We present Any-Modality Augmented Language Model (AnyMAL), a unified model that reasons over diverse input modality signals (i.e. text, image, video, audio, IMU motion sensor), and generates textual responses. AnyMAL inherits the powerful text-based reasoning abilities of the state-of-the-art LLMs including Llama-3 (70B), and converts modality-specific signals to the joint textual space through a pre-trained aligner module.In this paper, we provide details on the optimizations implemented to efficiently scale the training pipeline, and present a comprehensive recipe for model and training configurations. We conduct comprehensive empirical analysis comprising both human and automatic evaluations, and demonstrate state-of-the-art performance on various multimodal tasks compared to industry-leading models – albeit with a relatively small number of trainable parameters.

Vision-extended LLMs have made significant strides in Visual Question Answering (VQA). Despite these advancements, VLLMs still encounter substantial difficulties in handling queries involving long-tail entities, with a tendency to produce erroneous or hallucinated responses. In this work, we introduce a novel evaluative benchmark named SnapNTell, specifically tailored for entity-centric VQA. This task aims to test the models’ capabilities in identifying entities and providing detailed, entity-specific knowledge. We have developed the SnapNTell Dataset, distinct from traditional VQA datasets: (1) It encompasses a wide range of categorized entities, each represented by images and explicitly named in the answers; (2) It features QA pairs that require extensive knowledge for accurate responses. The dataset is organized into 22 major categories, containing 7,568 unique entities in total. For each entity, we curated 10 illustrative images and crafted 10 knowledge-intensive QA pairs. To address this novel task, we devised a scalable, efficient, and transparent retrieval-augmented multimodal LLM. Our approach markedly outperforms existing methods on the SnapNTell dataset, achieving a 66.5% improvement in the BELURT score.

Large Language models (LLMs) have shown remarkable success in assisting robot learning tasks, i.e., complex household planning.However, the performance of pretrained LLMs heavily relies on domain-specific templated text data, which may be infeasible in real-world robot learning tasks with image-based observations. Moreover, existing LLMs with text inputs lack the capability to evolve with non-expert interactions with environments.In this work, we introduce a novel learning paradigm that generates robots’ executable actions in the form of text, derived solely from visual observations. Our proposed paradigm stands apart from previous works, which utilized either language instructions or a combination of language and visual data as inputs. We demonstrate that our proposed method can employ two fine-tuning strategies, including imitation learning and reinforcement learning approaches, to adapt to the target test tasks effectively.We conduct extensive experiments involving various model selections, environments, and tasks across 7 house layouts in the VirtualHome environment. Our experimental results demonstrate that our method surpasses existing baselines, confirming the effectiveness of this novel learning paradigm.

Large language models (LLMs) are increasingly prevalent in conversational systems due to their advanced understanding and generative capabilities in general contexts. However, their effectiveness in task-oriented dialogues (TOD), which requires not only response generation but also effective dialogue state tracking (DST) within specific tasks and domains, remains less satisfying. In this work, we propose a novel approach FnCTOD for solving DST with LLMs through function calling. This method improves zero-shot DST, allowing adaptation to diverse domains without extensive data collection or model tuning. Our experimental results demonstrate that our approach achieves exceptional performance with both modestly sized open-source and also proprietary LLMs: with in-context prompting it enables various 7B or 13B parameter models to surpass the previous state-of-the-art (SOTA) achieved by ChatGPT, and improves ChatGPT’s performance beating the SOTA by 5.6% average joint goal accuracy (JGA). Individual model results for GPT-3.5 and GPT-4 are boosted by 4.8% and 14%, respectively. We also show that by fine-tuning on a small collection of diverse task-oriented dialogues, we can equip modestly sized models, specifically a 13B parameter LLaMA2-Chat model, with function-calling capabilities and DST performance comparable to ChatGPT while maintaining their chat capabilities. We have made the code publicly available at https://github.com/facebookresearch/FnCTOD.

