chainerrl
chainerrl copied to clipboard
chainer
ChainerRL is a deep reinforcement learning library built on top of Chainer.
``` =================================== FAILURES =================================== _____ TestTrainAgentAsync_param_1_{max_episode_len=None, num_envs=2}.test ______ self = def test(self): steps = 50 outdir = tempfile.mkdtemp() agent = mock.Mock() agent.shared_attributes = [] def _make_env(process_idx, test): env = mock.Mock()...
https://arxiv.org/abs/1911.02140
This is aimed to resolve #572 by making everything that is passed to subprocesses pickle-able. Still WIP.
``` % python examples/gym/train_a3c_gym.py 2 --env CartPole-v0 /Users/fujita-rental/.local/share/virtualenvs/chainerrl-TtJB_mwx/lib/python3.8/site-packages/chainer/_environment_check.py:33: UserWarning: Accelerate has been detected as a NumPy backend library. vecLib, which is a part of Accelerate, is known not to work...
From 12.6 OpenAI Gym started to support VectorEnv. Since we have our own VectorEnv, we should check the compatibility and hopefully support theirs.
``` 2019-09-17 15:58:06.381228 STDOUT 2012] | chainerrl/tests/links_tests/test_stateless_recurrent_sequential.py .F... [100%] -- | -- 2019-09-17 15:58:06.381228 STDOUT 2012] | 2019-09-17 15:58:06.381229 STDOUT 2012] | =================================== FAILURES =================================== 2019-09-17 15:58:06.381230 STDOUT 2012] |...
``` ______________ TestCastObservation_param_0.test_cast_observation _______________ self = def test_cast_observation(self): env = chainerrl.wrappers.CastObservation( gym.make(self.env_id), dtype=self.dtype) rtol = 1e-3 if self.dtype == np.float16 else 1e-7 obs = env.reset() self.assertEqual(env.original_observation.dtype, np.float64) self.assertEqual(obs.dtype, self.dtype) >...
