Run 1-D Conv using tensorflow

Question

I'm a beginner in TensorFlow. I want to train a 1-D conv model. I have one-rowed csv files for each row of my original data.

csv files look like this

csv_file1: 1.1, 1.3, 1.5, 1.5, 1
csv_file2: 2.1, 2.3, 2.7, 2.9, 0

The last column(containing 1 & 0) are the labels for the single-row csv files

Following the link I wrote the following pieces of code.

I converted the csv files to TFRecord using the following code

    with tf.python_io.TFRecordWriter(filename) as writer:
        features, label = df_values[:, 1:-1], df_values[:, -1:]
        example = tf.train.Example()
        example.features.feature["features"].float_list.value.extend(features[0])
        example.features.feature["label"].int64_list.value.append(label[0])
        writer.write(example.SerializeToString())

I want to now read the files and this is the code I'm using.

def _parse_function(data_record):
    features = {
        'label': tf.FixedLenSequenceFeature([], tf.int64, allow_missing = True),
        'features': tf.FixedLenSequenceFeature([], tf.float32, allow_missing = True),
    }
    sample = tf.parse_single_example(data_record, features)
    return sample['features'], sample['label']

filenames = glob.glob("*.tfrecords")
dataset = tf.data.TFRecordDataset(filenames)
dataset = dataset.map(_parse_function)  
dataset = dataset.shuffle(buffer_size=10000)
dataset = dataset.batch(batch_len)
# Create a one-shot iterator
iterator = dataset.make_one_shot_iterator()

X,y = iterator.get_next()

From here the problem starts, From the documentation I understand what session does but failing to put it into code. Assuming I'll later figure out how to use

tf.seesion.run()

I wrote the below code but don't know how to actually include it into my main script and further use it to train my model.

x_train_batch, y_train_batch = tf.train.shuffle_batch(
tensors=[X_train, y_train],
batch_size=batch_size,
capacity=capacity,
min_after_dequeue=min_after_dequeue,
enqueue_many=True,
num_threads=8)

x_train_batch = tf.cast(x_train_batch, tf.float32)
x_train_batch = tf.reshape(x_train_batch, shape=(batch_size, 1,65281))

y_train_batch = tf.cast(y_train_batch, tf.int64)
y_train_batch = tf.one_hot(y_train_batch, num_classes)

Any help regarding how to proceed further will help.

PS: Assuming my data was loaded into an np.array the dimension would be, (6571, 65281). Since it's astronomical data, each star has 65781 points.