نقدم طريقتان رواية غير منشأة لإزالة السمية في النص.تجمع أهميتنا الأولى بين الأفكار الحديثة: (1) إرشادات عملية التوليد مع نماذج اللغة الشرطية النمطية الصغيرة و (2) استخدام نماذج إعادة الصياغة لأداء نقل النمط.نحن نستخدم أداء أداء جيدا تسترشد نماذج لغة مدربة على الطراز للحفاظ على محتوى النص وإزالة السمية.تستخدم الطريقة الثانية لدينا بيرت لاستبدال الكلمات السامة مع مرادفاتها غير الهجومية.نحن نجعل الطريقة أكثر مرونة من خلال تمكين بيرت لتحل محل الرموز القناع مع عدد متغير من الكلمات.أخيرا، نقدم أول دراسة مقارنة واسعة النطاق لنماذج نقل النمط في مهمة إزالة السمية.نقارن نماذجنا بعدد من الطرق لنقل النمط.يتم تقييم النماذج بطريقة خالية من المرجع باستخدام مزيج من مقاييس نقل النمط غير المدقق.كلتا الطريقتين نقترح أن تسفر عن نتائج سوتا الجديدة.
We present two novel unsupervised methods for eliminating toxicity in text. Our first method combines two recent ideas: (1) guidance of the generation process with small style-conditional language models and (2) use of paraphrasing models to perform style transfer. We use a well-performing paraphraser guided by style-trained language models to keep the text content and remove toxicity. Our second method uses BERT to replace toxic words with their non-offensive synonyms. We make the method more flexible by enabling BERT to replace mask tokens with a variable number of words. Finally, we present the first large-scale comparative study of style transfer models on the task of toxicity removal. We compare our models with a number of methods for style transfer. The models are evaluated in a reference-free way using a combination of unsupervised style transfer metrics. Both methods we suggest yield new SOTA results.
References used
https://aclanthology.org/
Recent developments in natural language generation (NLG) have bolstered arguments in favor of re-introducing explicit coding of discourse relations in the input to neural models. In the Methodius corpus, a meaning representation (MR) is hierarchicall
Commonsense reasoning benchmarks have been largely solved by fine-tuning language models. The downside is that fine-tuning may cause models to overfit to task-specific data and thereby forget their knowledge gained during pre-training. Recent works o
Recently, fine-tuning pre-trained language models (e.g., multilingual BERT) to downstream cross-lingual tasks has shown promising results. However, the fine-tuning process inevitably changes the parameters of the pre-trained model and weakens its cro
Modern transformer-based language models are revolutionizing NLP. However, existing studies into language modelling with BERT have been mostly limited to English-language material and do not pay enough attention to the implicit knowledge of language,
Large language models benefit from training with a large amount of unlabeled text, which gives them increasingly fluent and diverse generation capabilities. However, using these models for text generation that takes into account target attributes, su