[GitHub] [incubator-mxnet] yzhliu commented on issue #18060: [Numpy] Dense does not support float32 + float64.

yzhliu commented on issue #18060: URL: https://github.com/apache/incubator-mxnet/issues/18060#issuecomment-621537539 We'll have an global default dtype switch, and it will be set to float32 if not changed by user. This is

[GitHub] [incubator-mxnet] yzhliu commented on issue #18060: [Numpy] Dense does not support float32 + float64.

yzhliu commented on issue #18060: URL: https://github.com/apache/incubator-mxnet/issues/18060#issuecomment-618102878 As discussed with @sxjscience before, one solution is to always convert to fp32 when reading from numpy array. what do you think? @reminisce @haojin2 @szha