search

TianhongDai / reinforcement-learning-algorithms

This repository contains most of pytorch implementation based classic deep reinforcement learning algorithms, including - DQN, DDQN, Dueling Network, DDPG, SAC, A2C, PPO, TRPO. (More algorithms are still in progress)
662 stars 109 forks source link

SAC Agent still wrong with "tuple index out of range" use "MountainCar-v0" #10

Open StewartTsai opened 2 years ago

StewartTsai commented 2 years ago

I try to use "MountainCar-v0" env in sac agent but still wrong with "tuple index out of range" Could you tell me how to fix it ? Thanks

Traceback (most recent call last): File "D:\RL_library/sac/train.py", line 14, in sac_trainer = sac_agent(env, args) File "D:RL_library\sac\sac_agent.py", line 30, in init self.qf1 = flatten_mlp(self.env.observation_space.shape[0], self.args.hidden_size, self.env.action_space.shape[0]) IndexError: tuple index out of range

TianhongDai commented 2 years ago

@StewartTsai Hi - my SAC implementation aims to solve the MuJoCo tasks with continuous control, thus I haven't tested MountainCar-v0 in the current version. But I will fix this problem in these days and notify you in this issue.