THE MODEL WE TRAINED
A small mind.
Your machine.
Give our language model a beginning. It writes the next part entirely in your browser.
ITSE KOULUTTAMAMME MALLI
Pieni malli.
Sinun koneesi.
Anna kielimallille tarinan alku. Malli jatkaa tekstiä kokonaan selaimessasi.
19.4M parameters512 bytes of contextTrained model
Ready when you are.
What it remembers
The prompt is processed together in one prefill pass. Then each new byte reuses the keys and values already calculated—the KV cache below.
Prompt prefill—
Generation—
KV cache0 MiB
Runtime—
0 / 512
This model reads UTF-8 bytes, not words. Generation stops at a total of 512 bytes; it does not silently forget the beginning.