LangChain Quiz

Answers and explanations appear as you take the quiz.

1. What is the central idea behind LangChain?
   • Composing and orchestrating multi-step LLM-powered applications
   • Training large language models from scratch
   • Hosting GPUs for model inference
   • Generating images from text prompts
2. Who created LangChain and roughly when?
   • Sam Altman in 2020
   • Harrison Chase in late 2022
   • Lukas Biewald in 2017
   • Jerry Liu in 2023
3. What is LangGraph used for within the LangChain ecosystem?
   • Drawing charts of training metrics
   • Compressing prompts to save tokens
   • Building stateful, controllable agent workflows as graphs
   • Converting code between programming languages
4. What is the primary role of LangSmith?
   • Generating synthetic datasets
   • Serving web pages
   • Encrypting API keys
   • Tracing, debugging, and evaluating LLM applications
5. In LangChain, what does the pipe syntax 'prompt | model | parser' represent?
   • Composing components via the shared Runnable interface
   • A shell command for running scripts
   • A database query
   • A way to delete chains
6. What happens when a LangGraph node invokes interrupt() with a checkpointer configured?
   • The graph discards its state and restarts training
   • The graph pauses with state saved, then can resume after external input
   • The graph converts the interrupt into a stateless chain
   • The graph changes the base model weights
7. What kind of evaluation does LangSmith describe as monitoring application traces in production?
   • Offline evaluation on a fixed pre-deployment dataset only
   • A test that fine-tunes the application’s foundation model
   • Online evaluation that checks selected production runs for quality signals
   • An evaluation that reports only GPU training throughput
8. What does an offline LangSmith evaluation test?
   • Only GPU utilization during model training
   • Every production user response with no reference examples
   • A guarantee of correctness on unseen inputs
   • An application against a curated dataset before deployment