Instructions to use mrrahman82/qwen-math-0.5b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use mrrahman82/qwen-math-0.5b with MLX:

# Make sure mlx-lm is installed
# pip install --upgrade mlx-lm

# Generate text with mlx-lm
from mlx_lm import load, generate

model, tokenizer = load("mrrahman82/qwen-math-0.5b")

prompt = "Write a story about Einstein"
messages = [{"role": "user", "content": prompt}]
prompt = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True
)

text = generate(model, tokenizer, prompt=prompt, verbose=True)

llama-cpp-python

How to use mrrahman82/qwen-math-0.5b with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="mrrahman82/qwen-math-0.5b",
	filename="qwen-math-Q4_K_M.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use mrrahman82/qwen-math-0.5b 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 mrrahman82/qwen-math-0.5b:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf mrrahman82/qwen-math-0.5b:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf mrrahman82/qwen-math-0.5b:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf mrrahman82/qwen-math-0.5b:Q4_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 mrrahman82/qwen-math-0.5b:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf mrrahman82/qwen-math-0.5b:Q4_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 mrrahman82/qwen-math-0.5b:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf mrrahman82/qwen-math-0.5b:Q4_K_M

Use Docker

docker model run hf.co/mrrahman82/qwen-math-0.5b:Q4_K_M

LM Studio
Jan

vLLM

How to use mrrahman82/qwen-math-0.5b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "mrrahman82/qwen-math-0.5b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "mrrahman82/qwen-math-0.5b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/mrrahman82/qwen-math-0.5b:Q4_K_M

Ollama
How to use mrrahman82/qwen-math-0.5b with Ollama:
```
ollama run hf.co/mrrahman82/qwen-math-0.5b:Q4_K_M
```

Unsloth Studio

How to use mrrahman82/qwen-math-0.5b 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 mrrahman82/qwen-math-0.5b 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 mrrahman82/qwen-math-0.5b to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for mrrahman82/qwen-math-0.5b to start chatting

How to use mrrahman82/qwen-math-0.5b with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "mrrahman82/qwen-math-0.5b"

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "mrrahman82/qwen-math-0.5b"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use mrrahman82/qwen-math-0.5b with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "mrrahman82/qwen-math-0.5b"

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 mrrahman82/qwen-math-0.5b

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use mrrahman82/qwen-math-0.5b with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "mrrahman82/qwen-math-0.5b"

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "mrrahman82/qwen-math-0.5b" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

MLX LM

How to use mrrahman82/qwen-math-0.5b with MLX LM:

Generate or start a chat session

# Install MLX LM
uv tool install mlx-lm
# Interactive chat REPL
mlx_lm.chat --model "mrrahman82/qwen-math-0.5b"

Run an OpenAI-compatible server

# Install MLX LM
uv tool install mlx-lm
# Start the server
mlx_lm.server --model "mrrahman82/qwen-math-0.5b"
# Calling the OpenAI-compatible server with curl
curl -X POST "http://localhost:8000/v1/chat/completions" \
   -H "Content-Type: application/json" \
   --data '{
     "model": "mrrahman82/qwen-math-0.5b",
     "messages": [
       {"role": "user", "content": "Hello"}
     ]
   }'

Docker Model Runner
How to use mrrahman82/qwen-math-0.5b with Docker Model Runner:
```
docker model run hf.co/mrrahman82/qwen-math-0.5b:Q4_K_M
```

Lemonade

How to use mrrahman82/qwen-math-0.5b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull mrrahman82/qwen-math-0.5b:Q4_K_M

Run and chat with the model

lemonade run user.qwen-math-0.5b-Q4_K_M

List all available models

lemonade list

qwen-math-0.5b

A LoRA supervised fine-tune of Qwen2.5-0.5B-Instruct that solves grade-school math word problems step by step and ends with a parseable #### <answer> line.

Trained with Apple MLX on a MacBook Air M5 (16 GB), in ~2 minutes, ~1.9 GB peak memory. Full pipeline is reproducible from mrrahman1517/apple-silicon-ml-lab.

Results

Held-out accuracy on 100 hard (3–5 step) word problems:

Model	Accuracy
Qwen2.5-0.5B-Instruct (base)	58%
qwen-math-0.5b (this model)	82% (+24 pts)

Runs at ~~215 tok/s in Ollama on an Apple M5 (~~0.6 GB RAM).

Files

File	Use
`qwen-math-Q4_K_M.gguf`	4-bit GGUF (379 MB) — llama.cpp / Ollama
`qwen-math-f16.gguf`	f16 GGUF (948 MB) — re-quantize as you like
`Modelfile`	Ollama Modelfile (Qwen2.5 ChatML + math system prompt)
`adapters/`	the MLX LoRA adapter (use on top of the 4-bit base)

Usage

Ollama (GGUF)

huggingface-cli download <repo_id> qwen-math-Q4_K_M.gguf Modelfile --local-dir qwen-math
cd qwen-math && ollama create qwen-math -f Modelfile
ollama run qwen-math "A jacket costs $200. Take 25% off, then add 10% tax. Final price?"
# -> ... #### 165

MLX (LoRA adapter on the 4-bit base)

pip install mlx-lm
python -m mlx_lm.generate \
  --model mlx-community/Qwen2.5-0.5B-Instruct-4bit \
  --adapter-path adapters \
  --prompt "A train travels at 60 km/h for 3 hours, then 50 km/h for 2 hours. Total distance?"

Training data

Synthetic GSM8K-style problems with deterministically-correct step-by-step solutions (discount+tax, percent-of-remainder, multi-leg trips, missing-average, two-equation systems). Generator and configs in the source repo.

Limitations

It's a 0.5B model — it still makes ~18% arithmetic slips on hard problems (e.g. summing a list incorrectly), and it's specialized to short numeric word problems. For general use prefer a larger base.

Base model: Qwen2.5-0.5B-Instruct (Apache-2.0).

Downloads last month: 72

MLX

Hardware compatibility

Quantized

GGUF

Model size

0.5B params

Architecture

qwen2

Hardware compatibility

4-bit

16-bit

Model tree for mrrahman82/qwen-math-0.5b

Base model

Qwen/Qwen2.5-0.5B

Finetuned

Qwen/Qwen2.5-0.5B-Instruct

Adapter

(708)

this model