zadaniamm/qwen-2.5-7b-mai-friend-merged-bf16-IQ4_NL-GGUF
This model was converted to GGUF format from zadaniamm/qwen-2.5-7b-mai-friend-merged-bf16 using llama.cpp via the ggml.ai's GGUF-my-repo space.
Refer to the original model card for more details on the model.
Kuantisasi & Kalibrasi (Importance Matrix)
Model ini dikuantisasi menggunakan metode modern IQ4_NL (Importance Matrix Non-Linear). Karena menggunakan arsitektur IQ, proses penghitungan bobot matriks kepentingannya dikalibrasi menggunakan dataset bawaan sistem (default fallback imatrix dataset):
- Dataset Kalibrasi:
wiki.train.raw(Bagian dari dataset standar WikiText-103 yang bersumber dari artikel pilihan Wikipedia bahasa Inggris). - Fungsi: Menjaga logika attention layers dan mencegah penurunan kecerdasan model secara drastis meskipun ukuran file telah dikompres secara signifikan ke tingkat 4-bit.
Use with llama.cpp
Install llama.cpp through brew (works on Mac and Linux)
brew install llama.cpp
Invoke the llama.cpp server or the CLI.
CLI:
llama-cli --hf-repo zadaniamm/qwen-2.5-7b-mai-friend-merged-bf16-IQ4_NL-GGUF --hf-file qwen-2.5-7b-mai-friend-merged-bf16-iq4_nl-imat.gguf -p "The meaning to life and the universe is"
Server:
llama-server --hf-repo zadaniamm/qwen-2.5-7b-mai-friend-merged-bf16-IQ4_NL-GGUF --hf-file qwen-2.5-7b-mai-friend-merged-bf16-iq4_nl-imat.gguf -c 2048
Note: You can also use this checkpoint directly through the usage steps listed in the Llama.cpp repo as well.
Step 1: Clone llama.cpp from GitHub.
git clone https://github.com/ggerganov/llama.cpp
Step 2: Move into the llama.cpp folder and build it with LLAMA_CURL=1 flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux).
cd llama.cpp && LLAMA_CURL=1 make
Step 3: Run inference through the main binary.
./llama-cli --hf-repo zadaniamm/qwen-2.5-7b-mai-friend-merged-bf16-IQ4_NL-GGUF --hf-file qwen-2.5-7b-mai-friend-merged-bf16-iq4_nl-imat.gguf -p "The meaning to life and the universe is"
or
./llama-server --hf-repo zadaniamm/qwen-2.5-7b-mai-friend-merged-bf16-IQ4_NL-GGUF --hf-file qwen-2.5-7b-mai-friend-merged-bf16-iq4_nl-imat.gguf -c 2048