This project provides more finegrained human-model comparison on lexical level - View it on GitHub