Moves saving of best agent per gen to right after sort

train_GeneticAgent.py

@@ -46,6 +46,7 @@ def train():
             scores = np.array(list(tqdm(pool.imap_unordered(agent.get_fitness, args), desc='Population done', leave=False)))
 
         agent.weights = agent.weights[scores.argsort()[::-1]]
+        best_per_gen[gen] = agent.weights[0]
 
         all_scores[gen] = scores
         survivors = int(np.ceil(elite_pct * population_size))
@@ -55,7 +56,6 @@ def train():
             idx = np.random.randint(0, len(parents), size=2)
             parent1, parent2 = parents[idx, :]
 
-            best_per_gen[gen] = agent.weights[scores.argmax()]
             if np.random.random() < crossover_rate:
                 child1, child2 = agent.breed(parent1, parent2)