What Makes My Model Perplexed? A Linguistic Investigation on Neural Language Models Perplexity
Alessio Miaschi, Dominique Brunato, Felice Dell’Orletta, Giulia Venturi
Abstract
This paper presents an investigation aimed at studying how the linguistic structure of a sentence affects the perplexity of two of the most popular Neural Language Models (NLMs), BERT and GPT-2. We first compare the sentence-level likelihood computed with BERT and the GPT-2’s perplexity showing that the two metrics are correlated. In addition, we exploit linguistic features capturing a wide set of morpho-syntactic and syntactic phenomena showing how they contribute to predict the perplexity of the two NLMs.- Anthology ID:
- 2021.deelio-1.5
- Volume:
- Proceedings of Deep Learning Inside Out (DeeLIO): The 2nd Workshop on Knowledge Extraction and Integration for Deep Learning Architectures
- Month:
- June
- Year:
- 2021
- Address:
- Online
- Venue:
- DeeLIO
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 40–47
- Language:
- URL:
- https://aclanthology.org/2021.deelio-1.5
- DOI:
- 10.18653/v1/2021.deelio-1.5
- Cite (ACL):
- Alessio Miaschi, Dominique Brunato, Felice Dell’Orletta, and Giulia Venturi. 2021. What Makes My Model Perplexed? A Linguistic Investigation on Neural Language Models Perplexity. In Proceedings of Deep Learning Inside Out (DeeLIO): The 2nd Workshop on Knowledge Extraction and Integration for Deep Learning Architectures, pages 40–47, Online. Association for Computational Linguistics.
- Cite (Informal):
- What Makes My Model Perplexed? A Linguistic Investigation on Neural Language Models Perplexity (Miaschi et al., DeeLIO 2021)
- PDF:
- https://preview.aclanthology.org/nodalida-main-page/2021.deelio-1.5.pdf
- Data
- Universal Dependencies