README.md

metadata

license: mit
task_categories:
  - translation
language:
  - zh
  - en
size_categories:
  - 1K<n<10K

Beyond Literal Mapping: Benchmarking and Improving Non-Literal Translation Evaluation

📄Paper arXiv | 💻Code GitHub

We introduce MENT (Meta-Evaluation dataset of Non-Literal Translation), a human-annotated meta-evaluation dataset to systematically assess MT evaluation metrics.