HumanEval with LangChain

📖 Related article: Evaluating code generation agents—LangChain and CodeChain

This is a demonstration of how to run HumanEval on GPT-3.5 and GPT-4 while taking advantage of LangSmith's visibility and tracing features:

Related repositories

• human-eval: Fork of OpenAI's HumanEval framework used in this workflow.
• humaneval-results: Repository of HumanEval solutions generated with this workflow.
• codechain: A simple library for generating code with LLMs.
• agenteval: Early version of a framework for evaluating code generation agents.