allenlsj/physics_world_rl

Implementation of CogSci 2019 paper 'Active physical learning via reinforcement learning'

physics_world_rl's Issues

ERROR: "python2 Qlearning.py --mode 1"

When I use commend "python2 Qlearning.py --mode 1" to train a q-value function, it produces error "TypeError: init() takes exactly 9 arguments (8 given)". I find that in Qlearning.py, the last line is " new_env = physic_env(cond, mass_list, force_list, init_mouse, T, args.mode, prior)" , it misses the last parameter: reward_stop. When I try to add this parameter, it produces another error.So how to solve this problem correctly? I'm waiting for your answer eagerly!