Model comparison
csm-1b vs s2-pro
These two publish different quantization sets; the table below has the exact sizes.
From the file· summed bytes, KV per layer
Side by side
KV cache by context
the term that decides long-context viability
| Context | csm-1b | s2-pro | Ratio |
|---|---|---|---|
| 4,096 | 0.13 GiB | — | — |
| 8,192 | 0.25 GiB | — | — |
| 16,384 | 0.50 GiB | — | — |
| 32,768 | 1.00 GiB | — | — |
| 65,536 | 2.00 GiB | — | — |
| 131,072 | 4.00 GiB | — | — |