Model comparison
SambaLingo-Japanese-Chat vs Qwen3-VL-8B-Instruct-abliterated-v1
These two publish different quantization sets; the table below has the exact sizes.
From the file· summed bytes, KV per layer
Side by side
| SambaLingo-Japanese-Chat | Qwen3-VL-8B-Instruct-abliterated-v1 | |
|---|---|---|
| Parameters | 6.9B | 8.8B |
| Architecture | llama | qwen3vl |
| Layers | 32 | — |
| Native context | 4,096 | — |
| Mixture of experts | no | no |
| Quantizations published | 22 | 48 |
| Smallest quantization | 1.52 GiB | 1.97 GiB |
| Q4_K_M | — | 4.68 GiB |
| Licence | llama2 | apache-2.0 |
KV cache by context
the term that decides long-context viability
| Context | SambaLingo-Japanese-Chat | Qwen3-VL-8B-Instruct-abliterated-v1 | Ratio |
|---|---|---|---|
| 4,096 | 2.00 GiB | — | — |
| 8,192 | 4.00 GiB | — | — |
| 16,384 | 8.00 GiB | — | — |
| 32,768 | 16.00 GiB | — | — |
| 65,536 | 32.00 GiB | — | — |
| 131,072 | 64.00 GiB | — | — |