Model comparison
LFM2-2.6B-Exp 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 1,563,669,120.
From the file· summed bytes, KV per layer
Side by side
| LFM2-2.6B-Exp | HyperCLOVAX-SEED-Text-Instruct-1.5B | |
|---|---|---|
| Parameters | 2.6B | 1.6B |
| Architecture | lfm2 | llama |
| Layers | 30 | — |
| Native context | 128,000 | — |
| Mixture of experts | no | no |
| Quantizations published | 42 | 1 |
| Smallest quantization | 0.88 GiB | 1.06 GiB |
| Q4_K_M | 1.46 GiB | 1.06 GiB |
| Licence | other | other |
KV cache by context
the term that decides long-context viability
| Context | LFM2-2.6B-Exp | HyperCLOVAX-SEED-Text-Instruct-1.5B | Ratio |
|---|---|---|---|
| 4,096 | 0.06 GiB | — | — |
| 8,192 | 0.13 GiB | — | — |
| 16,384 | 0.25 GiB | — | — |
| 32,768 | 0.50 GiB | — | — |
| 65,536 | 1.00 GiB | — | — |
| 131,072 | 2.00 GiB | — | — |