Quantization format
Q2_0
Q2_0 has a nominal rate of — bits per weight. We have no measured files carrying this label yet.
From the file· 7 files measured
Nominal bpw
—
from the block layout
Measured average
8.699
7 files
Range
2.132–18.098
varies by architecture
File sizes
0.43 GiB+
up to 7.68 GiB
Real files
one per model, most downloaded first
| Model | Size● | Effective bpw● | vs nominal● |
|---|---|---|---|
| Ternary-Bonsai-8B-unpacked | 2.03 GiB | 2.132 | — |
| Ternary-Bonsai-27B-gguf | 6.67 GiB | 15.722 | — |
| Ternary-Bonsai-4B-unpacked | 1.00 GiB | 2.138 | — |
| Ternary-Bonsai-1.7B-unpacked | 0.43 GiB | 2.155 | — |
| Nemotron-3-Embed-8B-BF16 | 2.36 GiB | 2.550 | — |
●From the filewhat these mean