Quantization publisher

Naphula

Naphula publishes 32 quantizations across 4 models in our index, averaging 4.488 effective bits per weight. Their files differ in size from other publishers' builds of the same nominal quantization on 3 of the pairs we can compare — the same label does not mean the same file.

From the file· summed file bytes
Repositories
4
Quantizations
32
Models covered
4
Avg effective bpw
4.488
across their files

Same model, same quant label, different bytes

largest disagreements first
ModelQuantNaphulavsTheirsDifference
Goetia-26B-A4B-v1.3-Absolute-Heretic-ARAQ8_025.02 GiBmradermacher25.02 GiB-0.0%
Slimaki-Tavern-24B-v1.3Q8_023.33 GiBmradermacher23.33 GiB-0.0%
Slimaki-Tavern-24B-v1.3Q5_K_M15.61 GiBmradermacher15.61 GiB-0.0%

A quantization label describes a target, not a recipe. Publishers make different choices about which tensors to keep at higher precision, and some apply an importance matrix while others don't — so two files both honestly labelled the same thing can differ measurably in size and in quality.

Models they publish