omjones92 comments

Repositories
Issues
Comments

Results 2 comments of


                                            omjones92

Depth Anything model - runs perfectly on CPU but on GPU Vulkan it only outputs one image the rest are scrambled

I modified the attention module to get pnnx to detect it and convert it. Check the google drive link above I have uploaded the modified version as a zip. The...

`convert_float_to_float16` results in `failed in shape inference <class 'Exception'>`

I'm having the same issue with a 4x span model in c++, fp32 version runs fine but when converted to fp16 with keep_io_types it gives : Non-zero status code returned...