Run your LLM evals every night on a GPU that only exists for a minute

· Dev.to

Read full story at source