📝 New preprint out on the limits of LLMs in lexical entailment recognition 🔬