DExperts: Decoding-Time Controlled Text Generation with Experts and Anti-Experts
This work addresses the challenge of attribute control in text generation for natural language processing applications, offering an efficient decoding-time approach.
The paper tackled the problem of controlling attributes in text generation by proposing DExperts, a decoding-time method that combines pretrained language models with expert and anti-expert models, achieving improved performance in language detoxification and sentiment-controlled generation over existing methods.
Despite recent advances in natural language generation, it remains challenging to control attributes of generated text. We propose DExperts: Decoding-time Experts, a decoding-time method for controlled text generation that combines a pretrained language model with "expert" LMs and/or "anti-expert" LMs in a product of experts. Intuitively, under the ensemble, tokens only get high probability if they are considered likely by the experts, and unlikely by the anti-experts. We apply DExperts to language detoxification and sentiment-controlled generation, where we outperform existing controllable generation methods on both automatic and human evaluations. Moreover, because DExperts operates only on the output of the pretrained LM, it is effective with (anti-)experts of smaller size, including when operating on GPT-3. Our work highlights the promise of tuning small LMs on text with (un)desirable attributes for efficient decoding-time steering.