naming convention doesnt like env key, leakyrelu is no bueno, running quick experiment to see which option sets can work with our time left. the options that look 4s ahead are definitely out of the questions, possibly the 2s ones aswell

sgail-ppo-options-setobs2.py

@@ -21,11 +21,10 @@ import json
 
 DIR = os.path.dirname(os.path.abspath(__file__))
 option_list = [[(vel, time) for vel in [0, 1, 2, 4, 6, 8, 10] for time in [5]],
-                [(vel, time) for vel in [0, 1, 2, 5, 7.5, 10] for time in [5, 20]],
+                [(vel, time) for vel in [0, 2.5, 5, 7.5, 10] for time in [5, 10]],
-                [(vel, time) for vel in [0, 1, 2, 4, 6, 8, 10] for time in [5, 10, 20]], # was the best in training with single hidden layer, but very slow
+                [(vel, time) for vel in [0, 2.5, 5, 7.5, 10] for time in [5, 20]],
-                [(vel, time) for vel in [0, 1, 2, 5, 7.5, 10] for time in [5, 20, 40]], 
+                [(vel, time) for vel in [0, 2.5, 5, 10] for time in [5, 10, 20]],
-                [(vel, time) for vel in [0, 2, 5, 10] for time in [5, 10, 20]],
+                [(vel, time) for vel in [0, 3, 10] for time in [5, 10, 20]]
-                [(vel, time) for vel in [0, 3, 10] for time in [5, 20, 40]]
 ]
 
 obs_min = np.array([
@@ -58,11 +57,12 @@ def training_function(config):
                 collision_penalty=0
             ),
             check_collisions=True,
-            stop_on_collision=config['env']['stop_on_collision'], track=track,
+            stop_on_collision=config['trainenv']['stop_on_collision'], track=track,
         ), collision_distance=6, collision_penalty=100), lambda obs: (obs - obs_min) / (obs_max - obs_min + 1e-10))
-    ), options=option_list[config['policy']['option']],
+        ), options=option_list[config['policy']['option']],
-        safe_actions_collision_method=config['env']['safe_actions_collision_method'], 
+        safe_actions_collision_method=config['trainenv']['safe_actions_collision_method'], 
-        abort_unsafe_collision_method=config['env']['abort_unsafe_collision_method']) for _ in range(20)] for track in range(4)],[])
+        abort_unsafe_collision_method=config['trainenv']['abort_unsafe_collision_method'],
+        ) for _ in range(20)] for track in range(4)],[])
 
     env_fn = lambda i: envs[i]
 
@@ -118,9 +118,9 @@ def training_function(config):
         value=value,
         v_opt=v_opt,
         v_iters=config['value']['iterations_per_epoch'],
-        epochs=301,
+        epochs=20, #300, #200 FIXME
         rollout_episodes=60,
-        rollout_steps=60,
+        rollout_steps=60, 
         gamma=0.99,
         gae_lambda=0.9,
         clip_ratio=config['policy']['clip_ratio'],
@@ -137,19 +137,19 @@ def training_function(config):
 analysis = tune.run(
     training_function,
     config={
-        'env': {
+        'trainenv': {
-            'stop_on_collision': False,
+            'stop_on_collision': False, #tune.grid_search([False, True]),
-            'safe_actions_collision_method': 'circle',
+            'safe_actions_collision_method': 'circle', 
-            'abort_unsafe_collision_method': 'circle',
+            'abort_unsafe_collision_method': 'circle', 
         },
         'policy': {
             'learning_rate': 3e-4, # tune.grid_search([3e-4]),
             'learning_rate_decay': 1.0, #tune.grid_search([1.0]),
             'clip_ratio': 0.2, #tune.grid_search([0.2]),
             'iterations_per_epoch': 100, #tune.grid_search([100]),
-            'hidden_layer_size': tune.grid_search([10, 20, 40]),
+            'hidden_layer_size': 10, #tune.grid_search([10, 20, 40]), FIXME
-            'n_hidden_layers': tune.grid_search([2, 3, 4]),
+            'n_hidden_layers': 2, #tune.grid_search([2, 3, 4]), FIXME
-            'activation':tune.grid_search([torch.nn.LeakyReLU, torch.nn.Tanh]),
+            'activation':torch.nn.Tanh,
             'option': tune.grid_search(list(range(len(option_list))))
         },
         'value': {