Model comparison

Marco-Nano-Instruct vs llama-3-youko-8b

These two publish different quantization sets; the table below has the exact sizes.

From the file· summed bytes, KV per layer

Side by side

Marco-Nano-Instructllama-3-youko-8b
Parameters8.0B8.0B
Architectureqwen3moellama
Layers2832
Native context32,7688,192
Mixture of expertsyes, 232 expertsno
Quantizations published362
Smallest quantization2.44 GiB5.34 GiB
Q4_K_M5.00 GiB
Licenceapache-2.0llama3

KV cache by context

the term that decides long-context viability
ContextMarco-Nano-Instructllama-3-youko-8bRatio
4,0960.44 GiB0.50 GiB1.14×
8,1920.88 GiB1.00 GiB1.14×
16,3841.75 GiB2.00 GiB1.14×
32,7683.50 GiB4.00 GiB1.14×
65,5367.00 GiB8.00 GiB1.14×
131,07214.00 GiB16.00 GiB1.14×