Donghyun Kang issues

Repositories
Issues
Comments

Results 1 issues of


                                            Donghyun Kang

caffe2 conv - cudnnAddTensor for bias generates wrong output! (HALF, CUDNN6.0, in special cases)

cudnnAddTensor generates an error on output blob :( specification: in my case, **float16** input tensor (NCHW) = (8, 256, 28, 28) output tensor (NCHW) = (8, 512, 28, 28) **1x1...

Downstream problem