The aim of this workshop is to bring together experts working on open-domain dialogue research. In this speedily advancing research area many challenges still exist, such as learning information from conversations, engaging in realistic and convincing simulation of human intelligence and reasoning. SCI-CHAT follows previous workshops on open domain dialogue but with a focus on the simulation of intelligent conversation as judged in a live human evaluation. Models aim to include the ability to follow a challenging topic over a multi-turn conversation, while positing, refuting and reasoning over arguments. The workshop included both a research track and shared task. The main goal of this paper is to provide an overview of the shared task and a link to an additional paper that will include an in depth analysis of the shared task results following presentation at the workshop.

State-of-the-art conversational AI systems raise concerns due to their potential risks of generating unsafe, toxic, unethical, or dangerous content. Previous works have developed datasets to teach conversational agents the appropriate social paradigms to respond effectively to specifically designed hazardous content. However, models trained on these adversarial datasets still struggle to recognize subtle unsafe situations that appear naturally in conversations or introduce an inappropriate response in a casual context. To understand the extent of this problem, we study prosociality in both adversarial and casual dialog contexts and audit the response quality of general-purpose language models in terms of propensity to produce unsafe content. We propose a dual-step fine-tuning process to address these issues using a socially aware n-pair contrastive loss. Subsequently, we train a base model that integrates prosocial behavior by leveraging datasets like Moral Integrity Corpus (MIC) and ProsocialDialog. Experimental results on several dialog datasets demonstrate the effectiveness of our approach in generating socially appropriate responses.

In the realm of dialogue systems, user simulation techniques have emerged as a game-changer, redefining the evaluation and enhancement of task-oriented dialogue (TOD) systems. These methods are crucial for replicating real user interactions, enabling applications like synthetic data augmentation, error detection, and robust evaluation. However, existing approaches often rely on rigid rule-based methods or on annotated data. This paper introduces DAUS, a Domain-Aware User Simulator. Leveraging large language models, we fine-tune DAUS on real examples of task-oriented dialogues. Results on two relevant benchmarks showcase significant improvements in terms of user goal fulfillment. Notably, we have observed that fine-tuning enhances the simulator’s coherence with user goals, effectively mitigating hallucinations—a major source of inconsistencies in simulator responses.

This paper introduces a novel approach to form-filling and dialogue system evaluation by leveraging Large Language Models (LLMs). The proposed method establishes a setup wherein multiple modules collaborate on addressing the form-filling task. The dialogue system is constructed on top of LLMs, focusing on defining specific roles for individual modules. We show that using multiple independent sub-modules working cooperatively on this task can improve performance and handle the typical constraints of using LLMs, such as context limitations. The study involves testing the modular setup on four selected forms of varying topics and lengths, employing commercial and open-access LLMs. The experimental results demonstrate that the modular setup consistently outperforms the baseline, showcasing the effectiveness of this approach. Furthermore, our findings reveal that open-access models perform comparably to commercial models for the specified task.

An effective multi-turn instruction-following assistant can be developed by creating a simulator that can generate useful interaction data. Apart from relying on its intrinsic weights, an ideal user simulator should also be able to bootstrap external knowledge rapidly in its raw form to simulate the multifarious diversity of text available over the internet. Previous user simulators generally lacked diversity, were mostly closed domain, and necessitated rigid schema making them inefficient to rapidly scale to incorporate external knowledge. In this regard, we introduce Kaucus, a Knowledge-Augmented User Simulator framework, to outline a process of creating diverse user simulators, that can seamlessly exploit external knowledge as well as benefit downstream assistant model training. Through two GPT-J based simulators viz., a Retrieval Augmented Simulator and a Summary Controlled Simulator we generate diverse simulator-assistant interactions. Through reward and preference model-based evaluations, we find that these interactions serve as useful training data and create more helpful downstream assistants. We also find that incorporating knowledge through retrieval augmentation or summary control helps create better assistants.

Conversational models often face challenges such as a lack of emotional temperament and a limited sense of humor when interacting with users. To address these issues, we have selected relevant data and fine-tuned the model to (i) humanize the chatbot based on the user’s emotional response and the context of the conversation using a dataset based on empathy and (ii) enhanced conversations while incorporating humor/sarcasm for better user engagement. We aspire to achieve more personalized and enhanced user-computer interactions with the help of varied datasets involving sarcasm together with empathy on top of already available state-of-the-art conversational systems.

This paper is the model description for the Emo-Gen BART dialogue generation architecture, as submitted to the SCI-CHAT 2024 Shared Task. The Emotion-Informed Dialogue Generation model is a multi-task BARTbased model which performs dimensional and categorical emotion detection and uses that information to augment the input to the generation models. Our implementation is trained and validated against the IEMOCAP dataset, and compared against contemporary architectures in both dialogue emotion classification and dialogue generation. We show that certain loss function ablations are competitive against the state-of-the-art single-task models.

This system paper describes our conversational AI agent developed for the SCI-CHAT competition. The goal is to build automated dialogue agents that can have natural, coherent conversations with humans over multiple turns. Our model is based on fine-tuning the Snorkel-Mistral-PairRM-DPO language model on podcast conversation transcripts. This allows the model to leverage Snorkel-Mistral-PairRMDPO’s linguistic knowledge while adapting it for multi-turn dialogue modeling using LoRA. During evaluation, human judges will converse with the agent on specified topics and provide ratings on response quality. Our system aims to demonstrate how large pretrained language models, when properly adapted and evaluated, can effectively converse on open-ended topics spanning multiple turns.