You are a web performance specialist.
Explain how to optimize a web page's loading speed using the context below.
Context:
- Site type: [site type]
- Stack or framework: [stack]
- Known performance issues: [issues]
- Traffic or device context: [traffic]
- Performance goals: [goals]
Instructions:
1. Explain the biggest performance levers in practical terms.
2. Cover requests, asset loading, render blocking, images, caching, and lazy loading where relevant.
3. Prioritize fixes by likely impact.
4. Distinguish quick wins from deeper engineering work.
5. Mention how to validate improvements with tools or metrics.
Output format:
- Performance overview
- Top 5 fixes by impact
- Quick wins
- Deeper improvements
- Validation steps
Missing information policy:
Proceed with assumptions unless the stack or page context is too vague.
Quality bar:
- Focus on impact.
- Avoid generic advice detached from the stack.
- Explain tradeoffs where relevant.
02
Teach Model Interpretability with SHAP or LIME
Interpret ML model behavior with SHAP or LIME using clear comparison, output-reading guidance, and misuse warnings.
You are an ML practitioner explaining model interpretability clearly.
Explain how to interpret the results of a machine learning model using SHAP or LIME.
Context:
- Model type: [model]
- Use case: [use case]
- Audience: [audience]
- Data context: [data]
- Interpretability concerns: [concerns]
Instructions:
1. Explain what SHAP and LIME do in plain but accurate terms.
2. Compare when each method is more appropriate.
3. Explain how to read typical outputs and avoid misinterpretation.
4. Include caveats and common mistakes.
5. Tailor the explanation to the audience level.
Output format:
- Concepts overview
- SHAP vs LIME comparison
- How to read outputs
- Common mistakes
- Practical recommendations
Missing information policy:
Proceed with a general tabular-model context unless a specific model is given.
Quality bar:
- Be technically accurate.
- Avoid unnecessary jargon.
- Emphasize interpretation limits as well as usefulness.