MTG: A Benchmark Suite for Multilingual Text Generation

TL;DR

MTG is a comprehensive multilingual text generation benchmark suite with extensive human-annotated data across five languages and four generation tasks, enabling evaluation of knowledge transfer and model performance.

Contribution

Introduces the first multilingual multiway text generation dataset with 400k annotations, facilitating diverse evaluation and analysis of multilingual models.

Findings

Models benefit from increased human-annotated data.

Multiway setup tests cross-language knowledge transfer.

Benchmark enables comprehensive evaluation of multilingual generation.

Abstract

We introduce MTG, a new benchmark suite for training and evaluating multilingual text generation. It is the first-proposed multilingual multiway text generation dataset with the largest human-annotated data (400k). It includes four generation tasks (story generation, question generation, title generation and text summarization) across five languages (English, German, French, Spanish and Chinese). The multiway setup enables testing knowledge transfer capabilities for a model across languages and tasks. Using MTG, we train and analyze several popular multilingual generation models from different aspects. Our benchmark suite fosters model performance enhancement with more human-annotated parallel data. It provides comprehensive evaluations with diverse generation scenarios. Code and data are available at \url{https://github.com/zide05/MTG}.