otherwhere1/Goetia-26B-A4B-v1.3-Absolute-Heretic-ARA-mlx-8Bit

Quantization vs Naphula/Goetia-26B-A4B-v1.3-Absolute-Heretic-ARA. otherwhere1/Goetia-26B-A4B-v1.3-Absolute-Heretic-ARA-mlx-8Bit is a quantized version of the other (U32/BF16).

1 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse