Measuring and Improving Consistency in Pretrained Language Models

Elazar, Yanai; Kassner, Nora; Ravfogel, Shauli; Ravichander, Abhilasha; Hovy, Eduard; Schütze, Hinrich; Goldberg, Yoav

doi:10.48550/arxiv.2102.01017

Cited by 14 publications

(9 citation statements)

References 64 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Instruction-based learning has also been used in few-shot settings; popular variants include in-context learning, where the model's parameters are fixed and examples are provided as additional context (Brown et al, 2020;Lu et al, 2021;Kumar and Talukdar, 2021;Min et al, 2021), finetuning the entire model (Schick and Schütze, 2021a,c;Gao et al, 2021;Tam et al, 2021), and prompt tuning, where only the instruction itself is optimized (Shin et al, 2020;Hambardzumyan et al, 2021;Li and Liang, 2021;. Several works investigating the limitations and drawbacks of instruction-based few-shot approaches find that current LMs are mostly unable to understand complex instructions that go beyond short prompts or simple questions (Efrat and Levy, 2020;Weller et al, 2020;Webson and Pavlick, 2021) and that they are highly sensitive to the exact wording of the instructions provided (Jiang et al, 2020;Schick and Schütze, 2021a;Elazar et al, 2021). In a similar vein, Perez et al (2021) andLogan IV et al (2021) argue that prior work overestimates few-shot performance as manual prompt tuning is required to achieve good performance.…”

Section: Related Workmentioning

confidence: 99%

True Few-Shot Learning with Prompts -- A Real-World Perspective

Schick¹,

Schütze²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

Prompt-based approaches are strong at fewshot learning. However, Perez et al. (2021) have recently cast doubt on their performance because they had difficulty getting good results in a "true" few-shot setting in which prompts and hyperparameters cannot be tuned on a dev set. In view of this, we conduct an extensive study of PET, a method that combines textual instructions with examplebased finetuning. We show that, if correctly configured, PET performs strongly in a true few-shot setting, i.e., without a dev set. Crucial for this strong performance is PET's ability to intelligently handle multiple prompts. We then put our findings to a real-world test by running PET on RAFT, a benchmark of tasks taken directly from realistic NLP applications for which no labeled dev or test sets are available. PET achieves a new state of the art on RAFT and performs close to nonexpert humans for 7 out of 11 tasks. These results demonstrate that prompt-based learners like PET excel at true few-shot learning and underpin our belief that learning from instructions will play an important role on the path towards human-like few-shot learning capabilities.

show abstract

Section: Related Workmentioning

confidence: 99%

True Few-Shot Learning with Prompts -- A Real-World Perspective

Schick¹,

Schütze²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…Subsequent work argues that this approach of using a single natural language prompt per relation underestimates the ability of language models to predict factual information given language models are very sensitive to the input prompts. To this end, Jiang et al (2020) and Elazar et al (2021) generate paraphrases of prompts in the LAMA dataset and ensemble them to get a score for each relation. Other methods argue that natural language prompts might not be the best, and instead optimize a discrete (Shin et al, 2020;Haviv et al, 2021) or continuous (Qin & Eisner, 2021;Zhong et al, 2021;Liu et al, 2021b) prompt for each relation.…”

Section: Related Workmentioning

confidence: 99%

“…We focus on the issue of consistency. Elazar et al (2021) thoroughly investigated this issue, finding models often predict different entities for semantically equivalent prompts. Further, Elazar et al (2021) propose to continue training LLMs with a consistency loss function to improve their robustness.…”

Section: Related Workmentioning

confidence: 99%

“…Elazar et al (2021) thoroughly investigated this issue, finding models often predict different entities for semantically equivalent prompts. Further, Elazar et al (2021) propose to continue training LLMs with a consistency loss function to improve their robustness. In this work, however, we focus on a user-centric and lightweight approach that does not require updating LLM parameters.…”

Section: Related Workmentioning

confidence: 99%

“…In order for LLMs to be effective knowledge bases, they have to be robust to different queries users could provide. Unfortunately, prior work has shown that LLMs are not robust: the queries they are prompted with greatly impact the model's predictions, and semantically equivalent prompts may lead to inconsistent predictions (Jiang et al, 2020;Elazar et al, 2021). For example, taking the prompts above, the first prompt does extract the correct answer "Ottawa" from BERT Base, but the second extracts the incorrect "Winnipeg".…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

P-Adapters: Robustly Extracting Factual Information from Language Models with Diverse Prompts

Newman¹,

Choubey²,

Rajani³

2021

Preprint

View full text Add to dashboard Cite

Recent work (e.g. LAMA (Petroni et al., 2019)) has found that the quality of the factual information extracted from Large Language Models (LLMs) depends on the prompts used to query them. This inconsistency is problematic because different users will query LLMs for the same information using different wording, but should receive the same, accurate responses regardless. In this work we aim to address this shortcoming by introducing P-Adapters: lightweight models that sit between the embedding layer and first attention layer of LLMs. They take LLM embeddings as input and output continuous prompts that are used to query the LLM. Additionally, we investigate Mixture of Experts (MoE) models that learn a set of continuous prompts ("experts") and select one to query the LLM. They require a separate classifier trained on human-annotated data to map natural language prompts to the continuous ones. P-Adapters perform comparably to the more complex MoE models in extracting factual information from BERT and RoBERTa while eliminating the need for additional annotations. P-Adapters show between 12-26% absolute improvement in precision and 36-50% absolute improvement in consistency over a baseline of only using natural language queries. Finally, we investigate what makes a P-Adapter successful and conclude that access to the LLM's embeddings of the original natural language prompt, particularly the subject of the entity pair being asked about, is a significant factor.

show abstract

RuleBERT: Teaching Soft Rules to Pre-Trained Language Models

Saeed¹,

Ahmadi²,

Nakov³

et al. 2021

Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing

View full text Add to dashboard Cite

While pre-trained language models (PLMs) are the go-to solution to tackle many natural language processing problems, they are still very limited in their ability to capture and to use common-sense knowledge. In fact, even if information is available in the form of approximate (soft) logical rules, it is not clear how to transfer it to a PLM in order to improve its performance for deductive reasoning tasks. Here, we aim to bridge this gap by teaching PLMs how to reason with soft Horn rules. We introduce a classification task where, given facts and soft rules, the PLM should return a prediction with a probability for a given hypothesis. We release the first dataset for this task, and we propose a revised loss function that enables the PLM to learn how to predict precise probabilities for the task. Our evaluation results show that the resulting fine-tuned models achieve very high performance, even on logical rules that were unseen at training. Moreover, we demonstrate that logical notions expressed by the rules are transferred to the finetuned model, yielding state-of-the-art results on external datasets.

show abstract

Measuring and Improving Consistency in Pretrained Language Models

Cited by 14 publications

References 64 publications

True Few-Shot Learning with Prompts -- A Real-World Perspective

True Few-Shot Learning with Prompts -- A Real-World Perspective

P-Adapters: Robustly Extracting Factual Information from Language Models with Diverse Prompts

RuleBERT: Teaching Soft Rules to Pre-Trained Language Models

Contact Info

Product

Resources

About