Generating intermediate steps, or Chain of Thought (CoT), is an effective way to significantly improve language models’ (LM) multi-step reasoning capability. However, the CoT lengths can grow rapidly with the problem complexity, easily exceeding the maximum context size. Instead of increasing the context limit, which has already been heavily investigated, we explore an orthogonal direction: making LMs divide a problem into multiple contexts. We propose a new inference framework, called Recursion of Thought (RoT), which introduces several special tokens that the models can output to trigger context-related operations. Extensive experiments with multiple architectures including GPT-3 show that RoT dramatically improves LMs’ inference capability to solve problems, whose solution consists of hundreds of thousands of tokens.

As short-form funny videos on social networks are gaining popularity, it becomes demanding for AI models to understand them for better communication with humans. Unfortunately, previous video humor datasets target specific domains such as speeches or sitcoms, and mostly focus on verbal cues. We curate a user-generated dataset of 10K multimodal funny videos from YouTube, called ExFunTube. Using a video filtering pipeline with GPT-3.5, we verify both verbal and visual elements contributing to humor. After filtering, we annotate each video with timestamps and text explanations for funny moments. Our ExFunTube is unique over existing datasets in that our videos cover a wide range of domains with various types of humor that necessitate a multimodal understanding of the content. Also, we develop a zero-shot video-to-text prompting to maximize video humor understanding of large language models (LLMs). With three different evaluation methods using automatic scores, rationale quality experiments, and human evaluations, we show that our prompting significantly improves LLMs’ ability for humor explanation.

The growing number of multimodal online discussions necessitates automatic summarization to save time and reduce content overload. However, existing summarization datasets are not suitable for this purpose, as they either do not cover discussions, multiple modalities, or both. To this end, we present mRedditSum, the first multimodal discussion summarization dataset. It consists of 3,033 discussion threads where a post solicits advice regarding an issue described with an image and text, and respective comments express diverse opinions. We annotate each thread with a human-written summary that captures both the essential information from the text, as well as the details available only in the image. Experiments show that popular summarization models—GPT-3.5, BART, and T5—consistently improve in performance when visual information is incorporated. We also introduce a novel method, cluster-based multi-stage summarization, that outperforms existing baselines and serves as a competitive baseline for future work.

Data scarcity has been a long standing issue in the field of open-domain social dialogue. To quench this thirst, we present SODA: the first publicly available, million-scale high-quality social dialogue dataset. By contextualizing social commonsense knowledge from a knowledge graph, we are able to distill an exceptionally broad spectrum of social interactions from a large language model. Human evaluation shows that conversations in SODA are more consistent, specific, and (surprisingly) natural than those in prior human-authored datasets. Using SODA, we train COSMO: a generalizable conversation model that is significantly more natural and consistent on unseen datasets than best-performing conversation models (e.g., GODEL, BlenderBot-1, Koala, Vicuna). Experiments reveal COSMO is sometimes even preferred to the original human-written gold responses. Additionally, our results shed light on the distinction between knowledge-enriched conversations and natural social chitchats. We plan to make our data, model, and code public.

Theory of mind (ToM) evaluations currently focus on testing models using passive narratives that inherently lack interactivity. We introduce FANToM, a new benchmark designed to stress-test ToM within information-asymmetric conversational contexts via question answering. Our benchmark draws upon important theoretical requisites from psychology and necessary empirical considerations when evaluating large language models (LLMs). In particular, we formulate multiple types of questions that demand the same underlying reasoning to identify illusory or false sense of ToM capabilities in LLMs. We show that FANToM is challenging for state-of-the-art LLMs, which perform significantly worse than humans even with chain-of-thought reasoning or fine-tuning.

In order to build self-consistent personalized dialogue agents, previous research has mostly focused on textual persona that delivers personal facts or personalities. However, to fully describe the multi-faceted nature of persona, image modality can help better reveal the speaker’s personal characteristics and experiences in episodic memory (Rubin et al., 2003; Conway, 2009). In this work, we extend persona-based dialogue to the multimodal domain and make two main contributions. First, we present the first multimodal persona-based dialogue dataset named MPCHAT, which extends persona with both text and images to contain episodic memories. Second, we empirically show that incorporating multimodal persona, as measured by three proposed multimodal persona-grounded dialogue tasks (i.e., next response prediction, grounding persona prediction, and speaker identification), leads to statistically significant performance improvements across all tasks. Thus, our work highlights that multimodal persona is crucial for improving multimodal dialogue comprehension, and our MPCHAT serves as a high-quality resource for this research.