Building an AI assistant that can seamlessly converse and instruct humans, in a user-centric situated scenario, requires several essential abilities:(1) spatial and temporal understanding of the situated and real-time user scenes,(2) capability of grounding the actively perceived visuals of users to conversation contexts,and (3) conversational reasoning over past utterances to perform just-in-time assistance. However, we currently lack a large-scale benchmark that captures user–assistant interactions with all of the aforementioned features. To this end, we propose SIMMC-VR, an extension of the SIMMC-2.0 dataset, to a video-grounded task-oriented dialog dataset that captures real-world AI-assisted user scenarios in VR.We propose a novel data collection paradigm that involves(1) generating object-centric multimodal dialog flows with egocentric visual streams and visually-grounded templates,and (2) manually paraphrasing the simulated dialogs for naturalness and diversity while preserving multimodal dependencies. To measure meaningful progress in the field, we propose four tasks to address the new challenges in SIMMC-VR, which require complex spatial-temporal dialog reasoning in active egocentric scenes. We benchmark the proposed tasks with strong multimodal models, and highlight the key capabilities that current models lack for future research directions.

Searching troves of videos with textual descriptions is a core multimodal retrieval task. Owing to the lack of a purpose-built dataset for text-to-video retrieval, video captioning datasets have been re-purposed to evaluate models by (1) treating captions as positive matches to their respective videos and (2) assuming all other videos to be negatives. However, this methodology leads to a fundamental flaw during evaluation: since captions are marked as relevant only to their original video, many alternate videos also match the caption, which introduces false-negative caption-video pairs. We show that when these false negatives are corrected, a recent state-of-the-art model gains 25% recall points—a difference that threatens the validity of the benchmark itself. To diagnose and mitigate this issue, we annotate and release 683K additional caption-video pairs. Using these, we recompute effectiveness scores for three models on two standard benchmarks (MSR-VTT and MSVD). We find that (1) the recomputed metrics are up to 25% recall points higher for the best models, (2) these benchmarks are nearing saturation for Recall@10, (3) caption length (generality) is related to the number of positives, and (4) annotation costs can be mitigated through sampling. We recommend retiring these benchmarks in their current form, and we make recommendations for future text-to-video retrieval benchmarks.

We present IMU2CLIP, a novel pre-training approach to align Inertial Measurement Unit (IMU) motion sensor recordings with text and video, by projecting them into the joint representation space of Contrastive Language-Image Pre-training (CLIP). The proposed approach allows IMU2CLIP to translate human motions (as measured by IMU sensors) into their corresponding textual descriptions and videos – while preserving the transitivity across these modalities. We introduce several new IMU-based Wearable AI applications such as motion-based media search, or an LM-based multimodal reasoning with motion sensor data – all using text as the grounding platform. In addition, we show that IMU2CLIP significantly improves downstream performances when fine-tuned for each application, demonstrating its universal usage as a new pre-trained resource. Our code and models will be released publicly.

With ever increasing interest in task-oriented dialog systems, the recent work on Situated and Interactive Multimodal Conversations (SIMMC 2.0) aims to develop personal assistants that interact with users, grounded in an immersive and co-observed setting of photo-realistic scenes. The dataset contains 11k task-oriented dialogs set in an interactive shopping scenario, spanning more than 117k utterances. In order to push research towards this next generation virtual assistants, the SIMMC 2.1 challenge was conducted at the Eleventh Dialog System Technology Challenge (DSTC) which had entries from across the world competing to achieve the state-of-the-art performance in the SIMMC 2.1 task. In this report, we present and compare 13 SIMMC 2.1 model entries from 5 trams across the world to understand the current progress made across the last three years (starting with SIMMC 1.0 and 2.0 challenges) for multimodal task-oriented dialog systems. We hope that our analysis throws light on components that showed promise in addition to identifying the gaps for future research towards this grand goal of an immersive multimodal conversational agent.

Cross-modal contrastive learning has led the recent advances in multimodal retrieval with its simplicity and effectiveness. In this work, however, we reveal that cross-modal contrastive learning suffers from incorrect normalization of the sum retrieval probabilities of each text or video instance. Specifically, we show that many test instances are either over- or under-represented during retrieval, significantly hurting the retrieval performance. To address this problem, we propose Normalized Contrastive Learning (NCL) which utilizes the Sinkhorn-Knopp algorithm to compute the instance-wise biases that properly normalize the sum retrieval probabilities of each instance so that every text and video instance is fairly represented during cross-modal retrieval. Empirical study shows that NCL brings consistent and significant gains in text-video retrieval on different model architectures, with new state-of-the-art multimodal retrieval metrics on the ActivityNet, MSVD, and MSR-VTT datasets without any architecture engineering.

