Model comparison

Tini-Cybersec-8B-A1B vs llama-3-youko-8b

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: Tini-Cybersec-8B-A1B's KV cache at 32K is 10.7× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

Tini-Cybersec-8B-A1Bllama-3-youko-8b
Parameters8.5B8.0B
Architecturelfm2moellama
Layers2432
Native context128,0008,192
Mixture of expertsyes, 32 expertsno
Quantizations published352
Smallest quantization2.71 GiB5.34 GiB
Q4_K_M4.80 GiB
Licenceapache-2.0llama3

KV cache by context

the term that decides long-context viability
ContextTini-Cybersec-8B-A1Bllama-3-youko-8bRatio
4,0960.05 GiB0.50 GiB10.66×
8,1920.09 GiB1.00 GiB10.66×
16,3840.19 GiB2.00 GiB10.66×
32,7680.38 GiB4.00 GiB10.67×
65,5360.75 GiB8.00 GiB10.67×
131,0721.50 GiB16.00 GiB10.67×