Chinese LLM NLP Best Practices: Prompt Engineering Guide

Chinese Prompt Characteristics

Chinese models have a natural advantage in Chinese understanding. Good Chinese prompts should be concise, direct, and structured — avoiding the lengthy formats common in English prompts.

Summarization Task

{"model":"qwen3.6-plus","messages":[{"role":"user","content":"Please compress the following article into a 200-word summary, highlighting key points:
{article_content}"}]}

Translation Task

{"model":"qwen3.6-plus","messages":[{"role":"user","content":"Please translate the following into a professional English business letter:
{content}"}]}

Q&A Task

{"model":"qwen3.6-plus","messages":[{"role":"system","content":"You are a professional technical support engineer. Answer user questions based on the following documents. If the document doesn't contain the relevant information, say so honestly."},{"role":"user","content":"Document: {document}

Question: {question}"}]}

General Tips

Structured output: use JSON format to specify output structure; Few-shot examples: provide 2-3 examples to help the model understand the task; Temperature: 0.7-0.9 for creative tasks, 0.1-0.3 for accuracy tasks.