Recent years have seen an increasing trend in the volume of personal media captured by users, thanks to the advent of smartphones and smart glasses, resulting in large media collections. Despite conversation being an intuitive human-computer interface, current efforts focus mostly on single-shot natural language based media retrieval to aid users query their media and re-live their memories. This severely limits the search functionality as users can neither ask follow-up queries nor obtain information without first formulating a single-turn query.In this work, we propose dialogs for connected memories as a powerful tool to empower users to search their media collection through a multi-turn, interactive conversation. Towards this, we collect a new task-oriented dialog dataset COMET, which contains 11.5k user↔assistant dialogs (totalling 103k utterances), grounded in simulated personal memory graphs. We employ a resource-efficient, two-phase data collection pipeline that uses: (1) a novel multimodal dialog simulator that generates synthetic dialog flows grounded in memory graphs, and, (2) manual paraphrasing to obtain natural language utterances. We analyze COMET, formulate four main tasks to benchmark meaningful progress, and adopt state-of-the-art language models as strong baselines, in order to highlight the multimodal challenges captured by our dataset.

Existing studies in dialogue system research mostly treat task-oriented dialogue and chit-chat as separate domains. Towards building a human-like assistant that can converse naturally and seamlessly with users, it is important to build a dialogue system that conducts both types of conversations effectively. In this work, we investigate how task-oriented dialogue and knowledge-grounded chit-chat can be effectively integrated into a single model. To this end, we create a new dataset, KETOD (Knowledge-Enriched Task-Oriented Dialogue), where we naturally enrich task-oriented dialogues with chit-chat based on relevant entity knowledge. We also propose two new models, SimpleToDPlus and Combiner, for the proposed task. Experimental results on both automatic and human evaluations show that the proposed methods can significantly improve the performance in knowledge-enriched response generation while maintaining a competitive task-oriented dialog performance. We believe our new dataset will be a valuable resource for future studies. Our dataset and code are publicly available at https://github.com/facebookresearch/ketod.

A video-grounded dialogue system is required to understand both dialogue, which contains semantic dependencies from turn to turn, and video, which contains visual cues of spatial and temporal scene variations. Building such dialogue systems is a challenging problem, involving various reasoning types on both visual and language inputs. Existing benchmarks do not have enough annotations to thoroughly analyze dialogue systems and understand their capabilities and limitations in isolation. These benchmarks are also not explicitly designed to minimise biases that models can exploit without actual reasoning. To address these limitations, in this paper, we present DVD, a Diagnostic Dataset for Video-grounded Dialogue. The dataset is designed to contain minimal biases and has detailed annotations for the different types of reasoning over the spatio-temporal space of video. Dialogues are synthesized over multiple question turns, each of which is injected with a set of cross-turn semantic relationships. We use DVD to analyze existing approaches, providing interesting insights into their abilities and limitations. In total, DVD is built from 11k CATER synthetic videos and contains 10 instances of 10-round dialogues for each video, resulting in more than 100k dialogues and 1M question-answer pairs. Our code and dataset are publicly available.

Existing dialogue corpora and models are typically designed under two disjoint motives: while task-oriented systems focus on achieving functional goals (e.g., booking hotels), open-domain chatbots aim at making socially engaging conversations. In this work, we propose to integrate both types of systems by Adding Chit-Chat to ENhance Task-ORiented dialogues (ACCENTOR), with the goal of making virtual assistant conversations more engaging and interactive. Specifically, we propose a Human <-> AI collaborative data collection approach for generating diverse chit-chat responses to augment task-oriented dialogues with minimal annotation effort. We then present our new chit-chat-based annotations to 23.8K dialogues from two popular task-oriented datasets (Schema-Guided Dialogue and MultiWOZ 2.1) and demonstrate their advantage over the originals via human evaluation. Lastly, we propose three new models for adding chit-chat to task-oriented dialogues, explicitly trained to predict user goals and to generate contextually relevant chit-chat responses. Automatic and human evaluations show that, compared with the state-of-the-art task-oriented baseline, our models can code-switch between task and chit-chat to be more engaging, interesting, knowledgeable, and humanlike, while maintaining competitive task performance.

