Quantization publisher

pmysl

pmysl publishes 11 quantizations across 1 models in our index, averaging 6.044 effective bits per weight. Their files differ in size from other publishers' builds of the same nominal quantization on 7 of the pairs we can compare — the same label does not mean the same file.

From the file· summed file bytes
Repositories
1
Quantizations
11
Models covered
1
Avg effective bpw
6.044
across their files

Same model, same quant label, different bytes

largest disagreements first
ModelQuantpmyslvsTheirsDifference
c4ai-command-r-plusQ3_K_M47.48 GiBdranger00347.48 GiB-0.0%
c4ai-command-r-plusQ5_K_M68.57 GiBdranger00368.57 GiB-0.0%
c4ai-command-r-plusQ5_K_S66.87 GiBdranger00366.87 GiB-0.0%
c4ai-command-r-plusQ3_K_L51.60 GiBdranger00351.60 GiB-0.0%
c4ai-command-r-plusQ4_K_M58.44 GiBdranger00358.44 GiB-0.0%
c4ai-command-r-plusQ2_K36.78 GiBdranger00336.78 GiB-0.0%
c4ai-command-r-plusQ4_K_S55.55 GiBdranger00355.55 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
c4ai-command-r-plus1136.78 GiB