Model comparison
CHEETAH-350M-Merged-FP16 vs LFM2.5-230M
At Q4_K_M, LFM2.5-230M is the smaller download — 153,406,048 bytes against 231,031,616.
From the file· summed bytes, KV per layer
Side by side
| CHEETAH-350M-Merged-FP16 | LFM2.5-230M | |
|---|---|---|
| Parameters | 354M | 230M |
| Architecture | lfm2 | lfm2 |
| Layers | 16 | 14 |
| Native context | 128,000 | 128,000 |
| Mixture of experts | no | no |
| Quantizations published | 47 | 25 |
| Smallest quantization | 0.10 GiB | 0.10 GiB |
| Q4_K_M | 0.22 GiB | 0.14 GiB |
| Licence | other | other |
KV cache by context
the term that decides long-context viability
| Context | CHEETAH-350M-Merged-FP16 | LFM2.5-230M | Ratio |
|---|---|---|---|
| 4,096 | 0.05 GiB | 0.05 GiB | 1.00× |
| 8,192 | 0.09 GiB | 0.09 GiB | 1.00× |
| 16,384 | 0.19 GiB | 0.19 GiB | 1.00× |
| 32,768 | 0.38 GiB | 0.38 GiB | 1.00× |
| 65,536 | 0.75 GiB | 0.75 GiB | 1.00× |
| 131,072 | 1.50 GiB | 1.50 GiB | 1.00× |