feat(mathy_alpha_sm): updated multi-task model

Author: justindujardin

libraries/mathy_alpha_sm/mathy_alpha_sm/model.config.json

@@ -6,47 +6,47 @@
   "author": "Justin DuJardin",
   "email": "justin@dujardinconsulting.com",
   "url": "https://mathy.ai",
-  "mathy_version": ">=0.0.0,<1.0.0",
+  "mathy_version": "0.3.5",
   "units": 512,
   "embedding_units": 512,
   "lstm_units": 128,
   "topics": [
-    "poly-combine"
+    "poly"
   ],
-  "difficulty": null,
-  "model_dir": "/var/folders/q3/j42n3r692gb2yl9cp1ncs3q00000gn/T/tmpa8tndnqz",
+  "difficulty": "easy",
+  "model_dir": "./training/alpha_piv_new",
   "model_name": "model",
   "init_model_from": null,
   "train": false,
   "verbose": false,
   "lr": 0.0003,
-  "max_eps": 1,
+  "max_eps": 15000,
   "summary_interval": 100,
   "gamma": 0.99,
-  "num_workers": 1,
+  "num_workers": 4,
   "td_lambda": 0.2,
   "print_training": false,
   "print_mode": "terminal",
   "use_term_order": false,
   "use_tree_complexity": false,
   "use_grouping_control": true,
-  "clip_grouping_control": false,
+  "clip_grouping_control": true,
   "use_env_features": false,
   "use_node_values": true,
   "update_gradients_every": 64,
   "num_thinking_steps_begin": 3,
-  "mcts_sims": 200,
-  "worker_wait": 0.01,
+  "mcts_sims": 10,
+  "worker_wait": 0.1,
   "profile": false,
   "action_strategy": "a3c",
   "main_worker_use_epsilon": false,
   "e_greedy_min": 0.01,
   "e_greedy_max": 0.1,
-  "entropy_loss_scaling": 1,
+  "entropy_loss_scaling": 0.05,
   "normalize_entropy_loss": true,
-  "aux_tasks_weight_scale": 1,
+  "aux_tasks_weight_scale": 0.1,
   "teacher_start_evaluations_at_episode": 50,
   "teacher_evaluation_steps": 20,
   "teacher_promote_wins": 0.95,
   "teacher_demote_wins": 0.6
-}
+}