Instructions to use aaardpark/Qwen2.5-32B-Instruct-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use aaardpark/Qwen2.5-32B-Instruct-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="aaardpark/Qwen2.5-32B-Instruct-GGUF",
	filename="Qwen2.5-32B-Instruct-aaardpark-Q3_K_M.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use aaardpark/Qwen2.5-32B-Instruct-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M
# Run inference directly in the terminal:
llama cli -hf aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M
# Run inference directly in the terminal:
llama cli -hf aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M
# Run inference directly in the terminal:
./llama-cli -hf aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M

Use Docker

docker model run hf.co/aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M

LM Studio
Jan
Ollama
How to use aaardpark/Qwen2.5-32B-Instruct-GGUF with Ollama:
```
ollama run hf.co/aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M
```

Unsloth Studio

How to use aaardpark/Qwen2.5-32B-Instruct-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for aaardpark/Qwen2.5-32B-Instruct-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for aaardpark/Qwen2.5-32B-Instruct-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for aaardpark/Qwen2.5-32B-Instruct-GGUF to start chatting

How to use aaardpark/Qwen2.5-32B-Instruct-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use aaardpark/Qwen2.5-32B-Instruct-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use aaardpark/Qwen2.5-32B-Instruct-GGUF with Docker Model Runner:
```
docker model run hf.co/aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M
```

Lemonade

How to use aaardpark/Qwen2.5-32B-Instruct-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull aaardpark/Qwen2.5-32B-Instruct-GGUF:Q3_K_M

Run and chat with the model

lemonade run user.Qwen2.5-32B-Instruct-GGUF-Q3_K_M

List all available models

lemonade list

Qwen2.5-32B-Instruct — GGUF (aaardpark)

15 GB Q3_K_M GGUF. Runs on any 24 GB machine at 22-25 tok/s with full reasoning capabilities.

Need more capability? See aaardpark/Qwen2.5-72B-Instruct-GGUF — 35 GB, 88% GSM8K.

Quick stats

File	Size	BPW	Min RAM	Speed (M5 Max, Metal)
`Qwen2.5-32B-Instruct-aaardpark-Q3_K_M.gguf`	15 GB	3.9	24 GB	22-25 tok/s

How to use

Download

huggingface-cli download aaardpark/Qwen2.5-32B-Instruct-GGUF \
  Qwen2.5-32B-Instruct-aaardpark-Q3_K_M.gguf --local-dir .

Run

llama.cpp:

llama-cli -m Qwen2.5-32B-Instruct-aaardpark-Q3_K_M.gguf -ngl 99 -p "Hello!"

LM Studio: Search for aaardpark/Qwen2.5-32B-Instruct-GGUF in the model browser.

Prompt format

This model uses the ChatML template:

<|im_start|>system
You are a helpful assistant.<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant

Benchmarks

Base model evaluation (lm-evaluation-harness)

Metric	FP16	This Quant (3-bit)	RTN 3-bit
Perplexity (wikitext-2)	3.049	3.84	3.94
GSM8K (5-shot)	82%	80%	78%
BoolQ	87%	85%	82%
TruthfulQA	56.4%	54.4%	54.5%

Measured on Qwen2.5-32B (base) with lm-evaluation-harness. The quantization method is identical for base and Instruct variants.

GGUF perplexity (wikitext-2, llama.cpp)

Variant	PPL
Instruct Q3_K_M	5.503

Why 32B at 3-bit matters

15 GB is smaller than most 7B FP16 models — but with 32B-class reasoning. This file fits on hardware most people already own:

A 16 GB MacBook Air (tight, short context)
Any 24 GB GPU (RTX 4090 / 4080 / 3090)
32 GB Mac mini / MacBook Pro (comfortable, long context)

Why this quant is different

Standard 3-bit quantization (RTN) rounds each weight to the nearest grid point uniformly. Our method uses calibration data to identify which weights are critical for model quality, then allocates quantization precision accordingly.

On the 72B variant of this method, the difference is dramatic: GSM8K drops from 90% to 16% with RTN at 3-bit, but only to 88% with our approach. Same bit budget, completely different reasoning quality.

Which file should I choose?

This file is 15 GB. Realistic RAM requirements:

≥32 GB RAM: comfortable, full 128K context window
24 GB RAM: works with 16K-32K context — typical for an RTX 4090
16 GB RAM: tight, short context only
More capability needed? The 72B variant (35 GB) handles harder reasoning tasks that this 32B can't.

On Apple Silicon with Metal offload (-ngl 99), expect 22-25 tok/s on M5 Max. NVIDIA GPUs need ~17 GB VRAM for full offload.

Method

Importance-weighted per-group optimization. Calibration data identifies which weights are critical for model quality, then quantization precision is allocated accordingly. ~20 minutes per quant on a single GPU. Output is standard Q3_K_M GGUF format — no custom kernels required.

Group size: 128
GGUF format: Q3_K_M (via llama.cpp)
Context: 128K tokens

Acknowledgments

Built on Qwen/Qwen2.5-32B-Instruct by Alibaba Cloud.

Downloads last month: 12

GGUF

Model size

33B params

Architecture

qwen2

Hardware compatibility

3-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for aaardpark/Qwen2.5-32B-Instruct-GGUF

Base model

Qwen/Qwen2.5-32B

Finetuned

Qwen/Qwen2.5-32B-Instruct

Quantized

(147)

this model