Rewrite state/reward normalizer

utils/config.py

@@ -3,6 +3,7 @@
 # Permission given to modify the code as long as you keep this        #
 # declaration at the top                                              #
 #######################################################################
+from .normalizer import *
 
 class Config:
     def __init__(self):
@@ -22,22 +23,18 @@ class Config:
         self.exploration_steps = 0
         self.logger = None
         self.history_length = 1
-        self.test_interval = 0
-        self.test_repetitions = 50
         self.double_q = False
         self.tag = 'vanilla'
         self.num_workers = 1
-        self.worker = None
         self.update_interval = 1
-        self.gradient_clip = 40
+        self.gradient_clip = 0.5
         self.entropy_weight = 0.01
         self.use_gae = False
         self.gae_tau = 1.0
         self.noise_decay_interval = 0
         self.target_network_mix = 0.001
-        self.action_shift_fn = lambda a: a
-        self.reward_shift_fn = lambda r: r
-        self.reward_weight = 1
+        self.state_normalizer = RescaleNormalizer()
+        self.reward_normalizer = RescaleNormalizer()
         self.hybrid_reward = False
         self.episode_limit = 0
         self.min_memory_size = 200