Thank you for supporting on aten::softplus.out, error on that operator has gone, but another show up!
aten::upsample_bicubic2d.out.
I'm still trying to use GPU Radeon 6800 on ZoeDepth model, use DirectML, until it works.

Full error message:

C:\Users[USER]\anaconda3\envs\zoe\lib\site-packages\torch\nn\functional.py:4073: UserWarning: The operator 'aten::upsample_bicubic2d.out' is not currently supported on the DML backend and will fall back to run on the CPU. This may have performance implications. (Triggered internally at C:__w\1\s\pytorch-directml-plugin\torch_directml\csrc\dml\dml_cpu_fallback.cpp:17.)
return torch._C._nn.upsample_bicubic2d(input, output_size, align_corners, scale_factors)
Traceback (most recent call last):

Line 187, in
compiled_tiles[x:x+M, y:y+N] += selected_filter * (np.mean(low_res_scaled_depth[x:x+M, y:y+N]) + np.std(low_res_scaled_depth[x:x+M, y:y+N]) * ((scaled_depth - np.mean(scaled_depth)) / np.std(scaled_depth)))
ValueError: operands could not be broadcast together with shapes (457,365) (447,375)