Zero-shot cross-domain dialogue state tracking (DST) enables us to handle unseen domains without the expense of collecting in-domain data. In this paper, we propose a slot descriptions enhanced generative approach for zero-shot cross-domain DST. Specifically, our model first encodes a dialogue context and a slot with a pre-trained self-attentive encoder, and generates slot value in auto-regressive manner. In addition, we incorporate Slot Type Informed Descriptions that capture the shared information of different slots to facilitates the cross-domain knowledge transfer. Experimental results on MultiWOZ shows that our model significantly improve existing state-of-the-art results in zero-shot cross-domain setting.

There is a growing interest in virtual assistants with multimodal capabilities, e.g., inferring the context of a conversation through scene understanding. The recently released situated and interactive multimodal conversations (SIMMC) dataset addresses this trend by enabling research to create virtual assistants, which are capable of taking into account the scene that user sees when conversing with the user and also interacting with items in the scene. The SIMMC dataset is novel in that it contains fully annotated user-assistant, task-orientated dialogs where the user and an assistant co-observe the same visual elements and the latter can take actions to update the scene. The SIMMC challenge, held as part of theNinth Dialog System Technology Challenge(DSTC9), propelled the development of various models which together set a new state-of-the-art on the SIMMC dataset. In this work, we compare and analyze these models to identify‘what worked?’, and the remaining gaps;‘whatnext?’. Our analysis shows that even though pretrained language models adapted to this set-ting show great promise, there are indications that multimodal context isn’t fully utilised, and there is a need for better and scalable knowledge base integration. We hope this first-of-its-kind analysis for SIMMC models provides useful insights and opportunities for further research in multimodal conversational agents

Existing conversational systems are mostly agent-centric, which assumes the user utterances will closely follow the system ontology. However, in real-world scenarios, it is highly desirable that users can speak freely and naturally. In this work, we attempt to build a user-centric dialogue system for conversational recommendation. As there is no clean mapping for a user’s free form utterance to an ontology, we first model the user preferences as estimated distributions over the system ontology and map the user’s utterances to such distributions. Learning such a mapping poses new challenges on reasoning over various types of knowledge, ranging from factoid knowledge, commonsense knowledge to the users’ own situations. To this end, we build a new dataset named NUANCED that focuses on such realistic settings, with 5.1k dialogues, 26k turns of high-quality user responses. We conduct experiments, showing both the usefulness and challenges of our problem setting. We believe NUANCED can serve as a valuable resource to push existing research from the agent-centric system to the user-centric system. The code and data are publicly available.

Next generation task-oriented dialog systems need to understand conversational contexts with their perceived surroundings, to effectively help users in the real-world multimodal environment. Existing task-oriented dialog datasets aimed towards virtual assistance fall short and do not situate the dialog in the user’s multimodal context. To overcome, we present a new dataset for Situated and Interactive Multimodal Conversations, SIMMC 2.0, which includes 11K task-oriented user<->assistant dialogs (117K utterances) in the shopping domain, grounded in immersive and photo-realistic scenes. The dialogs are collection using a two-phase pipeline: (1) A novel multimodal dialog simulator generates simulated dialog flows, with an emphasis on diversity and richness of interactions, (2) Manual paraphrasing of generating utterances to draw from natural language distribution. We provide an in-depth analysis of the collected dataset, and describe in detail the four main benchmark tasks we propose for SIMMC 2.0. Our baseline model, powered by the state-of-the-art language model, shows promising results, and highlights new challenges and directions for the community to study.

Continual learning in task-oriented dialogue systems allows the system to add new domains and functionalities overtime after deployment, without incurring the high cost of retraining the whole system each time. In this paper, we propose a first-ever continual learning benchmark for task-oriented dialogue systems with 37 domains to be learned continuously in both modularized and end-to-end learning settings. In addition, we implement and compare multiple existing continual learning baselines, and we propose a simple yet effective architectural method based on residual adapters. We also suggest that the upper bound performance of continual learning should be equivalent to multitask learning when data from all domain is available at once. Our experiments demonstrate that the proposed architectural method and a simple replay-based strategy perform better, by a large margin, compared to other continuous learning techniques, and only slightly worse than the multitask learning upper bound while being 20X faster in learning new domains. We also report several trade-offs in terms of parameter usage, memory size and training time, which are important in the design of a task-oriented dialogue system. The proposed benchmark is released to promote more research in this direction.

