Corpus ID: 215548699

BLEURT: Learning Robust Metrics for Text Generation

@article{Sellam2020BLEURTLR,
  title={BLEURT: Learning Robust Metrics for Text Generation},
  author={Thibault Sellam and Dipanjan Das and Ankur P. Parikh},
  journal={ArXiv},
  year={2020},
  volume={abs/2004.04696}
}
  • Thibault Sellam, Dipanjan Das, Ankur P. Parikh
  • Published 2020
  • Computer Science
  • ArXiv
  • Text generation has made significant advances in the last few years. Yet, evaluation metrics have lagged behind, as the most popular choices (e.g., BLEU and ROUGE) may correlate poorly with human judgments. We propose BLEURT, a learned evaluation metric based on BERT that can model human judgments with a few thousand possibly biased training examples. A key aspect of our approach is a novel pre-training scheme that uses millions of synthetic examples to help the model generalize. BLEURT… CONTINUE READING

    Figures, Tables, and Topics from this paper.

    Explore Further: Topics Discussed in This Paper

    Citations

    Publications citing this paper.
    SHOWING 1-2 OF 2 CITATIONS

    Evaluation of Text Generation: A Survey

    VIEW 6 EXCERPTS
    HIGHLY INFLUENCED

    ToTTo: A Controlled Table-To-Text Generation Dataset

    VIEW 2 EXCERPTS
    CITES BACKGROUND & METHODS

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 50 REFERENCES

    Attention is All you Need

    VIEW 4 EXCERPTS
    HIGHLY INFLUENTIAL

    ROUGE: A Package for Automatic Evaluation of Summaries

    VIEW 8 EXCERPTS
    HIGHLY INFLUENTIAL

    Results of the WMT17 Metrics Shared Task

    VIEW 7 EXCERPTS
    HIGHLY INFLUENTIAL

    Challenges in Data-to-Document Generation

    VIEW 4 EXCERPTS
    HIGHLY INFLUENTIAL