Quantization publisher

katafiek

katafiek publishes 4 quantizations across 1 models in our index, averaging 6.474 effective bits per weight. Their files differ in size from other publishers' builds of the same nominal quantization on 4 of the pairs we can compare — the same label does not mean the same file.

From the file· summed file bytes
Repositories
1
Quantizations
4
Models covered
1
Avg effective bpw
6.474
across their files

Same model, same quant label, different bytes

largest disagreements first
ModelQuantkatafiekvsTheirsDifference
Katarau-9B-ru-RP-nsfwQ4_K_M5.24 GiBmradermacher5.24 GiB-0.0%
Katarau-9B-ru-RP-nsfwQ5_K_M6.02 GiBmradermacher6.02 GiB-0.0%
Katarau-9B-ru-RP-nsfwQ6_K6.85 GiBmradermacher6.85 GiB-0.0%
Katarau-9B-ru-RP-nsfwQ8_08.87 GiBmradermacher8.87 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

ModelQuantizationsSmallest
Katarau-9B-ru-RP-nsfw45.24 GiB