Large Multimodal Models (LMMs) have ushered in a new era in artificial intelligence, merging capabilities in both language and vision to form highly capable \textbf{Visual Foundation Agents} that are postulated to excel across a myriad of tasks. However, existing benchmarks fail to sufficiently challenge or showcase the full potential of LMMs as visual foundation agents in complex, real-world environments. To address this gap, we introduce VisualAgentBench (VAB), a comprehensive and unified benchmark specifically designed to train and evaluate LMMs as visual foundation agents across diverse scenarios in one standard setting, including Embodied, Graphical User Interface, and Visual Design, with tasks formulated to probe the depth of LMMs' understanding and interaction capabilities. Through rigorous testing across 9 proprietary LMM APIs and 9 open models (18 in total), we demonstrate the considerable yet still developing visual agent capabilities of these models. Additionally, VAB explores the synthesizing of visual agent trajectory data through hybrid methods including Program-based Solvers, LMM Agent Bootstrapping, and Human Demonstrations, offering insights into obstacles, solutions, and trade-offs one may meet in developing open LMM agents. Our work not only aims to benchmark existing models but also provides an instrumental playground for future development into visual foundation agents. Code, train, and test data are available at \url{https://github.com/THUDM/VisualAgentBench}.

Agentic search such as Deep Research systems-where agents autonomously browse the web, synthesize information, and return comprehensive citation-backed answers-represents a major shift in how users interact with web-scale information. While promising greater efficiency and cognitive offloading, the growing complexity and open-endedness of agentic search have outpaced existing evaluation benchmarks and methodologies, which largely assume short search horizons and static answers. In this paper, we introduce Mind2Web 2, a benchmark of 130 realistic, high-quality, and long-horizon tasks that require real-time web browsing and extensive information synthesis, constructed with over 1000 hours of human labor. To address the challenge of evaluating time-varying and complex answers, we propose a novel Agent-as-a-Judge framework. Our method constructs task-specific judge agents based on a tree-structured rubric design to automatically assess both answer correctness and source attribution. We conduct a comprehensive evaluation of ten frontier agentic search systems and human performance, along with a detailed error analysis to draw insights for future development. The best-performing system, OpenAI Deep Research, can already achieve 50-70% of human performance while spending half the time, highlighting its great potential. Altogether, Mind2Web 2 provides a rigorous foundation for developing and benchmarking the next generation of agentic search systems.

The applications of large language models (LLMs) have expanded well beyond the confines of text processing, signaling a new era where LLMs are envisioned as generalist agents capable of operating within complex environments.These environments are often highly expansive, making it impossible for the LLM to process them within its short-term memory.Motivated by recent research on extending the capabilities of LLMs with tools, we seek to investigate the intriguing potential of tools to augment LLMs in handling such complexity by introducing a novel class of tools, termed middleware, to aid in the proactive exploration within these massive environments.Such specialized tools can serve as a middleware layer shielding the LLM from environmental complexity.In two representative complex environmentsknowledge bases (KBs) and databases-we demonstrate the significant potential of augmenting language agents with tools in complex environments.Notably, equipped with the middleware, GPT-4 achieves 2.8× the performance of the best baseline in tasks requiring access to database content and 2.2× in KB tasks.Our findings illuminate the path for advancing language agents in real-world applications.1

The potential of Large Language Model (LLM) as agents has been widely acknowledged recently. Thus, there is an urgent need to quantitatively evaluate LLMs as agents on challenging tasks in interactive environments. We present AgentBench, a multi-dimensional benchmark that consists of 8 distinct environments to assess LLM-as-Agent's reasoning and decision-making abilities. Our extensive test over 29 API-based and open-sourced (OSS) LLMs shows that, while top commercial LLMs present a strong ability of acting as agents in complex environments, there is a significant disparity in performance between them and many OSS competitors that are no larger than 70B. We identify the typical reasons of failures in environments and LLMs, showing that poor long-term reasoning, decision-making, and instruction following abilities are the main obstacles for developing usable LLM agents. Improving instruction following and training on high quality multi-round alignment data could improve agent performance. And different from existing assumptions, training on code present ambivalent impacts on different agent tasks. Datasets, environments, and an integrated evaluation package for AgentBench are released at https://github.com/THUDM/AgentBench.

