perf(quantize): AVX2 SIMD Q4_0/Q8_0 dequant — ~8-9× speedup (closes #386)#1691

noahgift merged 20 commits into

mainpaiml/aprender:mainfrom

perf/386-q4-q8-avx2-dequantpaiml/aprender:perf/386-q4-q8-avx2-dequantCopy head branch name to clipboard