在使用十折交叉验证时，通常可以计算每一折的性能指标（如RMSE、MAPE）并输出。这涉及到在每一折结束后对模型进行评估并记录性能指标的过程。在神经网络中，这通常通过在每一折中保存模型的预测结果，并在整个交叉验证过程结束后计算性能指标来实现。

以下是一个简单的伪代码示例，说明如何在十折交叉验证中计算每一折的性能指标：

from sklearn.model_selection import KFold
from sklearn.metrics import mean_squared_error, mean_absolute_error
import numpy as np

# Assume X_train, y_train are your training data and labels
# Assume model is your neural network model

kf = KFold(n_splits=10, shuffle=True, random_state=42)

rmse_scores = []
mape_scores = []

for train_index, val_index in kf.split(X_train):
    X_train_fold, X_val_fold = X_train[train_index], X_train[val_index]
    y_train_fold, y_val_fold = y_train[train_index], y_train[val_index]

    # Train your neural network model on X_train_fold and y_train_fold
    # Make predictions on X_val_fold
    y_pred_fold = model.predict(X_val_fold)

    # Calculate RMSE and MAPE for this fold
    rmse_fold = np.sqrt(mean_squared_error(y_val_fold, y_pred_fold))
    mape_fold = np.mean(np.abs((y_val_fold - y_pred_fold) / y_val_fold)) * 100

    # Append scores to the lists
    rmse_scores.append(rmse_fold)
    mape_scores.append(mape_fold)

# Calculate average scores across all folds
average_rmse = np.mean(rmse_scores)
average_mape = np.mean(mape_scores)

print("RMSE scores for each fold:", rmse_scores)
print("MAPE scores for each fold:", mape_scores)
print("Average RMSE:", average_rmse)
print("Average MAPE:", average_mape)

这段代码使用了scikit-learn库中的KFold类来进行十折交叉验证，然后在每一折中训练神经网络模型，计算预测结果的RMSE和MAPE，并将这些指标存储在相应的列表中。最后，计算平均的RMSE和MAPE，并输出每一折的性能指标。确保在实际代码中适配你的数据和神经网络模型。