Zero-shot transfer learning for dialogue state tracking (DST) enables us to handle a variety of task-oriented dialogue domains without the expense of collecting in-domain data. In this work, we propose to transfer the cross-task knowledge from general question answering (QA) corpora for the zero-shot DST task. Specifically, we propose TransferQA, a transferable generative QA model that seamlessly combines extractive QA and multi-choice QA via a text-to-text transformer framework, and tracks both categorical slots and non-categorical slots in DST. In addition, we introduce two effective ways to construct unanswerable questions, namely, negative question sampling and context truncation, which enable our model to handle none value slots in the zero-shot DST setting. The extensive experiments show that our approaches substantially improve the existing zero-shot and few-shot results on MultiWoz. Moreover, compared to the fully trained baseline on the Schema-Guided Dialogue dataset, our approach shows better generalization ability in unseen domains.

Next generation virtual assistants are envisioned to handle multimodal inputs (e.g., vision, memories of previous interactions, and the user’s utterances), and perform multimodal actions (, displaying a route while generating the system’s utterance). We introduce Situated Interactive MultiModal Conversations (SIMMC) as a new direction aimed at training agents that take multimodal actions grounded in a co-evolving multimodal input context in addition to the dialog history. We provide two SIMMC datasets totalling ~13K human-human dialogs (~169K utterances) collected using a multimodal Wizard-of-Oz (WoZ) setup, on two shopping domains: (a) furniture – grounded in a shared virtual environment; and (b) fashion – grounded in an evolving set of images. Datasets include multimodal context of the items appearing in each scene, and contextual NLU, NLG and coreference annotations using a novel and unified framework of SIMMC conversational acts for both user and assistant utterances. Finally, we present several tasks within SIMMC as objective evaluation protocols, such as structural API prediction, response generation, and dialog state tracking. We benchmark a collection of existing models on these SIMMC tasks as strong baselines, and demonstrate rich multimodal conversational interactions. Our data, annotations, and models will be made publicly available.

We study an end-to-end approach for conversational recommendation that dynamically manages and reasons over users’ past (offline) preferences and current (online) requests through a structured and cumulative user memory knowledge graph. This formulation extends existing state tracking beyond the boundary of a single dialog to user state tracking (UST). For this study, we create a new Memory Graph (MG) <-> Conversational Recommendation parallel corpus called MGConvRex with 7K+ human-to-human role-playing dialogs, grounded on a large-scale user memory bootstrapped from real-world user scenarios. MGConvRex captures human-level reasoning over user memory and has disjoint training/testing sets of users for zero-shot (cold-start) reasoning for recommendation. We propose a simple yet expandable formulation for constructing and updating the MG, and an end-to-end graph-based reasoning model that updates MG from unstructured utterances and predicts optimal dialog policies (eg recommendation) based on updated MG. The prediction of our proposed model inherits the graph structure, providing a natural way to explain policies. Experiments are conducted for both offline metrics and online simulation, showing competitive results.

Open-ended human learning and information-seeking are increasingly mediated by digital assistants. However, such systems often ignore the user’s pre-existing knowledge. Assuming a correlation between engagement and user responses such as “liking” messages or asking followup questions, we design a Wizard-of-Oz dialog task that tests the hypothesis that engagement increases when users are presented with facts related to what they know. Through crowd-sourcing of this experiment, we collect and release 14K dialogs (181K utterances) where users and assistants converse about geographic topics like geopolitical entities and locations. This dataset is annotated with pre-existing user knowledge, message-level dialog acts, grounding to Wikipedia, and user reactions to messages. Responses using a user’s prior knowledge increase engagement. We incorporate this knowledge into a multi-task model that reproduces human assistant policies and improves over a bert content model by 13 mean reciprocal rank points.

