This is a Supervised Fine-Tuned (SFT) model based on Qwen2.5-7B-Instruct, designed specifically for Traditional Chinese multiple-choice question answering and logical reasoning.

This model represents Step 1 of a two-stage training pipeline. It has been strictly fine-tuned to internalize a specific XML output format and build foundational domain knowledge.

Base Model: Qwen2.5-7B-Instruct
Language: Traditional Chinese (zh-TW)
Task: Multiple Choice Question Answering (MCQA)
Training Method: SFT with LoRA

Prompt Template & Output Format

The model is trained to strictly adhere to the following XML structure, ensuring that it generates a detailed reasoning process before outputting the final answer:

xml

<reasoning>
[Step-by-step logical deduction and elimination process]
</reasoning>
<answer>A/B/C/D</answer>

tony-shieh/Qwen2.5-7B-SFT-TW-QA

Qwen2.5-7B-SFT-TW-QA

Model Description

Prompt Template & Output Format

tony-shieh/
Qwen2.5-7B-SFT-TW-QA