Oh this is nice! Is this the same quantization method as GGML? It's just described as "Efficient custom 8 bit and 4 bit quantization" And thanks for the https:

Oh this is nice! Is this the same quantization method as GGML? It's just described as "Efficient custom 8 bit and 4 bit quantization" And thanks for the https://pypi.org/project/lm-eval/ https://pypi.org/project/lm-eval/ pointer, have to get this in tinygrad.