Can Apple M1 support quantization？
you could run the quantization APIs but the actual quantized model you get at the end doesn’t seem like it could run since none of the backends seem to work with M1, fbgemm needs x86 with AVX, and qnnpack needs ARM.
Do you know of any backends, except fbgemm and qnnpack, supporting quantization on M1/M2 chips?
I regret that I don’t know about it.