I've been trying to implement petastorm to load parquet straight into tensorflow. However I do not understand what is going on... After reading parquet file like:
with make_batch_reader(dataset_url_or_urls=filepath) as reader:
dataset = make_petastorm_dataset(reader)
it returns tensorflow.python.data.ops.dataset_ops.DatasetV1Adapter then when I try to iterate over it I get following error
UnknownError: RuntimeError: Trying to read a sample after a reader created by make_reader/make_batch_reader has stopped. This may happen if the make_reader/make_batch_reader context manager has exited but you try to fetch a sample from it anyway Traceback (most recent call last):
How do I fix this and also how to get tf.data.dataset instead of DatasetV1Adapter which I believe is old verson of tf.data.dataset