dennisonb/qwen25-tax-3b-v3-GGUF

qwen25-tax-3b-v3 (GGUF)

A Q8_0 quantized GGUF of the v3 IRS Tax Code model.

Model Description

Base model: Qwen/Qwen2.5-3B-Instruct Fine-tuning pipeline: SFT → DPO → GRPO Training data: IRC Title 26 (Internal Revenue Code), U.S. Code of Federal Regulations Title 26

This is v3 of the IRS Tax Code RL project. The model was trained in three stages:

1. SFT (Supervised Fine-Tuning): Grounded question-answer pairs from IRC Title 26
2. DPO (Direct Preference Optimization): Preference data generated from SFT model outputs
3. GRPO (Group Relative Policy Optimization): RL fine-tuning with reward signals based on citation accuracy and answer correctness

Files

File	Description
qwen25-tax-3b-v3-q8_0.gguf	Q8_0 quantized GGUF, ~3.1GB

Usage (Ollama)

ollama run hf.co/dennisonb/qwen25-tax-3b-v3-GGUF

Usage (llama.cpp)

./llama-cli -m qwen25-tax-3b-v3-q8_0.gguf -p "What is the standard deduction for 2024?"

Intended Use

This model is intended for research and educational purposes related to U.S. tax law (IRC Title 26). It is NOT a substitute for professional tax advice.