How to write scale and zero_point to fp32 tensor without doing quantization?

[alanzhai219 (Alan Zhai)]

Thanks for your reply. It is because our operation quantization compiler has too much rules.