Generating a set of text is a common challenge for many NLP applications, for example, automatically providing multiple keyphrases for a document to facilitate user reading. Existing generative models use a sequential decoder that generates a single sequence successively, and the set generation problem is converted to sequence generation via concatenating multiple text into a long text sequence. However, the elements of a set are unordered, which makes this scheme suffer from biased or conflicting training signals. In this paper, we propose a branching decoder, which can generate a dynamic number of tokens at each time-step and branch multiple generation paths. In particular, paths are generated individually so that no order dependence is required. Moreover, multiple paths can be generated in parallel which greatly reduces the inference time. Experiments on several keyphrase generation datasets demonstrate that the branching decoder is more effective and efficient than the existing sequential decoder.

In order to thrive in hostile and ever-changing natural environments, mammalian brains evolved to store large amounts of knowledge about the world and continually integrate new information while avoiding catastrophic forgetting. Despite the impressive accomplishments, large language models (LLMs), even with retrieval-augmented generation (RAG), still struggle to efficiently and effectively integrate a large amount of new experiences after pre-training. In this work, we introduce HippoRAG, a novel retrieval framework inspired by the hippocampal indexing theory of human long-term memory to enable deeper and more efficient knowledge integration over new experiences. HippoRAG synergistically orchestrates LLMs, knowledge graphs, and the Personalized PageRank algorithm to mimic the different roles of neocortex and hippocampus in human memory. We compare HippoRAG with existing RAG methods on multi-hop question answering (QA) and show that our method outperforms the state-of-the-art methods remarkably, by up to 20%. Single-step retrieval with HippoRAG achieves comparable or better performance than iterative retrieval like IRCoT while being 10-20 times cheaper and 6-13 times faster, and integrating HippoRAG into IRCoT brings further substantial gains. Finally, we show that our method can tackle new types of scenarios that are out of reach of existing methods.

A key missing capacity of current language models (LMs) is grounding to real-world environments.Most existing work for grounded language understanding uses LMs to directly generate plans that can be executed in the environment to achieve the desired effects.It thereby casts the burden of ensuring grammaticality, faithfulness, and controllability all on the LMs.We propose Pangu, a generic framework for grounded language understanding that capitalizes on the discriminative ability of LMs instead of their generative ability.Pangu consists of a symbolic agent and a neural LM working in a concerted fashion: The agent explores the environment to incrementally construct valid plans, and the LM evaluates the plausibility of the candidate plans to guide the search process.A case study on the challenging problem of knowledge base question answering (KBQA), which features a massive environment, demonstrates the remarkable effectiveness and flexibility of Pangu: A BERT-base LM is sufficient for setting a new record on standard KBQA datasets, and larger LMs further bring substantial gains.Pangu also enables, for the first time, effective few-shot in-context learning for KBQA with large LMs such as Codex. 1

Question answering over knowledge bases is considered a difficult problem due to the challenge of generalizing to a wide variety of possible natural language questions.Additionally, the heterogeneity of knowledge base schema items between different knowledge bases often necessitates specialized training for different knowledge base question-answering (KBQA) datasets.To handle questions over diverse KBQA datasets with a unified trainingfree framework, we propose KB-BINDER, which for the first time enables few-shot incontext learning over KBQA tasks.Firstly, KB-BINDER leverages large language models like Codex to generate logical forms as the draft for a specific question by imitating a few demonstrations.Secondly, KB-BINDER grounds on the knowledge base to bind the generated draft to an executable one with BM25 score matching.The experimental results on four public heterogeneous KBQA datasets show that KB-BINDER can achieve a strong performance with only a few in-context demonstrations.Especially on GraphQA and 3-hop MetaQA, KB-BINDER can even outperform the state-of-the-art trained models.On GrailQA and WebQSP, our model is also on par with other fully-trained models.We believe KB-BINDER can serve as an important baseline for future research.Our code is available at

We introduce Mind2Web, the first dataset for developing and evaluating generalist agents for the web that can follow language instructions to complete complex tasks on any website. Existing datasets for web agents either use simulated websites or only cover a limited set of websites and tasks, thus not suitable for generalist web agents. With over 2,000 open-ended tasks collected from 137 websites spanning 31 domains and crowdsourced action sequences for the tasks, Mind2Web provides three necessary ingredients for building generalist web agents: 1) diverse domains, websites, and tasks, 2) use of real-world websites instead of simulated and simplified ones, and 3) a broad spectrum of user interaction patterns. Based on Mind2Web, we conduct an initial exploration of using large language models (LLMs) for building generalist web agents. While the raw HTML of real-world websites are often too large to be fed to LLMs, we show that first filtering it with a small LM significantly improves the effectiveness and efficiency of LLMs. Our solution demonstrates a decent level of performance, even on websites or entire domains the model has never seen before, but there is still a substantial room to improve towards truly generalizable agents. We open-source our dataset, model implementation, and trained models (https://osu-nlp-group.github.io/Mind2Web) to facilitate further research on building a generalist agent for the web.

