The current version of the HANS dataset is missing the additional information provided for each example, including the sentence parses, heuristic and subcase.

How to reproduce the behaviour

1. Let's we have 0 and 1 labels' shortkeys.
2. Set the second shortkey to 0.
3. You get 400 error. It's ok.
4. But shortkey at the popup is 0.
5. Click outside of popup.
6. Shortkey at the table is 1. It's ok.
7. Click on shortkey.
8. But shortkey at the popup is 0.

Your Environment

Colour could use a definition like that to interact easily with OpenColorIO:

def ocio_colourspace_process(a, config, source, target, use_cpu=True):
    a = colour.utilities.as_float_array(a, dtype=np.float32)
    height, width, channels = a.shape

    processor = config.getProcessor(source, target)
    processor = (
        processor.getDefaultCPUProcessor()
        if use_cpu

Issue to track tutorial requests:

• Deep Learning with PyTorch: A 60 Minute Blitz - #69

What?

Currently, API manually throws its own messages and errors. We should move them to werkzeug exceptions.

Datasets can be listed with lots of filters, but strangely not by uploader user id.