added more hyperparams sets for lstms

hyperpars/rppo-asm2o.yml

@@ -16,24 +16,92 @@ id: "1"
 repo: "cboettig/rl-ecology"
 save_path: "/home/rstudio/rl4fisheries/saved_agents"
 
-# algo hyperpars taken from BipedalWalker-v3 hyperpars:
+# algo hyperpars taken from:
 # https://github.com/DLR-RM/rl-baselines3-zoo/blob/master/hyperparams/ppo_lstm.yml
+
+# # BIPEDAL WALKER
+
+# algo_config:
+#     # normalize: True # not clear what this one actually does -- from the source code it seems to 'activate' VecNormalize, but more care & examination needed
+#     policy: 'MlpLstmPolicy'
+#     tensorboard_log: "/home/rstudio/logs"
+#     n_steps: 256
+#     batch_size: 256
+#     gae_lambda: 0.95
+#     gamma: 0.999
+#     n_epochs: 10
+#     ent_coef: 0.0
+#     learning_rate: !!float 3e-4
+#     clip_range: 0.18
+#     policy_kwargs: "dict(
+#                     ortho_init=False,
+#                     activation_fn=torch.nn.ReLU,
+#                     lstm_hidden_size=64,
+#                     enable_critic_lstm=True,
+#                     net_arch=dict(pi=[64], vf=[64])
+#                   )"
+
+# # HALF CHEETAH V4
+
 algo_config:
-    # normalize: True # not clear what this one actually does -- from the source code it seems to 'activate' VecNormalize, but more care & examination needed
-    policy: 'MlpLstmPolicy'
-    tensorboard_log: "/home/rstudio/logs"
-    n_steps: 256
-    batch_size: 256
-    gae_lambda: 0.95
-    gamma: 0.999
-    n_epochs: 10
-    ent_coef: 0.0
-    learning_rate: !!float 3e-4
-    clip_range: 0.18
-    policy_kwargs: "dict(
+  normalize: true
+  n_envs: 1
+  policy: 'MlpLstmPolicy'
+  tensorboard_log: "/home/rstudio/logs"
+  batch_size: 64
+  n_steps: 512
+  gamma: 0.98
+  learning_rate: 2.0633e-05
+  ent_coef: 0.000401762
+  clip_range: 0.1
+  n_epochs: 20
+  gae_lambda: 0.92
+  max_grad_norm: 0.8
+  vf_coef: 0.58096
+  policy_kwargs: "dict(
+                    log_std_init=-2,
                     ortho_init=False,
-                    activation_fn=torch.nn.ReLU,
-                    lstm_hidden_size=64,
-                    enable_critic_lstm=True,
-                    net_arch=dict(pi=[64], vf=[64])
-                  )"
+                    activation_fn=nn.ReLU,
+                    net_arch=dict(pi=[256, 256], vf=[256, 256])
+                  )"
+
+
+
+# # INVERTED PENDULUM
+
+# algo_config:
+#   n_envs: 8
+#   tensorboard_log: "/home/rstudio/logs"
+#   policy: 'MlpLstmPolicy'
+#   n_steps: 2048
+#   batch_size: 64
+#   gae_lambda: 0.95
+#   gamma: 0.99
+#   n_epochs: 10
+#   ent_coef: 0.0
+#   learning_rate: 2.5e-4
+#   clip_range: 0.2
+
+
+# # MOUNTAIN CAR NO VEL
+
+# algo_config:
+#   n_envs: 8
+#   tensorboard_log: "/home/rstudio/logs"
+#   policy: 'MlpLstmPolicy'
+#   batch_size: 256
+#   n_steps: 1024
+#   gamma: 0.9999
+#   learning_rate: !!float 7.77e-05
+#   ent_coef: 0.00429
+#   clip_range: 0.1
+#   n_epochs: 10
+#   gae_lambda: 0.9
+#   max_grad_norm: 5
+#   vf_coef: 0.19
+#   use_sde: True
+#   sde_sample_freq: 8
+#   policy_kwargs: "dict(log_std_init=0.0, ortho_init=False,
+#                        lstm_hidden_size=32,
+#                        enable_critic_lstm=True,
+#                        net_arch=dict(pi=[64], vf=[64]))"