Glossary / KV cache
KV cacheKV cache
Memory where the model keeps what it already read, so it does not recompute it for every new word.
Memori tempat model menyimpan yang sudah dibaca, supaya tidak menghitung ulang untuk setiap kata baru.
Example
Contoh
It grows with the length of the context.
Ukurannya naik seiring panjang konteks.
Why it matters
Kenapa penting
A long context can need as much memory as the model itself.
Konteks panjang bisa butuh memori sebesar modelnya.
See it in the calculator →Lihat di kalkulator →