Negated and Misprimed Probes for Pretrained Language Models: Birds Can Talk, But Cannot Fly

By Nora Kassner, Hinrich Schütze. 2019

382 citations recorded by the source corpus.

Building on Petroni et al. 2019, we propose two new probing tasks analyzing factual knowledge stored in Pretrained Language Models (PLMs). (1) Negation. We find that PLMs do not distinguish between negated (‘‘Birds cannot [MASK]”) and non-negated (‘‘Birds can [MASK]”) cloze questions. (2) Mispriming. Inspired by priming methods in human psychology, we add “misprimes” to cloze questions (‘‘Talk? Birds can [MASK]”). We find that PLMs are easily distracted by misprimes. These results suggest that PLMs still have a long way to go to adequately learn human-like factual knowledge.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Natural Language Processing, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Natural Language Processing

In Evaluation & Benchmarks

Citations in the corpus

Exploiting Cloze-Questions for Few-Shot Text Classification and Natural Language Inference (2020, background): “, 2020), linguistic capabilities (Ettinger, 2020; Kassner and Schütze, 2020), understanding of rare words (Schick and Schütze, 2020), and ability to perform symbolic reasoning (Talmor et al.”
RuleBERT: Teaching Soft Rules to Pre-Trained Language Models (2021, methodology): “We can see that RULEBERT performs better on both evaluation measures used in (Kassner and Schütze, 2020).”
Measuring and Improving Consistency in Pretrained Language Models (2021, background): “For instance, previous work has studied the effect of negation on factual statements, which can also be seen as consistency (Ettinger, 2020; Kassner and Schütze, 2020).”
Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLP (2021, background): “Unfortunately, negation is often not understood well by current generations of language models (Kassner and Schütze, 2020).”
Unsupervised Commonsense Question Answering with Self-Talk (2020, background): “(ii) insufficient precision: the distributional training objective increases the probability of non-facts that are semantically similar to true facts, as in negation (“birds cannot fly”; Kassner and Schütze, 2020).”
KEPLER: A Unified Model for Knowledge Embedding and Pre-trained Language Representation (2019, methodology): “This may be due to different vocabularies used in BERT and RoBERTa, which presents the vulnerability of LAMA-style probing again (Kassner and Schütze, 2020).”
Thinking Aloud: Dynamic Context Generation Improves Zero-Shot Reasoning Performance of GPT-2 (2021, methodology): “Language models quickly learn to master logical fragments given appropriate training data [Kassner and Schütze, 2020, Richardson and Sabharwal, 2019, Richardson et al., 2020], and can be ﬁne-tuned to correctly draw complex deductive inferences [Clark et al., 2020, Betz et al., 2020] and to generate…”
Taxonomy of Risks posed by Language Models (2022, background): “Kassner and Schütze [98] found that masked language models ELMo and BERT fail to distinguish such nuances.”
DirectProbe: Studying Representations without Classifiers (2021, methodology): “…commonly used strategy calls for training classiﬁers on them to predict linguistic properties such as syntax, or cognitive skills like numeracy (e.g. Kassner and Schütze, 2020; Perone et al., 2018; Yaghoobzadeh et al., 2019; Krasnowska-Kiera´s and Wróblewska, 2019; Wallace et al., 2019;…”
Are Pretrained Language Models Symbolic Reasoners over Knowledge? (2020, result): “This is in agreement with prior work (Ettinger, 2020; Kassner and Schütze, 2020) showing that BERT trained on natural language corpora is as”
Understanding by Understanding Not: Modeling Negation in Language Models (2021, methodology): “We test our model on the negated LAMA dataset (Kassner and Schütze, 2019), which is the negated version of knowledge probing dataset LAMA, introduced in Petroni et al.”