2021-01-01 · conference-paper · Topic Modeling · en
highly citedfrom the source record
Some NLP tasks can be solved in a fully unsupervised fashion by providing a pretrained language model with "task descriptions" in natural language (e.g., Radford et al., 2019).While this approach underperforms its supervised counterpart, we show in this work that the two ideas can be combined: We introduce Pattern-Exploiting Training (PET), a semi-supervised training procedure that reformulates input examples as cloze-style phrases to help language models understand a given task.These phrases are then used to assign soft labels to a large set of unlabeled examples.Finally, standard supervised training is performed on the resulting training set.For several tasks and languages, PET outperforms supervised training and strong semi-supervised approaches in lowresource settings by a large margin. 1
Claims drawn from cited facts, not live model generation.
This paper is a conference paper centered on topic modeling. That conference-paper status means this work has passed peer review and should be weighed as published research rather than a preprint.
4 cited facts
This work was produced by a small team of two researchers.
1 cited fact
With 1302 citations, this work is already an established reference point in its field. It draws on 47 referenced works.
3 cited facts
Exploiting Cloze-Questions for Few-Shot Text Classification and Natural Language Inference is a conference-paper, focusing on Topic Modeling, published in 2021.
Exploiting Cloze-Questions for Few-Shot Text Classification and Natural Language Inference has 1,302 citations in our record, placing it among the highly cited works we track.
Our record lists 2 authors for Exploiting Cloze-Questions for Few-Shot Text Classification and Natural Language Inference and references 47 other works. Author names aren't in our data yet — the source link on the page lists the full byline.
Source facts, citations, and refresh stamp for this record.
Sources: openalex_works