navyaparesh/llama-3.2-3b-q8-k-m

Llama 3.2B — Quantized (q8_0, GGUF)

This repository provides an 8-bit quantized version of Meta's Llama 3.2B for efficient deployment on resource-constrained environments (CPU and small GPUs). The GGUF file uses q8_0 quantization (8-bit) — a good tradeoff between size and quality for small models.Please refer to the original model card for full details on its capabilities and limitations.

Base model: Llama 3.2B (Meta AI)
Quantization: 8-bit Post-Training Quantization (q8_0) — GGUF.
Format: GGUF (compatible with llama.cpp, GPT4All, Ollama).
Model file: llama_3.2_3b_q8_k_m.gguf

---

Usage (llama.cpp)

# Run with llama.cpp
git clone https://github.com/ggerganov/llama.cpp
cd llama.cpp
make

# then:
./main -m ./llama_3.2_3b_q8_k_m.gguf -p "Hello, how are you?"

Download

You can download this model directly via:

git lfs install
git clone https://huggingface.co/navyaparesh/llama-3.2-3b-q8-k-m

Or programmatically:

from huggingface_hub import snapshot_download
snapshot_download(repo_id="navyaparesh/llama-3.2-3b-q4-k-m", local_dir="models/llama3-quantized")