Model comparison

Llama-3.1-8B-Instruct-heretic 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

Llama-3.1-8B-Instruct-hereticllama-3-youko-8b
Parameters8.0B8.0B
Architecturellamallama
Layers32
Native context8,192
Mixture of expertsnono
Quantizations published592
Smallest quantization2.75 GiB5.34 GiB
Q4_K_M4.58 GiB
Licencellama3.1llama3

KV cache by context

the term that decides long-context viability
ContextLlama-3.1-8B-Instruct-hereticllama-3-youko-8bRatio
4,0960.50 GiB
8,1921.00 GiB
16,3842.00 GiB
32,7684.00 GiB
65,5368.00 GiB
131,07216.00 GiB
Llama-3.1-8B-Instruct-heretic vs llama-3-youko-8b — size, memory and hardware fit — ossmodeldb