Views
No views yet
unsloth/llama-3.2-3b-instruct-bnb-4bit, fine-tuned on Rimyy/problemMath-Llama3.5K, a high-quality dataset of math reasoning and problem-solving questions. The model is tailored for math instruction, step-by-step reasoning, and educational applications.🚨 Designed to reason, not just regurgitate. Small model, big brain.
| Feature | Value |
|---|---|
| Base | unsloth/llama-3.2-3b-instruct-bnb-4bit |
| Finetuned Dataset | Rimyy/problemMath-Llama3.5K |
| Quantization | 4-bit GGUF (compatible with llama.cpp/text-generation-webui) |
| Format | GGUF |
| Language | English |
| Instruction Tuned | ✅ Yes |
Rimyy/problemMath-Llama3.5K1./main -m llama-3.2-3b-math.gguf --prompt "### Question: What is the value of x if x + 3 = 7?
2### Answer:"To solve for x, subtract 3 from both sides of the equation:
x + 3 = 7
x = 7 - 3
x = 4
Answer: 41from llama_cpp import Llama
2
3llm = Llama(
4 model_path="llama-3.2-3b-instruct-math.q4_K.gguf",
5 n_ctx=2048,
6 n_gpu_layers=32, # adjust based on your GPU
7)
8
9prompt = (
10 "### Question: If a rectangle has length 10 and width 5, what is its area?
11"
12 "### Answer:"
13)
14response = llm(prompt)
15print(response["choices"][0]["text"])| Task Type | Performance |
|---|---|
| Simple Arithmetic | ✅ Excellent |
| One-Step Algebra | ✅ Strong |
| Multi-Step CoT | ⚠️ Good (some drift) |
| Logic Puzzles | ⚠️ Mixed |
📌 Quantitative benchmarks forthcoming.
1@misc{rimyy2025math,
2 author = {Rimyy},
3 title = {ProblemMath-Llama3.5K: A Dataset for Math Problem Solving},
4 year = {2025},
5 url = {https://huggingface.co/datasets/Rimyy/problemMath-Llama3.5K}
6}