Error "TypeError: object of type 'float' has no len()" when adding parameter "weights" to sum_model in catboost classifier

Question

I am building a CatBoost model to simulate the federated learning system at first I am training the local models on the same Catboost model using a for loop code:

for j in range(len(Train_Datasets)): 
    
    df_train= Train_Datasets[j]
    df_test= Test_Datasets[j]
    x_train = df_train.drop(['index', 'binary_attack'], axis=1)
    x_test = df_test.drop(['index', 'binary_attack'], axis=1)
    y_train = df_train['binary_attack']
    y_test = df_test["binary_attack"] 
    
    local_model = CatBoostClassifier(**params)
    local_model.fit(x_train, y_train)

Later I am appending all the local models and sum them using Sum_model function from the Catboost package:

models.append(local_model)
global_model = sum_models(models, weights=0.7)
global_model= to_classifier(global_model)

I am getting this error whenever I add the weights=0.7 parameter:

TypeError: object of type 'float' has no len()

However, I tried once the same code with adding the weights parameter, and it worked. Do you have any suggestions on what might be the problem?