The Adapter-Bot: All-In-One Controllable Conversational Model

In this paper, we present the Adapter-Bot, a generative chat-bot that uses fixed backbone conversational model such as DialGPT (Zhang et al. 2019) and triggers on-demand dialogue skills via different adapters (Houlsby 2019). Each adapter can be trained independently, thus allowing continual integration of without retraining entire model. Depending on skills, is able to process multiple knowledge types, text, tables, graphs, in seamless manner. The triggered automatically manager, or manually, high-level control generated responses. At current stage, have implemented 12 response styles (e.g., positive, negative etc.), 6 goal-oriented (e.g. weather information, movie recommendation, personalized emphatic