The potential social harms that large language models pose, such as generating offensive content and reinforcing biases, are steeply rising. Existing works focus on coping with this concern while interacting with ill-intentioned users, such as those who explicitly make hate speech or elicit harmful responses. However, discussions on sensitive issues can become toxic even if the users are well-intentioned. For safer models in such scenarios, we present the Sensitive Questions and Acceptable Response (SQuARe) dataset, a large-scale Korean dataset of 49k sensitive questions with 42k acceptable and 46k non-acceptable responses. The dataset was constructed leveraging HyperCLOVA in a human-in-the-loop manner based on real news headlines. Experiments show that acceptable response generation significantly improves for HyperCLOVA and GPT-3, demonstrating the efficacy of this dataset.

Large language models (LLMs) not only learn natural text generation abilities but also social biases against different demographic groups from real-world data. This poses a critical risk when deploying LLM-based applications. Existing research and resources are not readily applicable in South Korea due to the differences in language and culture, both of which significantly affect the biases and targeted demographic groups. This limitation requires localized social bias datasets to ensure the safe and effective deployment of LLMs. To this end, we present KosBi, a new social bias dataset of 34k pairs of contexts and sentences in Korean covering 72 demographic groups in 15 categories. We find that through filtering-based moderation, social biases in generated content can be reduced by 16.47%p on average for HyperClova (30B and 82B), and GPT-3.

Most existing dialogue systems fail to respond properly to potentially unsafe user utterances by either ignoring or passively agreeing with them. To address this issue, we introduce ProsocialDialog, the first large-scale multi-turn dialogue dataset to teach conversational agents to respond to problematic content following social norms. Covering diverse unethical, problematic, biased, and toxic situations, ProsocialDialog contains responses that encourage prosocial behavior, grounded in commonsense social rules (i.e., rules-of-thumb, RoTs). Created via a human-AI collaborative framework, ProsocialDialog consists of 58K dialogues, with 331K utterances, 160K unique RoTs, and 497K dialogue safety labels accompanied by free-form rationales.With this dataset, we introduce a dialogue safety detection module, Canary, capable of generating RoTs given conversational context, and a socially-informed dialogue agent, Prost. Empirical results show that Prost generates more socially acceptable dialogues compared to other state-of-the-art language and dialogue models in both in-domain and out-of-domain settings. Additionally, Canary effectively guides conversational agents and off-the-shelf language models to generate significantly more prosocial responses. Our work highlights the promise and importance of creating and steering conversational AI to be socially responsible.

Knowledge is now starting to power neural dialogue agents. At the same time, the risk of misinformation and disinformation from dialogue agents also rises. Verifying the veracity of information from formal sources are widely studied in computational fact checking. In this work, we ask: How robust are fact checking systems on claims in colloquial style? We aim to open up new discussions in the intersection of fact verification and dialogue safety. In order to investigate how fact checking systems behave on colloquial claims, we transfer the styles of claims from FEVER (Thorne et al., 2018) into colloquialism. We find that existing fact checking systems that perform well on claims in formal style significantly degenerate on colloquial claims with the same semantics. Especially, we show that document retrieval is the weakest spot in the system even vulnerable to filler words, such as “yeah” and “you know”. The document recall of WikiAPI retriever (Hanselowski et al., 2018) which is 90.0% on FEVER, drops to 72.2% on the colloquial claims. We compare the characteristics of colloquial claims to those of claims in formal style, and demonstrate the challenging issues in them.

