search

stanfordnmbl / osim-rl

Reinforcement learning environments with musculoskeletal models
http://osim-rl.stanford.edu/
MIT License
877 stars 248 forks source link

ValueError while executing act_and_train in TRPO #224

Open mcmips opened 3 years ago

mcmips commented 3 years ago

Traceback (most recent call last): File "TRPO.py", line 169, in action = agent.act_and_train(obs, reward) File "C:\Anaconda3\envs\osim-rl2\lib\site-packages\chainerrl\agents\trpo.py", line 680, in act_and_train b_state = self.obs_normalizer(b_state, update=False) File "C:\Anaconda3\envs\osim-rl2\lib\site-packages\chainerrl\links\empirical_normalization.py", line 93, in call mean = xp.broadcast_to(self._mean, x.shape) File "", line 6, in broadcast_to File "C:\Anaconda3\envs\osim-rl2\lib\site-packages\numpy\lib\stride_tricks.py", line 180, in broadcast_to return _broadcast_to(array, shape, subok=subok, readonly=True) File "C:\Anaconda3\envs\osim-rl2\lib\site-packages\numpy\lib\stride_tricks.py", line 125, in _broadcast_to op_flags=['readonly'], itershape=shape, order='C') ValueError: operands could not be broadcast together with remapped shapes [original->remapped]: (1,158) and requested shape (1,160)

please help.. any hints are welcome..

length of obs returned me 160, so i assume req shape is (1,160)

I see that obs_space = env.observation_space returns a space size of (158)

mcmips commented 3 years ago

any hints, how can i fix this while using google colab for this code.