Multiple-Attribute Text Rewriting

By Guillaume Lample, Sandeep Subramanian, Eric Michael Smith, Ludovic Denoyer, Marc'Aurelio Ranzato, Y-Lan Boureau. 2018

245 citations recorded by the source corpus.

Subfields: Natural Language Processing

How researchers see this paper

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

In Natural Language Processing

Citations in the corpus

Open-Domain Conversational Agents: Current Progress, Open Problems, and Future Directions (2020, methodology): “Promising methods for controlled text generation (See et al., 2019) and text rewriting (Lample et al., 2019; Smith et al., 2019) could be refined to provide models more personally tailored to individual preferences, but are still not mature enough for that application.”
Plug and Play Language Models: A Simple Approach to Controlled Text Generation (2019, background): “We then ask annotators to rank the pair on the desired attribute (e.g. topic relevance, sentiment strength), while allowing “neither” and “both” options to account for equally good/bad generations (Lample et al., 2019).”
Controlling Style in Generated Dialogue (2020, methodology): “based on added noise and back-translation (Lample et al., 2019; Smith et al., 2019).”
Demoting Racial Bias in Hate Speech Detection (2020, background): “…for successfully adapting models to learn representations that are invariant to undesired attributes, such as demo-graphics and topics, though they rarely disentangle attributes completely (Li et al., 2018; Elazar and Goldberg, 2018; Kumar et al., 2019; Lample et al., 2019; Landeiro et al., 2019).”
PowerTransformer: Unsupervised Controllable Revision for Biased Language Correction (2020, background): “…to disentangle style from content and steer generations (e.g., Shen et al., 2017; Zhang et al., 2018; Fu et al., 2018; Yang et al., 2018; Niu and Bansal, 2018; Romanov et al., 2019; Dai et al., 2019; John et al., 2019) or backtranslation setups (Prabhumoye et al., 2018; Lample et al., 2018).”
Prompt-and-Rerank: A Method for Zero-Shot and Few-Shot Arbitrary Textual Style Transfer with Small Language Models (2022, background): “Some recent studies (Lample et al., 2019; Dai et al., 22For the AMAZON and YELP datasets, we show the benefit of data-cleaning on overall performance.”
Drinking From a Firehose: Continual Learning With Web-Scale Natural Language (2020, background): “small set of control factors such as sentiments [54], [55], sentence tense and length [56], [57], or topics and domains [58], [59].”
What happens if you treat ordinal ratings as interval data? Human evaluations in NLP are even more under-powered than you think (2021, methodology): “Collected by Dathathri et al. (2020), this data includes 5-point rating scale judgements assessing ﬂuency ranging from 1 = “not ﬂuent at all” to 5 = “very ﬂuent” as in (Lample et al., 2019).”
Linguistic calibration through metacognition: aligning dialogue agent responses with expected correctness (2020, methodology): “Lample et al. (2018) proposed a method to exert simultaneous control over multiple attributes based on concatenated learned control tokens.”
Mimic and Rephrase: Reflective Listening in Open-Ended Dialogue (2019, background): “Other work in this area includes generating text conditioned on a sentiment to express (Li et al., 2018), or controlled generation (Hu et al., 2017) by editing attributes (Shen et al., 2017; Logeswaran et al., 2018; Lample et al., 2019).”
LEWIS: Levenshtein Editing for Unsupervised Text Style Transfer (2021, methodology): “…al. (2017); John et al. (2019) or (2) using unsupervised generative approaches from retrieval (Li et al., 2018), tagging using style phrases (Madaan et al., 2020), to back-translation and unsupervised machine translation techniques (Prabhumoye et al., 2018; Lample et al., 2019; He et al., 2020b).”
Zero-Shot Fine-Grained Style Transfer: Leveraging Distributed Continuous Style Representations to Transfer To Unseen Styles (2019, methodology): “The style attribute is similarly speciﬁed here by providing an initial token to the model to specify the target attribute, but the generated text is also conditioned much more strongly on a source sentence, as was done in Lample et al. (2019).”
Style Transformer: Unpaired Text Style Transfer without Disentangled Latent Representation (2019, methodology): “As reported in (Elazar and Goldberg, 2018; Lample et al., 2019), the style information can be still recovered from the latent representation even the model has trained adversarially.”
Gradient-based Constrained Sampling from Language Models (2022, background): “∇ E + 2 η ( t − 1) βz ( t 1) ) ; Update et al., 2019) or fine-tuning (Gururangan et al., 2020; Chan et al., 2021) for prompt based generation and generative models for tasks such for style transfer (Lample et al., 2019; Ziegler et al., 2020; Prabhumoye et al., 2018; Yu et al., 2017).”
Few-shot Sequence Learning with Transformers (2020, background): “Previous works have shown that the behavior of a transformer model can be conditioned by appending special tokens describing the task to be performed on the input sequence (Lample et al., 2019; Zellers et al., 2019).”
Topics to Avoid: Demoting Latent Confounds in Text Classification (2019, background): “hx still ends up leaking information about the confound (Lample et al., 2019; Elazar and Goldberg, 2018).”
On Variational Learning of Controllable Representations for Text without Supervision (2019, methodology): “…controllable text generation, previous methods either assume annotated attributes or multiple text datasets with different known styles (Hu et al., 2017; Shen et al., 2017; Zhao et al., 2018; Fu et al., 2018; Li et al., 2018; Sudhakar et al., 2019; Logeswaran et al., 2018; Lample et al., 2018).”
Detect and Perturb: Neutral Rewriting of Biased and Sensitive Text via Gradient-based Decoding (2021, unknown): “, 2021a,b), backtranslation (Lample et al., 2018; Prabhumoye et al., 2018; Xu et al., 2019), pretraining (Duan et al.”
Civil Rephrases Of Toxic Texts With Self-Supervised Transformers (2021, methodology): “Lample et al. (2019) applied UNMT techniques from Conneau and Lample (2019) to several attribute transfer tasks, including social media datasets.”