gaspersic/baselines
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

lstm network builders using tf lstm

Browse Source
This commit is contained in:
Peter Zhokhov
2018-08-10 14:23:45 -07:00
parent 217b111c88
commit dbcc4e0252
2 changed files with 7 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
baselines/common/tests/test_fixed_sequence.py
View File

@@ -6,7 +6,8 @@ from baselines.run import get_learn_function
common_kwargs = dict( common_kwargs = dict(
seed=0, seed=0,
total_timesteps=50000, total_timesteps=20000,
nlstm=64
) )
learn_kwargs = { learn_kwargs = {
@@ -19,7 +20,7 @@ learn_kwargs = {
alg_list = learn_kwargs.keys() alg_list = learn_kwargs.keys()
rnn_list = ['lstm'] rnn_list = ['lstm', 'tflstm', 'tflstm_static']
@pytest.mark.slow @pytest.mark.slow
@pytest.mark.parametrize("alg", alg_list) @pytest.mark.parametrize("alg", alg_list)
@@ -41,11 +42,11 @@ def test_fixed_sequence(alg, rnn):
**kwargs **kwargs
) )
simple_test(env_fn, learn, 0.7) simple_test(env_fn, learn, 0.3)
if __name__ == '__main__': if __name__ == '__main__':
test_fixed_sequence('ppo2', 'lstm') test_fixed_sequence('ppo2', 'tflstm')

3
baselines/common/tests/util.py
View File

@@ -2,6 +2,7 @@ import tensorflow as tf
import numpy as np import numpy as np
from gym.spaces import np_random from gym.spaces import np_random
from baselines.common.vec_env.dummy_vec_env import DummyVecEnv from baselines.common.vec_env.dummy_vec_env import DummyVecEnv
from baselines.bench.monitor import Monitor
N_TRIALS = 10000 N_TRIALS = 10000
N_EPISODES = 100 N_EPISODES = 100
@@ -10,7 +11,7 @@ def simple_test(env_fn, learn_fn, min_reward_fraction, n_trials=N_TRIALS):
np.random.seed(0) np.random.seed(0)
np_random.seed(0) np_random.seed(0)
env = DummyVecEnv([env_fn]) env = DummyVecEnv([lambda: Monitor(env_fn(), None, allow_early_resets=True)])
with tf.Graph().as_default(), tf.Session(config=tf.ConfigProto(allow_soft_placement=True)).as_default(): with tf.Graph().as_default(), tf.Session(config=tf.ConfigProto(allow_soft_placement=True)).as_default():