Don't Trust: Verify -- Grounding LLM Quantitative Reasoning with Autoformalization