Pitfalls in the Evaluation of Sentence Embeddings

Eger, Steffen; Rücklé, Andreas; Gurevych, Iryna

doi:10.18653/v1/w19-4308

Cited by 19 publications

(13 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The relation of probing to downstream tasks is also unclear, as our multilingual results show. This is supported by recent findings giving contradictory claims regarding, e.g., the importance of the Word Content probing task for downstream performances (Eger et al, 2019;Wang and Kuo, 2020;Perone et al, 2018). Our findings further add to contemporaneous work by Ravichander et al (2020) and Elazar et al (2020), who showed that probes do not necessarily identify linguistic properties required for solving an actual task, thus questioning a common interpretation of probing itself.…”

Section: Downstream Taskssupporting

confidence: 55%

How to Probe Sentence Embeddings in Low-Resource Languages: On Structural Design Choices for Probing Task Evaluation

Eger

Daxenberger

Gurevych

2020

Proceedings of the 24th Conference on Computational Natural Language Learning

Self Cite

View full text Add to dashboard Cite

Sentence encoders map sentences to real valued vectors for use in downstream applications. To peek into these representations-e.g., to increase interpretability of their resultsprobing tasks have been designed which query them for linguistic knowledge. However, designing probing tasks for lesser-resourced languages is tricky, because these often lack largescale annotated data or (high-quality) dependency parsers as a prerequisite of probing task design in English. To investigate how to probe sentence embeddings in such cases, we investigate sensitivity of probing task results to structural design choices, conducting the first such large scale study. We show that design choices like size of the annotated probing dataset and type of classifier used for evaluation do (sometimes substantially) influence probing outcomes. We then probe embeddings in a multilingual setup with design choices that lie in a 'stable region', as we identify for English, and find that results on English do not transfer to other languages. Fairer and more comprehensive sentence-level probing evaluation should thus be carried out on multiple languages in the future.

show abstract

Section: Downstream Taskssupporting

confidence: 55%

How to Probe Sentence Embeddings in Low-Resource Languages: On Structural Design Choices for Probing Task Evaluation

Eger

Daxenberger

Gurevych

2020

Proceedings of the 24th Conference on Computational Natural Language Learning

Self Cite

View full text Add to dashboard Cite

show abstract

“…It is worthwhile to emphasize that we use only 768dimension vectors for sentence embedding while InferSent uses 4096-dimension vectors. As explained in [14], [30], [52], the increase in the embedding dimension leads to increased performance for almost all models. This may explain SBERT-WK is slightly inferior to InferSent on the SICK-R dataset.…”

Section: A Semantic Textural Similaritymentioning

confidence: 68%

SBERT-WK: A Sentence Embedding Method by Dissecting BERT-based Word Models

Wang

Kuo

2020

Preprint

View full text Add to dashboard Cite

Sentence embedding is an important research topic in natural language processing (NLP) since it can transfer knowledge to downstream tasks. Meanwhile, a contextualized word representation, called BERT, achieves the state-of-the-art performance in quite a few NLP tasks. Yet, it is an open problem to generate a high quality sentence representation from BERTbased word models. It was shown in previous study that different layers of BERT capture different linguistic properties. This allows us to fusion information across layers to find better sentence representation. In this work, we study the layer-wise pattern of the word representation of deep contextualized models. Then, we propose a new sentence embedding method by dissecting BERT-based word models through geometric analysis of the space spanned by the word representation. It is called the SBERT-WK method. No further training is required in SBERT-WK. We evaluate SBERT-WK on semantic textual similarity and downstream supervised tasks. Furthermore, ten sentencelevel probing tasks are presented for detailed linguistic analysis. Experiments show that SBERT-WK achieves the state-of-the-art performance. Our codes are publicly available at the Github 1 .

show abstract

“…Hewitt and Liang (2019) pointed out that a simple linear probe is not enough to evaluate a representation. Recently, we have also seen non-linear probes Eger et al, 2019). There are also efforts to inspect the representations from a geometric persepctive (e.g.…”

Section: Related Workmentioning

confidence: 99%

A Closer Look at How Fine-tuning Changes BERT

Zhou¹,

Srikumar²

2021

Preprint

View full text Add to dashboard Cite

Given the prevalence of pre-trained contextualized representations in today's NLP, there have been several efforts to understand what information such representations contain. A common strategy to use such representations is to fine-tune them for an end task. However, how fine-tuning for a task changes the underlying space is less studied. In this work, we study the English BERT family and use two probing techniques to analyze how fine-tuning changes the space. Our experiments reveal that fine-tuning improves performance because it pushes points associated with a label away from other labels. By comparing the representations before and after fine-tuning, we also discover that fine-tuning does not change the representations arbitrarily; instead, it adjusts the representations to downstream tasks while preserving the original structure. Finally, using carefully constructed experiments, we show that fine-tuning can encode training sets in a representation, suggesting an overfitting problem of a new kind.

show abstract

Pitfalls in the Evaluation of Sentence Embeddings

Cited by 19 publications

References 16 publications

How to Probe Sentence Embeddings in Low-Resource Languages: On Structural Design Choices for Probing Task Evaluation

How to Probe Sentence Embeddings in Low-Resource Languages: On Structural Design Choices for Probing Task Evaluation

SBERT-WK: A Sentence Embedding Method by Dissecting BERT-based Word Models

A Closer Look at How Fine-tuning Changes BERT

Contact Info

Product

Resources

About