Model comparison

gemma-2b vs HyperCLOVAX-SEED-Text-Instruct-1.5B

At Q4_K_M, HyperCLOVAX-SEED-Text-Instruct-1.5B is the smaller download — 1,133,974,368 bytes against 1,495,245,728.

From the file· summed bytes, KV per layer

Side by side

gemma-2bHyperCLOVAX-SEED-Text-Instruct-1.5B
Parameters2.5B1.6B
Architecturegemmallama
Layers18
Native context8,192
Mixture of expertsnono
Quantizations published381
Smallest quantization0.84 GiB1.06 GiB
Q4_K_M1.39 GiB1.06 GiB
Licencecc-by-nc-4.0other

KV cache by context

the term that decides long-context viability
Contextgemma-2bHyperCLOVAX-SEED-Text-Instruct-1.5BRatio
4,0960.07 GiB
8,1920.14 GiB
16,3840.28 GiB
32,7680.56 GiB
65,5361.13 GiB
131,0722.25 GiB