3 ms·

Halving the precision of the weights is not a free lunch...

• Catloafdev 3mo ago
  Q8 is virtually lossless. The quantization is much more noticeable around Q4 and below. FP16->Q8 on consumer hardware is 2x the speed at ~99.99% the quality.
  • rvba 3mo ago
    Any source that confirms the 99.99% quality?
    • Catloafdev 3mo ago
      I don't have a 'source' off-hand but I recommend reading up on it if you want to learn more. A lot of models on HF show a card demonstrating the different quality trade-offs between quants.