Added TODO for run_ml_sim. Updated valid_config to match modern configuration files

Author: bozo-bud

run_ml_sim.py

@@ -11,7 +11,8 @@
 from config_scripts.setup_config import read_config
 from helper_scripts.ml_helpers import process_data, plot_confusion
 from helper_scripts.ml_helpers import save_model
-
+#TODO: Raise Not Implemented Error
+#TODO: This node is still in active development
 
 def _train_test_knn(df_processed: pd.DataFrame, sim_dict: dict, erlang: str):
     predictor_df = df_processed['num_segments']

tests/fixtures/valid_config.ini

@@ -48,31 +48,39 @@ xt_noise = False
 requested_xt = {"QPSK": -26.19, "16-QAM": -36.69, "64-QAM": -41.69}
 
 [rl_settings]
+n_trials = 2
 device = cpu
 optimize = False
 is_training = True
-path_algorithm = ucb_bandit
-path_model = greedy_bandit/NSFNet/0617/16_47_22_694727/state_vals_e750.0_routes_c4.json
+
+path_algorithm = ppo
+path_model = None
 core_algorithm = first_fit
-core_model = greedy_bandit/NSFNet/0617/16_57_13_315030/state_vals_e750.0_cores_c4.json
+core_model = None
 spectrum_algorithm = first_fit
-spectrum_model = ppo/NSFNet/0512/12_57_55_484293
-# Only for DRL
+spectrum_model = None
+
 render_mode = None
 super_channel_space = 3
-# Only for q-learning
-learn_rate = 0.01
-discount_factor = 0.95
-epsilon_start = 0.2
-epsilon_end = 0.05
-reward = 1
-penalty = -100
-dynamic_reward = False
-# TODO: Sim helpers has not been updated for this! (Only support for 2)
+
+alpha_start = 0.1
+alpha_end = 0.01
+alpha_update = linear_decay
+
+gamma = 0.1
+
+epsilon_start = 0.0
+epsilon_end = 0.0
+epsilon_update = linear_decay
+
 path_levels = 2
 decay_factor = 0.01
+
+reward = 1
+penalty = -10
+dynamic_reward = False
 core_beta = 0.1
-gamma = 0.1
+
 
 [ml_settings]
 deploy_model = False