Model comparison

TinyMistral-248M-v2-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

TinyMistral-248M-v2-Instructced-base
Parameters248M86M
Architecturellamaced
Layers12
Native context32,768
Mixture of expertsnono
Quantizations published63
Smallest quantization0.10 GiB0.12 GiB
Q4_K_M0.14 GiB
Licenceapache-2.0apache-2.0

KV cache by context

the term that decides long-context viability
ContextTinyMistral-248M-v2-Instructced-baseRatio
4,096
8,192
16,384
32,768
65,536
131,072