We study a conversational reasoning model that strategically traverses through a large-scale common fact knowledge graph (KG) to introduce engaging and contextually diverse entities and attributes. For this study, we collect a new Open-ended Dialog <-> KG parallel corpus called OpenDialKG, where each utterance from 15K human-to-human role-playing dialogs is manually annotated with ground-truth reference to corresponding entities and paths from a large-scale KG with 1M+ facts. We then propose the DialKG Walker model that learns the symbolic transitions of dialog contexts as structured traversals over KG, and predicts natural entities to introduce given previous dialog contexts via a novel domain-agnostic, attention-based graph path decoder. Automatic and human evaluations show that our model can retrieve more natural and human-like responses than the state-of-the-art baselines or rule-based models, in both in-domain and cross-domain tasks. The proposed model also generates a KG walk path for each entity retrieved, providing a natural way to explain conversational reasoning.

We introduce Episodic Memory QA, the task of answering personal user questions grounded on memory graph (MG), where episodic memories and related entity nodes are connected via relational edges. We create a new benchmark dataset first by generating synthetic memory graphs with simulated attributes, and by composing 100K QA pairs for the generated MG with bootstrapped scripts. To address the unique challenges for the proposed task, we propose Memory Graph Networks (MGN), a novel extension of memory networks to enable dynamic expansion of memory slots through graph traversals, thus able to answer queries in which contexts from multiple linked episodes and external knowledge are required. We then propose the Episodic Memory QA Net with multiple module networks to effectively handle various question types. Empirical results show improvement over the QA baselines in top-k answer prediction accuracy in the proposed task. The proposed model also generates a graph walk path and attention vectors for each predicted answer, providing a natural way to explain its QA reasoning.

We demonstrate a conversational system which engages the user through a multi-modal, multi-turn dialog over the user’s memories. The system can perform QA over memories by responding to user queries to recall specific attributes and associated media (e.g. photos) of past episodic memories. The system can also make proactive suggestions to surface related events or facts from past memories to make conversations more engaging and natural. To implement such a system, we collect a new corpus of memory grounded conversations, which comprises human-to-human role-playing dialogs given synthetic memory graphs with simulated attributes. Our proof-of-concept system operates on these synthetic memory graphs, however it can be trained and applied to real-world user memory data (e.g. photo albums, etc.) We present the architecture of the proposed conversational system, and example queries that the system supports.

We introduce a new task called Multimodal Named Entity Recognition (MNER) for noisy user-generated data such as tweets or Snapchat captions, which comprise short text with accompanying images. These social media posts often come in inconsistent or incomplete syntax and lexical notations with very limited surrounding textual contexts, bringing significant challenges for NER. To this end, we create a new dataset for MNER called SnapCaptions (Snapchat image-caption pairs submitted to public and crowd-sourced stories with fully annotated named entities). We then build upon the state-of-the-art Bi-LSTM word/character based NER models with 1) a deep image network which incorporates relevant visual context to augment textual information, and 2) a generic modality-attention module which learns to attenuate irrelevant modalities while amplifying the most informative ones to extract contexts from, adaptive to each sample and token. The proposed MNER model with modality attention significantly outperforms the state-of-the-art text-only NER models by successfully leveraging provided visual contexts, opening up potential applications of MNER on myriads of social media platforms.

We introduce the new Multimodal Named Entity Disambiguation (MNED) task for multimodal social media posts such as Snapchat or Instagram captions, which are composed of short captions with accompanying images. Social media posts bring significant challenges for disambiguation tasks because 1) ambiguity not only comes from polysemous entities, but also from inconsistent or incomplete notations, 2) very limited context is provided with surrounding words, and 3) there are many emerging entities often unseen during training. To this end, we build a new dataset called SnapCaptionsKB, a collection of Snapchat image captions submitted to public and crowd-sourced stories, with named entity mentions fully annotated and linked to entities in an external knowledge base. We then build a deep zeroshot multimodal network for MNED that 1) extracts contexts from both text and image, and 2) predicts correct entity in the knowledge graph embeddings space, allowing for zeroshot disambiguation of entities unseen in training set as well. The proposed model significantly outperforms the state-of-the-art text-only NED models, showing efficacy and potentials of the MNED task.