Model comparison

Falcon3-1B-Instruct vs ced-base

These two publish different quantization sets; the table below has the exact sizes.

From the file· summed bytes, KV per layer

Side by side

Falcon3-1B-Instructced-base
Parameters1.7B86M
Architecturellamaced
Layers18
Native context8,192
Mixture of expertsnono
Quantizations published303
Smallest quantization0.21 GiB0.12 GiB
Q4_K_M0.98 GiB
Licenceotherapache-2.0

KV cache by context

the term that decides long-context viability
ContextFalcon3-1B-Instructced-baseRatio
4,0960.28 GiB
8,1920.56 GiB
16,3841.13 GiB
32,7682.25 GiB
65,5364.50 GiB
131,0729.00 GiB