Write a Python function that reads a CSV file, filters rows where the
"status" column equals "active", and returns the results as a list of
dictionaries. Include error handling for missing files and malformed CSV data.

# 從終端機用 Gemma 4 作為程式設計助理
cat myfile.py | ollama run gemma4:12b "Review this code and suggest improvements:"

I'm going to paste a 10-page contract below. Please:
1. Summarize the key terms in 5 bullet points
2. Identify any unusual clauses that might be concerning
3. List all deadlines and dates mentioned
4. Flag any ambiguous language that should be clarified

[paste contract text here]

Translate the following English marketing copy to natural, conversational
Japanese. The target audience is tech-savvy professionals in their 30s.
Avoid overly formal language — aim for a friendly but professional tone.

"Our new app helps you organize your work and life in one place.
No more switching between five different tools just to get through your day."

import ollama

response = ollama.chat(
    model="gemma4:12b",
    messages=[{
        "role": "user",
        "content": "Describe what's in this image and extract any visible text.",
        "images": ["./screenshot.png"]
    }]
)
print(response["message"]["content"])

Write a 300-word blog introduction about the benefits of running AI models
locally instead of using cloud APIs. The tone should be practical and
slightly opinionated — not corporate or generic. Target audience: developers
who are curious about local AI but haven't tried it yet.

Avoid clichés like "in today's rapidly evolving landscape" or
"game-changing". Start with a concrete scenario, not an abstract statement.

Extract all line items from this invoice and format them as JSON:

[paste invoice text here]

Expected format:
{
  "invoice_number": "...",
  "date": "...",
  "items": [
    {"description": "...", "quantity": 0, "unit_price": 0.00, "total": 0.00}
  ],
  "subtotal": 0.00,
  "tax": 0.00,
  "total": 0.00
}

You are a helpful customer support agent for TechCo, a software company.

Rules:
- Always be polite and professional
- If you don't know the answer, say so honestly and suggest contacting
  support@techco.com
- Never make up product features or pricing
- Keep responses concise (2-3 sentences for simple questions)
- For billing issues, always recommend contacting the billing team directly

Product info:
- TechCo Pro: $29/month, includes 5 users, 100GB storage
- TechCo Enterprise: $99/month, unlimited users, 1TB storage
- Free trial: 14 days, no credit card required

Review the following Python code for:
1. Potential bugs or edge cases
2. Security vulnerabilities
3. Performance issues
4. Code style improvements
5. Missing error handling

Be specific — point to exact lines and explain why each issue matters.

```python
def get_user_data(user_id):
    conn = sqlite3.connect("users.db")
    cursor = conn.cursor()
    result = cursor.execute(f"SELECT * FROM users WHERE id = {user_id}")
    data = result.fetchone()
    return {"id": data[0], "name": data[1], "email": data[2]}

Gemma 4 會正確識別 SQL injection 漏洞、缺少的連線清理、對 null 結果缺少錯誤處理，並建議使用參數化查詢和 context manager。

---

## 9. 研究摘要

**建議模型：** Gemma 4 12B 或 27B

將長篇研究論文、技術文件或文章集合濃縮為可行動的摘要。

**範例提示詞：**

**批次處理技巧：** 使用 Ollama API 以程式方式處理多篇論文：

```python
import ollama

papers = ["paper1.txt", "paper2.txt", "paper3.txt"]

for paper_path in papers:
    with open(paper_path) as f:
        content = f.read()

    response = ollama.chat(
        model="gemma4:12b",
        messages=[{
            "role": "user",
            "content": f"Summarize this paper in 3 bullet points:\n\n{content}"
        }]
    )
    print(f"\n--- {paper_path} ---")
    print(response["message"]["content"])

Summarize the following patient visit notes into a structured format:
- Chief complaint
- Key findings
- Diagnosis
- Treatment plan
- Follow-up needed (yes/no, when)

Keep the summary under 100 words. Use medical terminology where appropriate.

[paste notes here]

ollama pull gemma4:12b
ollama run gemma4:12b