Relevance search over knowledge graphs seeks top-ranked answer entities that are most relevant to a query entity. Since the semantics of relevance varies with the user need and its formalization is difficult for non-experts, existing methods infer semantics from user-provided example answer entities. However, a user may provide very few examples, even none at the beginning of interaction, thereby limiting the effectiveness of such methods. In this paper, we vision a more practical scenario called labeling-based iterative relevance search: instead of effortfully inputting example answer entities, the user effortlessly (e.g., implicitly) labels current answer entities, and is rewarded with improved answer entities in the next iteration. To realize the scenario, our approach TRAVERS incorporates two rankers: a diversity-oriented ranker for supporting cold start and avoiding converging to sub-optimum caused by noisy labels, and a relevance-oriented ranker capable of handling unbalanced labels. Moreover, the two rankers and their combination dynamically evolve over iterations. TRAVERS outperformed a variety of baselines in experiments with simulated and real user behavior.

Question answering on knowledge bases (KBQA) poses a unique challenge for semantic parsing research due to two intertwined challenges: large search space and ambiguities in schema linking. Conventional ranking-based KBQA models, which rely on a candidate enumeration step to reduce the search space, struggle with flexibility in predicting complicated queries and have impractical running time. In this paper, we present ArcaneQA, a novel generation-based model that addresses both the large search space and the schema linking challenges in a unified framework with two mutually boosting ingredients: dynamic program induction for tackling the large search space and dynamic contextualized encoding for schema linking. Experimental results on multiple popular KBQA datasets demonstrate the highly competitive performance of ArcaneQA in both effectiveness and efficiency.

Zixian Huang, Ao Wu, Jiaying Zhou, Yu Gu, Yue Zhao, Gong Cheng. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

Ad hoc dataset retrieval is a trending topic in IR research. Methods and systems are evolving from metadata-based to content-based ones which exploit the data itself for improving retrieval accuracy but thus far lack a specialized test collection. In this paper, we build and release the first test collection for ad hoc content-based dataset retrieval, where content-oriented dataset queries and content-based relevance judgments are annotated by human experts who are assisted with a dashboard designed specifically for comprehensively and conveniently browsing both the metadata and data of a dataset. We conduct extensive experiments on the test collection to analyze its difficulty and provide insights into the underlying task.

Vardaan Pahuja, Yu Gu, Wenhu Chen, Mehdi Bahrami, Lei Liu, Wei-Peng Chen, Yu Su. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Existing studies on question answering on knowledge bases (KBQA) mainly operate with the standard i.i.d. assumption, i.e., training distribution over questions is the same as the test distribution. However, i.i.d. may be neither achievable nor desirable on large-scale KBs because 1) true user distribution is hard to capture and 2) randomly sampling training examples from the enormous space would be data-inefficient. Instead, we suggest that KBQA models should have three levels of built-in generalization: i.i.d., compositional, and zero-shot. To facilitate the development of KBQA models with stronger generalization, we construct and release a new large-scale, high-quality dataset with 64,331 questions, GrailQA, and provide evaluation settings for all three levels of generalization. In addition, we propose a novel BERT-based KBQA model. The combination of our dataset and model enables us to thoroughly examine and demonstrate, for the first time, the key role of pre-trained contextual embeddings like BERT in the generalization of KBQA.1

Relevance search over a knowledge graph (KG) has gained much research attention. Given a query entity in a KG, the problem is to find its most relevant entities. However, the relevance function is hidden and dynamic. Different users for different queries may consider relevance from different angles of semantics. The ambiguity in a query is more noticeable in the presence of thousands of types of entities and relations in a schema-rich KG, which has challenged the effectiveness and scalability of existing methods. To meet the challenge, our approach called RelSUE requests a user to provide a small number of answer entities as examples, and then automatically learns the most likely relevance function from these examples. Specifically, we assume the intent of a query can be characterized by a set of meta-paths at the schema level. RelSUE searches a KG for diversified significant meta-paths that best characterize the relevance of the user-provided examples to the query entity. It reduces the large search space of a schema-rich KG using distance and degree-based heuristics, and performs reasoning to deduplicate meta-paths that represent equivalent query-specific semantics. Finally, a linear model is learned to predict meta-path based relevance. Extensive experiments demonstrate that RelSUE outperforms several state-of-the-art methods.