Python 类型错误：'<'；在'；dict'；和'；int'；使用gridsearch保存参数时

我正在使用gridsearch查找最佳参数，我希望保存这些参数并将其用于测试。保存best^参数并将其提供给algo时，我遇到了一个错误：

BEST PARAMETERS:
 {'metric': 'euclidean', 'n_neighbors': 5, 'weights': 'uniform'}
BEST SCORE:
 0.631578947368421
---Filename in processed corpus_ix_test_FMC_
small
Traceback (most recent call last):
  File "training_and_cross_validate_tuning.py", line 161, in <module>
    testing(testdir, f, root, dict_sm)
  File "training_and_cross_validate_tuning.py", line 138, in testing
    model_final.fit(Xtrain, Ytrain)
  File "/home/g/kee/anaconda3/lib/python3.7/site-packages/sklearn/neighbors/base.py", line 917, in fit
    return self._fit(X)
  File "/home/g/ke/anaconda3/lib/python3.7/site-packages/sklearn/neighbors/base.py", line 238, in _fit
    self.n_neighbors < self._fit_X.shape[0] // 2) and
TypeError: '<' not supported between instances of 'dict' and 'int'

    Xtrain_emb, mdlname = get_flaubert_layer(data)
    
    for mdl in dict_model:
        print('Training model: ', mdl)
        clf = GridSearchCV(estimator=dict_model[mdl], param_grid=parameters_dict[mdl], verbose = 1, n_jobs = -1, return_train_score=True)
        best_model = clf.fit(Xtrain_emb, label)
        print("Best parameters set found on development set:")  
        best_model_param = best_model.best_params_
        print(type(best_model_param))
        print('BEST PARAMETERS:\n', best_model_param)
        best_model_score = best_model.best_score_
        print('BEST SCORE:\n',best_model_score)
                
        # create, train and test
    return best_model_param, Xtrain_emb, label
        
def testing(path2, file, path, dict_model):

    param, Xtrain, Ytrain = training_data(file, path, dict_model) 
    # test
    filename_t = path2[21:-5]   # corpus_or_AB_FMC
    print("---Filename in processed", filename_t)
    f = pd.read_excel(os.path.join(path2), sheet_name= 0)
    data_idd = f.identifiant
    Xtest = f.verbatim
    Ytest = f.etiquette
    Xtest_emb, mdlname = get_flaubert_layer(Xtest)
    model_final = KNeighborsClassifier(param)
    model_final.fit(Xtrain, Ytrain)
    preds = model_final.predict(Xtest_emb)
    print('The best model from grid search scores {:.5f} ROC AUC on the test set.'.format(roc_auc_score(Ytest, preds)))