Empathy is a complex cognitive ability based on the reasoning of others’ affective states. In order to better understand others and express stronger empathy in dialogues, we argue that two issues must be tackled at the same time: (i) identifying which word is the cause for the other’s emotion from his or her utterance and (ii) reflecting those specific words in the response generation. However, previous approaches for recognizing emotion cause words in text require sub-utterance level annotations, which can be demanding. Taking inspiration from social cognition, we leverage a generative estimator to infer emotion cause words from utterances with no word-level label. Also, we introduce a novel method based on pragmatics to make dialogue models focus on targeted words in the input during generation. Our method is applicable to any dialogue models with no additional training on the fly. We show our approach improves multiple best-performing dialogue agents on generating more focused empathetic responses in terms of both automatic and human evaluation.

We present a novel data augmentation technique, CRA (Contextual Response Augmentation), which utilizes conversational context to generate meaningful samples for training. We also mitigate the issues regarding unbalanced context lengths by changing the input output format of the model such that it can deal with varying context lengths effectively. Specifically, our proposed model, trained with the proposed data augmentation technique, participated in the sarcasm detection task of FigLang2020, have won and achieves the best performance in both Reddit and Twitter datasets.

We explore the task of improving persona consistency of dialogue agents. Recent models tackling consistency often train with additional Natural Language Inference (NLI) labels or attach trained extra modules to the generative agent for maintaining consistency. However, such additional labels and training can be demanding. Also, we find even the best-performing persona-based agents are insensitive to contradictory words. Inspired by social cognition and pragmatics, we endow existing dialogue agents with public self-consciousness on the fly through an imaginary listener. Our approach, based on the Rational Speech Acts framework (Frank and Goodman, 2012), can enforce dialogue agents to refrain from uttering contradiction. We further extend the framework by learning the distractor selection, which has been usually done manually or randomly. Results on Dialogue NLI (Welleck et al., 2019) and PersonaChat (Zhang et al., 2018) dataset show that our approach reduces contradiction and improves consistency of existing dialogue models. Moreover, we show that it can be generalized to improve context-consistency beyond persona in dialogues.

We explore the problem of Audio Captioning: generating natural language description for any kind of audio in the wild, which has been surprisingly unexplored in previous research. We contribute a large-scale dataset of 46K audio clips with human-written text pairs collected via crowdsourcing on the AudioSet dataset. Our thorough empirical studies not only show that our collected captions are indeed faithful to audio inputs but also discover what forms of audio representation and captioning models are effective for the audio captioning. From extensive experiments, we also propose two novel components that help improve audio captioning performance: the top-down multi-scale encoder and aligned semantic attention.

We address the problem of abstractive summarization in two directions: proposing a novel dataset and a new model. First, we collect Reddit TIFU dataset, consisting of 120K posts from the online discussion forum Reddit. We use such informal crowd-generated posts as text source, in contrast with existing datasets that mostly use formal documents as source such as news articles. Thus, our dataset could less suffer from some biases that key sentences usually located at the beginning of the text and favorable summary candidates are already inside the text in similar forms. Second, we propose a novel abstractive summarization model named multi-level memory networks (MMN), equipped with multi-level memory to store the information of text from different levels of abstraction. With quantitative evaluation and user studies via Amazon Mechanical Turk, we show the Reddit TIFU dataset is highly abstractive and the MMN outperforms the state-of-the-art summarization models.

Variational autoencoders (VAE) combined with hierarchical RNNs have emerged as a powerful framework for conversation modeling. However, they suffer from the notorious degeneration problem, where the decoders learn to ignore latent variables and reduce to vanilla RNNs. We empirically show that this degeneracy occurs mostly due to two reasons. First, the expressive power of hierarchical RNN decoders is often high enough to model the data using only its decoding distributions without relying on the latent variables. Second, the conditional VAE structure whose generation process is conditioned on a context, makes the range of training targets very sparse; that is, the RNN decoders can easily overfit to the training data ignoring the latent variables. To solve the degeneration problem, we propose a novel model named Variational Hierarchical Conversation RNNs (VHCR), involving two key ideas of (1) using a hierarchical structure of latent variables, and (2) exploiting an utterance drop regularization. With evaluations on two datasets of Cornell Movie Dialog and Ubuntu Dialog Corpus, we show that our VHCR successfully utilizes latent variables and outperforms state-of-the-art models for conversation generation. Moreover, it can perform several new utterance control tasks, thanks to its hierarchical latent structure.