OutOfRangeError when creating batch from tfrecord file - tensorflow

Question

I'm writing a script that saves certain features of my data to tfrecord. The features are numpy arrays (float32). When I read the tfrecord file I get the following error:
OutOfRangeError (see above for traceback): RandomShuffleQueue '_1_shuffle_batch/random_shuffle_queue' is closed and has insufficient elements (requested 20, current size 0)
[[Node: shuffle_batch = QueueDequeueManyV2[component_types=[DT_UINT8, DT_UINT8], timeout_ms=-1, _device="/job:localhost/replica:0/task:0/device:CPU:0"](shuffle_batch/random_shuffle_queue, shuffle_batch/n)]]
I searched a lot, and apparently this error can be caused by different things. So far, I was not able to fix it. I recreated the problem with the following minimal code:
saving the toy data:
def _bytes_feature(value):
return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))
writer = tf.python_io.TFRecordWriter('stuff.tfrecords')
for i in range(100):
seq = np.random.uniform(size=(500,300)).astype(np.float32)
lbl = np.random.uniform(size=(90,1)).astype(np.float32)
feature = {'train/lbl': _bytes_feature(tf.compat.as_bytes(lbl.tostring())),
'train/seq': _bytes_feature(tf.compat.as_bytes(seq.tostring()))}
example = tf.train.Example(features=tf.train.Features(feature=feature))
writer.write(example.SerializeToString())
writer.close()
sys.stdout.flush()
Reading the data:
def read_and_decode_single_example(filename):
filename_queue = tf.train.string_input_producer([filename], num_epochs=1)
reader = tf.TFRecordReader()
_, serialized_example = reader.read(filename_queue)
f = {'train/lbl': tf.FixedLenFeature([], tf.string),
'train/seq': tf.FixedLenFeature([], tf.string)}
features = tf.parse_single_example(serialized_example, features=f)
seq = tf.decode_raw(features['train/seq'], tf.float32)
lbl = tf.decode_raw(features['train/lbl'], tf.float32)
seq = tf.reshape(seq, [ 500,300 ])
lbl = tf.reshape(lbl, [ 90,1 ])
sbatch, lbatch = tf.train.shuffle_batch([seq, lbl],
batch_size= batch_size,
capacity=3*batch_size,
min_after_dequeue=batch_size)
return sbatch, lbatch
sbatch, lbatch = read_and_decode_single_example("stuff.tfrecords" )
with tf.Session() as sess:
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(sess=sess, coord=coord)
sess.run(tf.global_variables_initializer())
sess.run(tf.local_variables_initializer())
s,l = sess.run([sbatch, lbatch])
coord.request_stop()
coord.join(threads)
I'm using Tensorflow-GPU v. 1.4.0.
Here is some error code, that may be informative:
Caused by op 'shuffle_batch', defined at:
File "teststuff.py", line 59, in <module>
sbatch, lbatch = read_and_decode_single_example("stuff.tfrecords" )
File "teststuff.py", line 54, in read_and_decode_single_example
min_after_dequeue=batch_size)
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/training/input.py", line 1225, in shuffle_batch
name=name)
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/training/input.py", line 796, in _shuffle_batch
dequeued = queue.dequeue_many(batch_size, name=name)
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/ops/data_flow_ops.py", line 464, in dequeue_many
self._queue_ref, n=n, component_types=self._dtypes, name=name)
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/ops/gen_data_flow_ops.py", line 2418, in _queue_dequeue_many_v2
component_types=component_types, timeout_ms=timeout_ms, name=name)
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/framework/op_def_library.py", line 787, in _apply_op_helper
op_def=op_def)
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/framework/ops.py", line 2956, in create_op
op_def=op_def)
File "/usr/local/lib/python3.5/dist-packages/tensorflow/python/framework/ops.py", line 1470, in __init__
self._traceback = self._graph._extract_stack() # pylint: disable=protected-access