Quantization publisher
sabafallah
sabafallah publishes 9 quantizations across 3 models in our index, averaging 8.710 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
3
Quantizations
9
Models covered
3
Avg effective bpw
8.710
across their files
Same model, same quant label, different bytes
largest disagreements first
| Model | Quant | sabafallah | vs | Theirs | Difference |
|---|---|---|---|---|---|
| DeepSeek-OCR | Q8_0 | 2.91 GiB | NexaAI | 2.90 GiB | +0.3% |
| DeepSeek-OCR | BF16 | 5.47 GiB | NexaAI | 5.47 GiB | +0.1% |
| Unlimited-OCR | Q8_0 | 2.91 GiB | DevQuasar | 2.91 GiB | 0.0% |
| Unlimited-OCR | Q4_K_M | 1.82 GiB | DevQuasar | 1.82 GiB | 0.0% |
| Unlimited-OCR | Q8_0 | 2.91 GiB | sahilchachra | 2.91 GiB | -0.0% |
| Unlimited-OCR | Q4_K_M | 1.82 GiB | sahilchachra | 1.82 GiB | -0.0% |
| Unlimited-OCR | BF16 | 5.47 GiB | sahilchachra | 5.47 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 |
|---|---|---|
| Unlimited-OCR | 3 | 1.82 GiB |
| DeepSeek-OCR-2 | 3 | 1.82 GiB |
| DeepSeek-OCR | 3 | 1.82 GiB |