The GEM Benchmark: Natural Language Generation, its Evaluation and Metrics
This benchmark addresses the problem of inconsistent and anglo-centric evaluation for the natural language generation research community, providing an incremental step towards standardized and multilingual assessment.
The GEM benchmark was introduced to address the problem of divergent evaluation practices in Natural Language Generation (NLG). It provides a standardized environment for applying models to various tasks and testing evaluation strategies, aiming to make NLG research more multilingual and to evolve alongside new models.
We introduce GEM, a living benchmark for natural language Generation (NLG), its Evaluation, and Metrics. Measuring progress in NLG relies on a constantly evolving ecosystem of automated metrics, datasets, and human evaluation standards. Due to this moving target, new models often still evaluate on divergent anglo-centric corpora with well-established, but flawed, metrics. This disconnect makes it challenging to identify the limitations of current models and opportunities for progress. Addressing this limitation, GEM provides an environment in which models can easily be applied to a wide set of tasks and in which evaluation strategies can be tested. Regular updates to the benchmark will help NLG research become more multilingual and evolve the challenge alongside models. This paper serves as the description of the data for which we are organizing a shared task at our ACL 2021 Workshop and to which we invite the entire NLG community to participate.