Numpy arrays are stored as text arrays by default. This is readable but takes a lot of space.
When the file is compressed, it perhaps makes more sense to use the new compact format introduced in #9.
Since this is a backwards-incompatible change, it will have to wait until the next major version (4.0.0).
For now the compressed format can be used like this (compression is optional):
json = dumps(data, compression=True, properties={'ndarray_compact': True})
There's also a warning announcing this change, which you can silence by explicitly disabling the new format:
json = dumps(data, properties={'ndarray_compact': False})
Numpy arrays are stored as text arrays by default. This is readable but takes a lot of space.
When the file is compressed, it perhaps makes more sense to use the new compact format introduced in #9.
Since this is a backwards-incompatible change, it will have to wait until the next major version (4.0.0).
For now the compressed format can be used like this (compression is optional):
There's also a warning announcing this change, which you can silence by explicitly disabling the new format: