Instructions to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath 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 U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M

Use Docker

docker model run hf.co/U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M

SGLang

How to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath with Ollama:
```
ollama run hf.co/U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M
```
Unsloth Desktop

How to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath with Docker Model Runner:
```
docker model run hf.co/U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M
```

Lemonade

How to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M

Run and chat with the model

lemonade run user.Qwen2.5-Math-1.5B-HigherMath-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_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 U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M

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 "U-Ug1kn/Qwen2.5-Math-1.5B-HigherMath:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

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

qwen-math-higher-math

qwen-math-higher-math is a fine-tuned version of Qwen2.5-Math-1.5B-Instruct, specialized for university-level mathematics: calculus, linear algebra, ordinary differential equations, series, multivariable calculus, complex analysis, and probability.

The model produces step-by-step Chain-of-Thought solutions in English, uses LaTeX notation, and finishes with the final answer in \boxed{}.

Model details

Property	Value
Base model	Qwen/Qwen2.5-Math-1.5B-Instruct
Parameters	1.5B
Fine-tuning data	1,500 filtered examples from `AI-MO/NuminaMath-CoT`
Target domain	University-level mathematics
Output format	Step-by-step reasoning, LaTeX, final answer in `\boxed{}`
Inference mode	Chain-of-Thought only; no tool use / Python / calculator
Quantized artifact	GGUF Q4_K_M, approximately 0.92 GB
Runtime	Ollama, CPU inference

Intended use

This model is intended for:

University-level math problem solving
Calculus: limits, derivatives, integrals, series
Linear algebra: matrices, determinants, eigenvalues
Ordinary differential equations
Multivariable calculus and vector calculus
Complex analysis
Probability and statistics
Generating step-by-step explanations with LaTeX

It is not intended as a general-purpose assistant, coding assistant, or tool-using agent.

Benchmarks

All results were obtained with greedy decoding (temperature = 0), one generation per problem (pass@1), and answer extraction from the final \boxed{} expression. The model was evaluated locally through Ollama using a Q4_K_M GGUF quantization.

Benchmark	Baseline: Qwen2.5-Math-1.5B-Instruct	qwen-math-higher-math	Difference
Custom University Math-200	98.50%	99.50%	+1.0%
MATH-500	74.60%	73.60%	-1.0%
GSM8K-100	86.00%	82.00%	-4.0%

Custom University Math-200

Custom University Math-200 is a held-out set of 200 university-level problems covering:

Limits and continuity
Derivatives
Integrals
Multivariable calculus
Vector calculus
Linear algebra
ODEs and PDEs
Series
Complex analysis
Probability and statistics

This is a custom evaluation set, not the official CollegeMath benchmark. The result should be interpreted as a domain-specific evaluation rather than a standardized public benchmark score.

MATH-500

MATH-500 is a 500-problem subset of the MATH benchmark. Answers are evaluated by exact match after extracting the final answer from \boxed{}.

GSM8K-100

GSM8K-100 is a 100-problem subset of GSM8K, used as a general mathematical reasoning sanity check.

Results interpretation

The fine-tuned model improved on the custom university-mathematics set by 1 percentage point, while showing a small decline on MATH-500 and GSM8K.

Because the baseline already scored 98.5% on the custom set, the available headroom was limited. The results suggest that the fine-tune specializes the model for higher mathematics, but does not improve general competition-level math performance.

The differences are small relative to the benchmark sizes, particularly for the 100-problem GSM8K subset. Larger evaluations and an error analysis are needed before drawing strong conclusions.

Usage

Ollama

ollama create qwen-math-higher-math -f Modelfile
ollama run qwen-math-higher-math

Example Modelfile:

FROM ./qwen-math-q4_k_m.gguf
PARAMETER temperature 0
PARAMETER num_ctx 4096
PARAMETER stop "<|im_end|>"

Python

from ollama import chat

response = chat(
    model="qwen-math-higher-math",
    messages=[
        {
            "role": "system",
            "content": (
                "Please reason step by step. "
                "Use LaTeX for mathematical expressions. "
                "Do not use Python, a calculator, or any external tools. "
                "Put your final answer inside \\boxed{}."
            )
        },
        {
            "role": "user",
            "content": "Find the derivative of f(x) = x^3 ln(x)."
        }
    ],
    options={
        "temperature": 0,
        "num_predict": 2048,
    },
)

print(response["message"]["content"])

Recommended prompt

Please reason step by step.
Use LaTeX for mathematical expressions.
Do not use Python, a calculator, or any external tools.
Put your final answer inside \boxed{}.

Problem:
{problem}

Example

Input:

Find the derivative of f(x) = x^3 ln(x).

Expected output format:

To find the derivative of f(x) = x^3 ln(x), we use the product rule.

f'(x) = 3x^2 ln(x) + x^3 * (1/x)
      = x^2(3 ln(x) + 1)

\boxed{x^2(3\ln(x)+1)}

Limitations

The model is specialized for university-level mathematics and may perform worse on general conversation, coding, and non-mathematical tasks.
It does not use external tools, calculators, or code execution.
Long Chain-of-Thought outputs may be truncated if the context or generation limit is too small.
The model may produce plausible but incorrect reasoning, especially on unfamiliar or highly complex problems.
Benchmark scores are sensitive to prompt format, decoding settings, answer extraction, and answer-equivalence checking.
The custom university-math benchmark has not yet been fully checked for overlap with the fine-tuning data.

Training data

The model was fine-tuned on 1,500 problems selected from AI-MO/NuminaMath-CoT.

The selected problems focus on:

Limits and continuity
Derivatives and differentiation rules
Definite, indefinite, improper, and multiple integrals
Series and convergence tests
First- and second-order ordinary differential equations
Matrices, determinants, and eigenvalues
Probability, Bayes’ theorem, and combinatorics

The training data uses Chain-of-Thought solutions. No tool-integrated reasoning, Python execution, or calculator calls were used.

Evaluation methodology

Model: qwen-math-higher-math, Q4_K_M GGUF
Runtime: Ollama on CPU
Decoding: greedy, temperature = 0
Generation limit: 2,048 tokens
Evaluation: one generation per problem, pass@1
Answer extraction: last valid \boxed{} expression
Answer checking: mathematical equivalence via math-verify
Baseline: original Qwen/Qwen2.5-Math-1.5B-Instruct

Citation

If you use this model, please cite the underlying Qwen2.5-Math work and NuminaMath dataset:

@article{qwen2.5math,
  title={Qwen2.5-Math Technical Report: Toward Mathematical Expert Model via Self-Improvement},
  author={Qwen Team},
  journal={arXiv preprint arXiv:2409.12122},
  year={2024}
}

@misc{numina,
  title={NuminaMath},
  author={AI-MO},
  howpublished={\url{[https://huggingface.co/datasets/AI-MO/NuminaMath-CoT](https://huggingface.co/datasets/AI-MO/NuminaMath-CoT)}},
  year={2024}
}