Fetching the paper…

What I cannot execute, I do not understand: Training and Evaluating LLMs on Program Execution Traces · Around