Is magnum-v2.5-12b-kto or llama-3-youko-8b smaller?

Both publish quantizations; compare the table for the exact sizes.

Which uses less memory at long context?

llama-3-youko-8b — its KV cache at 32K context is 1.3x smaller, because of how its attention layers cache.

Model comparison

magnum-v2.5-12b-kto vs llama-3-youko-8b

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: llama-3-youko-8b's KV cache at 32K is 1.3× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	magnum-v2.5-12b-kto	llama-3-youko-8b
Parameters	12.2B	8.0B
Architecture	llama	llama
Layers	40	32
Native context	1,024,000	8,192
Mixture of experts	no	no
Quantizations published	46	2
Smallest quantization	2.79 GiB	5.34 GiB
Q4_K_M	6.96 GiB	—
Licence	apache-2.0	llama3

KV cache by context

the term that decides long-context viability

Context	magnum-v2.5-12b-kto	llama-3-youko-8b	Ratio
4,096	0.63 GiB	0.50 GiB	1.25×
8,192	1.25 GiB	1.00 GiB	1.25×
16,384	2.50 GiB	2.00 GiB	1.25×
32,768	5.00 GiB	4.00 GiB	1.25×
65,536	10.00 GiB	8.00 GiB	1.25×
131,072	20.00 GiB	16.00 GiB	1.25×