para_config.ini

[simulation]
total_episodes = 100
max_step =3600
green_duration = 5
yellow_duration = 2
sim_seed = 42
num_states_obs = 4
num_states_phase = 4
num_states_lanes = 12
num_actions = 4
num_neighbors = 5
coef_reward = 0.1

[model]
agent_type = IQL
seq_len = 4
hidden_dim = 64
n_heads = 8

[memory]
buffer_size = 1000
batch_size = 64

[agent]
update_step = 40
gamma = 0.9
epsilon_final = 0.1
epsilon_init = 1
tau = 0.01
lr = 0.001