb.init()".
In demo_IsaacGym.py, if I change the Agent such as AgentSAC, there will be an typeerror in agents/AgentBse.py that init() takes 4 positional arguments but 5 were given.