cleaning

README.md

@@ -8,14 +8,11 @@
 ## Available algorithms:
 - Behavior Cloning as described in [ArXiv](https://arxiv.org/abs/1903.10527) `python3 train.py cfg/cloning.cfg`
 - DAGGER imitation learning `python3 train.py cfg/dagger.cfg`
-- Deep Deterministic Policy Gradients (TODO)
+
+## To test:
+- `python3 test_model.py cfg/dagger.cfg`
 
 ## Other code:
 - `python3 flocking_gym_test.py` provides test code for the Gym Flock environments
 
-## To Do
-- Plot centralized and decentralized baselines
-- Obstacle avoidance
-- Follow 1 or 2 leaders
-- Play with control costs
-- DDPG
+

cfg/cloning.cfg

@@ -9,6 +9,11 @@ updates_per_step = 200
 seed = 11
 actor_lr = 5e-5
 
+n_train_episodes = 400
+beta_coeff = 0.993
+test_interval = 40
+n_test_episodes = 20
+
 # architecture parameters
 k = 2
 hidden_size = 32
@@ -22,7 +27,9 @@ comm_radius = 1.0
 n_agents = 100
 n_actions = 2
 n_states = 6
-debug = False
+debug = True
 header = reward
+dt = 0.01
 
-[test]
+[test]
+fname = cloning_k3

cfg/dagger.cfg

@@ -33,4 +33,4 @@ dt = 0.01
 
 [test]
 
-fname = k3
+fname = dagger_k3