Model comparison

t5-v1_1-xxl 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 2,896,123,072.

From the file· summed bytes, KV per layer

Side by side

t5-v1_1-xxlHyperCLOVAX-SEED-Text-Instruct-1.5B
Parameters4.8B1.6B
Architecturet5encoderllama
Layers
Native context
Mixture of expertsnono
Quantizations published431
Smallest quantization1.65 GiB1.06 GiB
Q4_K_M2.70 GiB1.06 GiB
Licenceapache-2.0other

KV cache by context

the term that decides long-context viability
Contextt5-v1_1-xxlHyperCLOVAX-SEED-Text-Instruct-1.5BRatio
4,096
8,192
16,384
32,768
65,536
131,072