Quantization publisher
INSAIT-Institute
INSAIT-Institute publishes 4 quantizations across 1 models in our index, averaging 5.962 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
5.962
across their files
Same model, same quant label, different bytes
largest disagreements first
| Model | Quant | INSAIT-Institute | vs | Theirs | Difference |
|---|---|---|---|---|---|
| gemma-2-9b | Q4_K_M | 5.37 GiB | QuantFactory | 5.37 GiB | 0.0% |
| gemma-2-9b | Q4_K_S | 5.10 GiB | QuantFactory | 5.10 GiB | 0.0% |
| gemma-2-9b | Q5_K_S | 6.04 GiB | QuantFactory | 6.04 GiB | 0.0% |
| gemma-2-9b | Q8_0 | 9.15 GiB | QuantFactory | 9.15 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
| Model | Quantizations | Smallest |
|---|---|---|
| gemma-2-9b | 4 | 5.10 GiB |