Model comparison
Gemma-SEA-LION-v3-9B-IT vs HyperCLOVAX-SEED-Text-Instruct-1.5B
At Q4_K_M, HyperCLOVAX-SEED-Text-Instruct-1.5B is the smaller download — 1,133,974,368 bytes against 5,761,057,920.
From the file· summed bytes, KV per layer
Side by side
| Gemma-SEA-LION-v3-9B-IT | HyperCLOVAX-SEED-Text-Instruct-1.5B | |
|---|---|---|
| Parameters | 9.2B | 1.6B |
| Architecture | gemma2 | llama |
| Layers | 42 | — |
| Native context | 8,192 | — |
| Mixture of experts | no | no |
| Quantizations published | 33 | 1 |
| Smallest quantization | 2.22 GiB | 1.06 GiB |
| Q4_K_M | 5.37 GiB | 1.06 GiB |
| Licence | gemma | other |
KV cache by context
the term that decides long-context viability
| Context | Gemma-SEA-LION-v3-9B-IT | HyperCLOVAX-SEED-Text-Instruct-1.5B | Ratio |
|---|---|---|---|
| 4,096 | 1.31 GiB | — | — |
| 8,192 | 2.05 GiB | — | — |
| 16,384 | 3.36 GiB | — | — |
| 32,768 | 5.99 GiB | — | — |
| 65,536 | 11.24 GiB | — | — |
| 131,072 | 21.74 GiB | — | — |