Merge branch 'development'

Author: weidler

.gitignore

@@ -1,20 +1,21 @@
+*venv/
 docs/analysis/*
 MUJOCO_LOG.TXT
 
-dexterity/monitor/static/experiments/
+angorapy/monitor/static/experiments/
 storage/
 !storage/saved_models
 storage/saved_models/*
 !/storage/saved_models/*.zip
 
-dexterity/analysis/monitor/experiments/
-dexterity/analysis/storage/
+angorapy/analysis/monitor/experiments/
+angorapy/analysis/storage/
 
 docs/monitor/experiments/
 docs/benchmarks/backup/*
 docs/benchmarks/*.pdf
 docs/storage/
-dexterity/monitor/experiments/
+angorapy/monitor/experiments/
 rnnconstruction/firsttrainhistory
 rnnconstruction/saved/
 

README.md

@@ -0,0 +1,4 @@
+import angorapy.agent
+from angorapy.common.wrappers import make_env
+from angorapy.models import get_model_builder
+from angorapy.utilities.monitoring import Monitor

angorapy/__init__.py

@@ -0,0 +1,2 @@
+from angorapy.agent.ppo_agent import PPOAgent
+from angorapy.agent.gather import BaseGatherer, Gatherer, EpsilonGreedyGatherer

angorapy/agent/__init__.py

@@ -7,7 +7,7 @@
 import tensorflow as tf
 from scipy.signal import lfilter
 
-from dexterity.common.const import NP_FLOAT_PREC
+from angorapy.common.const import NP_FLOAT_PREC
 
 
 def get_discounted_returns(reward_trajectory, discount_factor: float):

dexterity/agent/core.py renamed to angorapy/agent/core.py

@@ -10,10 +10,10 @@
 import tensorflow as tf
 from mpi4py import MPI
 
-from dexterity.common.const import STORAGE_DIR
-from dexterity.common.senses import Sensation
-from dexterity.utilities.datatypes import StatBundle
-from dexterity.common.data_buffers import ExperienceBuffer, TimeSequenceExperienceBuffer
+from angorapy.common.const import STORAGE_DIR
+from angorapy.common.senses import Sensation
+from angorapy.utilities.datatypes import StatBundle
+from angorapy.common.data_buffers import ExperienceBuffer, TimeSequenceExperienceBuffer
 
 
 def _float_feature(value):

dexterity/agent/dataio.py renamed to angorapy/agent/dataio.py

@@ -9,16 +9,16 @@
 import tensorflow as tf
 from gym.spaces import Box
 
-from dexterity.agent.core import estimate_episode_advantages
-from dexterity.agent.dataio import tf_serialize_example, make_dataset_and_stats, serialize_sample
-from dexterity.common.data_buffers import ExperienceBuffer, TimeSequenceExperienceBuffer
-from dexterity.common.policies import BasePolicyDistribution
-from dexterity.common.senses import Sensation
-from dexterity.common.wrappers import BaseWrapper, make_env
-from dexterity.common.const import STORAGE_DIR, DETERMINISTIC
-from dexterity.utilities.datatypes import StatBundle
-from dexterity.utilities.model_utils import is_recurrent_model
-from dexterity.utilities.util import add_state_dims, flatten, env_extract_dims
+from angorapy.agent.core import estimate_episode_advantages
+from angorapy.agent.dataio import tf_serialize_example, make_dataset_and_stats, serialize_sample
+from angorapy.common.data_buffers import ExperienceBuffer, TimeSequenceExperienceBuffer
+from angorapy.common.policies import BasePolicyDistribution
+from angorapy.common.senses import Sensation
+from angorapy.common.wrappers import BaseWrapper, make_env
+from angorapy.common.const import STORAGE_DIR, DETERMINISTIC
+from angorapy.utilities.datatypes import StatBundle
+from angorapy.utilities.model_utils import is_recurrent_model
+from angorapy.utilities.util import add_state_dims, flatten, env_extract_dims
 
 
 class BaseGatherer(abc.ABC):

dexterity/agent/gather.py renamed to angorapy/agent/gather.py

@@ -2,10 +2,10 @@
 
 import tensorflow as tf
 
-from dexterity.agent.core import extract_discrete_action_probabilities
-from dexterity.agent.ppo import loss
-from dexterity.common.policies import BasePolicyDistribution
-from dexterity.common.senses import Sensation
+from angorapy.agent.core import extract_discrete_action_probabilities
+from angorapy.agent.ppo import loss
+from angorapy.common.policies import BasePolicyDistribution
+from angorapy.common.senses import Sensation
 
 
 def learn_on_batch(batch,