当我有多个TFRecords时，如何使用slim.dataset_data_provider？

import tensorflow as tf
import tensorflow.contrib.slim as slim

# Step 1: Create a TFRecords file list
tfrecords_files = tf.train.match_filenames_once("path/to/tfrecords/*.tfrecords")

# Step 2: Create an input queue
input_queue = tf.train.string_input_producer(tfrecords_files)

# Step 3: Read TFRecords files
reader = tf.TFRecordReader()
_, serialized_example = reader.read(input_queue)

# Step 4: Provide data using slim.dataset_data_provider
def parse_fn(serialized_example):
    # Parse features from serialized example
    features = tf.parse_single_example(serialized_example, features={
        'image': tf.FixedLenFeature([], tf.string),
        'label': tf.FixedLenFeature([], tf.int64),
    })
    # Decode image and preprocess
    image = tf.image.decode_jpeg(features['image'], channels=3)
    image = tf.image.resize_images(image, [224, 224])
    image = tf.image.per_image_standardization(image)
    # Convert label to one-hot encoding
    label = tf.one_hot(features['label'], depth=10)
    return image, label

data_provider = slim.dataset_data_provider.DatasetDataProvider(
    dataset=serialized_example,
    num_readers=4,
    common_queue_capacity=32,
    common_queue_min=8,
    shuffle=True,
    num_epochs=100,
    parser_fn=parse_fn
)

# Get data from data provider
image, label = data_provider.get(['image', 'label'])

# Use the data in your model
# ...

# Start the input queue threads
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)

# Run your TensorFlow session
with tf.Session() as sess:
    # Initialize variables and TFRecords file list
    sess.run([tf.global_variables_initializer(), tf.local_variables_initializer(), tfrecords_files.initializer])
    
    # Train or evaluate your model
    # ...
    
    # Stop the input queue threads
    coord.request_stop()
    coord.join(threads)