Conchylicultor

Yes, however, the `ragged_tensor.numpy()` remove information on the existing ragged tensor. Depending on your use-case might, it cam make it more complicated to use. Because you would now have to...

Also note you can also use `ds.as_numpy_iterator()` which should directly returns `tf.RaggedTensor` as numpy. `as_numpy_iterator` has some difference with `tfds.as_numpy` though. Like not supporting `None`, `len(ds)`,... but it's been a...

Thanks for reporting, the issue is that we forgot to upload the file here: https://github.com/tensorflow/datasets/blob/e5ca78ccef0c7715600e014934198afc30988a95/setup.py#L103

Thank you for the suggestion. I think it would be best to add directly those features inside the `DownloadManager.download` method. It would ensure downloads are correctly tracked by the rest...

You can use TFDS pathlib-like API which works with GCS paths: ```python path = tfds.core.as_path('gs://reu/data/corpus.0.tfrecord') with path.open('rb') as f: pass content = path.read_bytes() assert path.exists() assert path.name == 'corpus.0.tfrecord' ```...

It is possible that the code to auto-detect corrupted images do not works on windows: https://github.com/tensorflow/datasets/blob/921c0f86b8eeba863ce0af6523f34ac75d3d7529/tensorflow_datasets/image_classification/cats_vs_dogs.py#L94 Or maybe there are additional corrupted images on windows that works on linux ?...

@tomergt45 Thanks for looking into this. The generation code is only executed once the first time the dataset is generated, afterward, the generated files are reused. To force executing the...

What is the full stacktrace ?

> there is no stacktrace. Does that mean the dataset is correctly generated and loaded ? TFDS tries to lookup on GCS to see if the dataset can be directly...

To generate dataset with beam, you should try to follow our instructions: https://www.tensorflow.org/datasets/beam_datasets#on_google_cloud_dataflow Likely you'll have to use Dataflow or similar.