Is Huihui-gemma-3n-E4B-it-abliterated or llama-3-youko-8b smaller?

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

Which uses less memory at long context?

Huihui-gemma-3n-E4B-it-abliterated — its KV cache at 32K context is 8.1x smaller, because of how its attention layers cache.

Model comparison

Huihui-gemma-3n-E4B-it-abliterated vs llama-3-youko-8b

These two publish different quantization sets; the table below has the exact sizes. At long context the gap widens: Huihui-gemma-3n-E4B-it-abliterated's KV cache at 32K is 8.1× smaller, which usually matters more than the difference in weights.

From the file· summed bytes, KV per layer

Side by side

	Huihui-gemma-3n-E4B-it-abliterated	llama-3-youko-8b
Parameters	7.8B	8.0B
Architecture	gemma3n	llama
Layers	35	32
Native context	32,768	8,192
Mixture of experts	no	no
Quantizations published	33	2
Smallest quantization	2.57 GiB	5.34 GiB
Q4_K_M	3.95 GiB	—
Licence	gemma	llama3

KV cache by context

the term that decides long-context viability

Context	Huihui-gemma-3n-E4B-it-abliterated	llama-3-youko-8b	Ratio
4,096	0.11 GiB	0.50 GiB	4.57×
8,192	0.16 GiB	1.00 GiB	6.10×
16,384	0.27 GiB	2.00 GiB	7.31×
32,768	0.49 GiB	4.00 GiB	8.13×
65,536	0.93 GiB	8.00 GiB	8.61×
131,072	1.80 GiB	16.00 GiB	8.87×