GGUFGGUF

A single-file format for running models locally. Used by llama.cpp and tools built on it, such as Ollama. It holds the weights (often quantized) plus metadata.

Format satu file untuk menjalankan model secara lokal. Dipakai llama.cpp dan alat turunannya seperti Ollama. Isinya bobot model (sering sudah dikuantisasi) plus metadata.

Example

Contoh

A file named model-Q4_K_M.gguf is a 4-bit version.

Nama file model-Q4_K_M.gguf berarti versi 4-bit.

Why it matters

Kenapa penting

If you download a model for a laptop, you usually want the GGUF file.

Kalau mengunduh model untuk laptop, biasanya yang kamu cari file GGUF.

More terms

Istilah lain