Recent work in open-domain conversational agents has demonstrated that significant improvements in humanness and user preference can be achieved via massive scaling in both pre-training data and model size (Adiwardana et al., 2020; Roller et al., 202
0). However, if we want to build agents with human-like abilities, we must expand beyond handling just text. A particularly important topic is the ability to see images and communicate about what is perceived. With the goal of getting humans to engage in multi-modal dialogue, we investigate combining components from state-of-the-art open-domain dialogue agents with those from state-of-the-art vision models. We study incorporating different image fusion schemes and domain-adaptive pre-training and fine-tuning strategies, and show that our best resulting model outperforms strong existing models in multi-modal dialogue while simultaneously performing as well as its predecessor (text-only) BlenderBot (Roller et al., 2020) in text-based conversation. We additionally investigate and incorporate safety components in our final model, and show that such efforts do not diminish model performance with respect to human preference.
Conversational agents trained on large unlabeled corpora of human interactions will learn patterns and mimic behaviors therein, which include offensive or otherwise toxic behavior. We introduce a new human-and-model-in-the-loop framework for evaluati
ng the toxicity of such models, and compare a variety of existing methods in both the cases of non-adversarial and adversarial users that expose their weaknesses. We then go on to propose two novel methods for safe conversational agents, by either training on data from our new human-and-model-in-the-loop framework in a two-stage system, or ''baking-in'' safety to the generative model itself. We find our new techniques are (i) safer than existing models; while (ii) maintaining usability metrics such as engagingness relative to state-of-the-art chatbots. In contrast, we expose serious safety issues in existing standard systems like GPT2, DialoGPT, and BlenderBot.
We seek to create agents that both act and communicate with other agents in pursuit of a goal. Towards this end, we extend LIGHT (Urbanek et al. 2019)---a large-scale crowd-sourced fantasy text-game---with a dataset of quests. These contain natural l
anguage motivations paired with in-game goals and human demonstrations; completing a quest might require dialogue or actions (or both). We introduce a reinforcement learning system that (1) incorporates large-scale language modeling-based and commonsense reasoning-based pre-training to imbue the agent with relevant priors; and (2) leverages a factorized action space of action commands and dialogue, balancing between the two. We conduct zero-shot evaluations using held-out human expert demonstrations, showing that our agents are able to act consistently and talk naturally with respect to their motivations.
How can we design Natural Language Processing (NLP) systems that learn from human feedback? There is a growing research body of Human-in-the-loop (HITL) NLP frameworks that continuously integrate human feedback to improve the model itself. HITL NLP r
esearch is nascent but multifarious---solving various NLP problems, collecting diverse feedback from different people, and applying different methods to learn from human feedback. We present a survey of HITL NLP work from both Machine Learning (ML) and Human-computer Interaction (HCI) communities that highlights its short yet inspiring history, and thoroughly summarize recent frameworks focusing on their tasks, goals, human interactions, and feedback learning methods. Finally, we discuss future studies for integrating human feedback in the NLP development loop.
In this paper we argue that embodied multimodal agents, i.e., avatars, can play an important role in moving natural language processing toward deep understanding.'' Fully-featured interactive agents, model encounters between two people,'' but a langu
age-only agent has little environmental and situational awareness. Multimodal agents bring new opportunities for interpreting visuals, locational information, gestures, etc., which are more axes along which to communicate. We propose that multimodal agents, by facilitating an embodied form of human-computer interaction, provide additional structure that can be used to train models that move NLP systems closer to genuine understanding'' of grounded language, and we discuss ongoing studies using existing systems.
This research traces, after conducting a wide literature survey, the areas not covered by prominent agent oriented software engineering (AOSE) methodologies. Each methodology has its strength and weakness and focuses on some stages of software
devel
opment lifecycle but not all stages. This paper presents an addition to a well established AOSE methodology (MaSE). MaSE is considered one of the strongest in the field, it does not, however, support handling early requirements. This work integrates MaSE with another methodology known for its strength in early requirement representation. The integration implied the development of a wide set of translation rules between two different environments of notations and graphical representations. A software tool was developed to automate the translation and a case study is used to demonstrate the work.
