LLM fundamentals & prompting
Tokens, context windows, sampling, prompt structure, structured output, caching, model selection, fine-tuning vs prompting.
- A customer asks why their bill is 30% higher after moving to a newer model with the same prompts. What is going on, and how do you reason about tokens? · easyFree
- Explain the context window to a non-technical customer exec, and tell me what actually counts against it. · easy · reported 1×Free
- A customer's agent runs long tasks with lots of chat history and tool output, and quality drops over time. How do you manage context? · medium · reported 1×Free
- What does temperature do, and how would you set sampling parameters for a customer's extraction pipeline? · easyFree
- Walk me through how you structure a production prompt for a customer workflow. · medium · reported 2×Free
- The customer's downstream system needs valid JSON every time. How do you get reliable structured output? · mediumPro
- What is prompt caching, and how much could it save a customer with a large, repeated system prompt? · mediumPro
- You turned on prompt caching but the bill did not drop. How do you debug it? · hardPro
- How do you choose which model to use for a customer's use case? · mediumPro
- The customer wants to fine-tune a model on their data. How do you decide between prompting, RAG and fine-tuning? · hardPro
- The model keeps making things up in a customer-facing summary. What do you do about hallucinations? · easyPro
- How do you make an LLM's output consistent enough for production? · medium · reported 1×Pro
- A customer will process 100,000 documents a day. Estimate the LLM cost and tell me how you would bring it down. · hardPro
- When would you turn on extended reasoning or raise the effort setting, and what does it cost you? · mediumPro
- A prompt works great in your playground tests but fails about 5% of the time in the customer's production traffic. How do you find out why? · hardPro