chore: release v0.3.0

motrix_rl/template/skrl_config.yaml

@@ -0,0 +1,75 @@
+seed: 42
+
+# Models are instantiated using skrl's model instantiator utility
+# https://skrl.readthedocs.io/en/latest/api/utils/model_instantiators.html
+models:
+    separate: False
+    policy: # gaussian model
+        class: "GaussianMixin"
+        clip_actions: True
+        clip_log_std: True
+        initial_log_std: 0.0
+        min_log_std: -20.0
+        max_log_std: 2.0
+        input: "STATES"
+        hiddens: [32, 32]
+        hidden_activation: ["elu", "elu"]
+        output: "ACTIONS"
+        output_activation: "tanh"
+        output_scale: 1.0
+    value: # deterministic model
+        class: "DeterministicMixin"
+        clip_actions: False
+        input: "STATES"
+        hiddens: [32, 32]
+        hidden_activation: ["elu", "elu"]
+        output: "ONE"
+        output_activation: ""
+        output_scale: 1.0
+
+# Memory
+# https://skrl.readthedocs.io/en/latest/api/memories/random.html
+memory:
+    class: "RandomMemory"
+    memory_size: -1 # -1: automatically determined value
+
+# PPO agent configuration (field names are from PPO_DEFAULT_CONFIG)
+# https://skrl.readthedocs.io/en/latest/api/agents/ppo.html
+agent:
+    class: "PPO"
+    rollouts: 16
+    learning_epochs: 8
+    mini_batches: 1
+    discount_factor: 0.99
+    lambda: 0.95
+    learning_rate: 3.e-4
+    learning_rate_scheduler: "KLAdaptiveLR"
+    learning_rate_scheduler_kwargs:
+        kl_threshold: 0.008
+    random_timesteps: 0 # random exploration steps
+    learning_starts: 0 # learning starts after this many steps
+    grad_norm_clip: 1.0
+    ratio_clip: 0.2
+    value_clip: 0.2
+    clip_predicted_values: True
+    entropy_loss_scale: 0.0
+    value_loss_scale: 2.0
+    kl_threshold: 0
+    rewards_shaper_scale: 1.0
+    time_limit_bootstrap: False
+    # logging and checkpoint
+    experiment:
+        directory: "runs"
+        experiment_name: ""
+        write_interval: 16
+        checkpoint_interval: 80
+        store_separately: False
+        wandb: False
+        wandb_kwargs: null
+
+# Sequential trainer
+# https://skrl.readthedocs.io/en/latest/api/trainers/sequential.html
+trainer:
+    class: "SequentialTrainer"
+    timesteps: 1600
+    environment_info: "log"