Ziyan Jiang

Yancheng Wang, Ziyan Jiang, Zheng Chen et al. · amazon-science

While the recommendation system (RS) has advanced significantly through deep learning, current RS approaches usually train and fine-tune models on task-specific datasets, limiting their generalizability to new recommendation tasks and their ability to leverage external knowledge due to model scale and data size constraints. Thus, we designed an LLM-powered autonomous recommender agent, RecMind, which is capable of leveraging external knowledge, utilizing tools with careful planning to provide zero-shot personalized recommendations. We propose a Self-Inspiring algorithm to improve the planning ability. At each intermediate step, the LLM self-inspires to consider all previously explored states to plan for the next step. This mechanism greatly improves the model's ability to comprehend and utilize historical information in planning for recommendation. We evaluate RecMind's performance in various recommendation scenarios. Our experiment shows that RecMind outperforms existing zero/few-shot LLM-based recommendation baseline methods in various tasks and achieves comparable performance to a fully trained recommendation model P5.

Jialin Yang, Dongfu Jiang, Lipeng He et al. · amazon-science, utoronto

As Large Language Models (LLMs) become integral to software development workflows, their ability to generate structured outputs has become critically important. We introduce StructEval, a comprehensive benchmark for evaluating LLMs' capabilities in producing both non-renderable (JSON, YAML, CSV) and renderable (HTML, React, SVG) structured formats. Unlike prior benchmarks, StructEval systematically evaluates structural fidelity across diverse formats through two paradigms: 1) generation tasks, producing structured output from natural language prompts, and \textbf{2)} conversion tasks, translating between structured formats. Our benchmark encompasses 18 formats and 44 types of task, with novel metrics for format adherence and structural correctness. Results reveal significant performance gaps-even state-of-the-art models like o1-mini achieve only 75.58 average score, with open-source alternatives lagging approximately 10 points behind. We find generation tasks more challenging than conversion tasks, and producing correct visual content more difficult than generating text-only structures.

Zekun Xi, Yichen Nie, Ziyan Jiang et al.

Despite the rapid advancement of generative agents, their deployment in real-world industry scenarios often encounters significant challenges due to a lack of domain-specific knowledge. To address this gap, we present KnowPilot: a Domain-Specific Knowledge Augmented Generative Agent System. KnowPilot is an open-source framework that integrates task-specific priors, explicit knowledge, and experiential knowledge to enhance agent performance in specialized applications. It combines knowledge retrieval from structured repositories with a memory system capable of capturing expert experience through human AI interaction. Taking domain-specific writing generation as a representative case, KnowPilot enables private deployment, supports injection of task requirements, loads private knowledge bases, and stores tacit expert knowledge as persistent memory. Experimental results demonstrate that KnowPilot achieves superior performance in domain-oriented text generation and is applicable across fields such as medicine, finance and industry.

Jizhan Fang, Tianhe Lu, Yunzhi Yao et al.

Chinese, as a linguistic system rich in depth and complexity, is characterized by distinctive elements such as ancient poetry, proverbs, idioms, and other cultural constructs. However, current Large Language Models (LLMs) face limitations in these specialized domains, highlighting the need for the development of comprehensive datasets that can assess, continuously update, and progressively improve these culturally-grounded linguistic competencies through targeted training optimizations. To address this gap, we introduce CKnowEdit, the first-ever Chinese knowledge editing dataset designed to correct linguistic, factual, and logical errors in LLMs. We collect seven types of knowledge from a wide range of sources, including classical texts, idioms, and content from Baidu Tieba Ruozhiba, taking into account the unique polyphony, antithesis, and logical structures inherent in the Chinese language. By analyzing this dataset, we highlight the challenges current LLMs face in mastering Chinese. Furthermore, our evaluation of state-of-the-art knowledge editing techniques reveals opportunities to advance the correction of Chinese knowledge. Code and dataset are available at https://github.com/zjunlp/EasyEdit.

Elan Markowitz, Ziyan Jiang, Fan Yang et al. · amazon-science

This work explores unifying knowledge enhanced recommendation with multi-domain recommendation systems in a conversational AI assistant application. Multi-domain recommendation leverages users' interactions in previous domains to improve recommendations in a new one. Knowledge graph enhancement seeks to use external knowledge graphs to improve recommendations within a single domain. Both research threads incorporate related information to improve the recommendation task. We propose to unify these approaches: using information from interactions in other domains as well as external knowledge graphs to make predictions in a new domain that would not be possible with either information source alone. We develop a new model and demonstrate the additive benefit of these approaches on a dataset derived from millions of users' queries for content across three domains (videos, music, and books) in a live virtual assistant application. We demonstrate significant improvement on overall recommendations as well as on recommendations for new users of a domain.

Niranjan Uma Naresh, Ziyan Jiang, Ankit et al. · amazon-science

Conversational understanding is an integral part of modern intelligent devices. In a large fraction of the global traffic from customers using smart digital assistants, frictions in dialogues may be attributed to incorrect understanding of the entities in a customer's query due to factors including ambiguous mentions, mispronunciation, background noise and faulty on-device signal processing. Such errors are compounded by two common deficiencies from intelligent devices namely, (1) the device not being tailored to individual customers, and (2) the device responses being unaware of the context in the conversation session. Viewing this problem via the lens of retrieval-based search engines, we build and evaluate a scalable entity correction system, PENTATRON. The system leverages a parametric transformer-based language model to learn patterns from in-session customer-device interactions coupled with a non-parametric personalized entity index to compute the correct query, which aids downstream components in reasoning about the best response. In addition to establishing baselines and demonstrating the value of personalized and context-aware systems, we use multitasking to learn the domain of the correct entity. We also investigate the utility of language model prompts. Through extensive experiments, we show a significant upward movement of the key metric (Exact Match) by up to 500.97% (relative to the baseline).

Jinglun Cai, Mingda Li, Ziyan Jiang et al. · amazon-science

Query Rewriting (QR) plays a critical role in large-scale dialogue systems for reducing frictions. When there is an entity error, it imposes extra challenges for a dialogue system to produce satisfactory responses. In this work, we propose KG-ECO: Knowledge Graph enhanced Entity COrrection for query rewriting, an entity correction system with corrupt entity span detection and entity retrieval/re-ranking functionalities. To boost the model performance, we incorporate Knowledge Graph (KG) to provide entity structural information (neighboring entities encoded by graph neural networks) and textual information (KG entity descriptions encoded by RoBERTa). Experimental results show that our approach yields a clear performance gain over two baselines: utterance level QR and entity correction without utilizing KG information. The proposed system is particularly effective for few-shot learning cases where target entities are rarely seen in training or there is a KG relation between the target entity and other contextual entities in the query.

Yifei He, Haoxiang Wang, Ziyan Jiang et al. · amazon-science

Reward models (RM) capture the values and preferences of humans and play a central role in Reinforcement Learning with Human Feedback (RLHF) to align pretrained large language models (LLMs). Traditionally, training these models relies on extensive human-annotated preference data, which poses significant challenges in terms of scalability and cost. To overcome these limitations, we propose Semi-Supervised Reward Modeling (SSRM), an approach that enhances RM training using unlabeled data. Given an unlabeled dataset, SSRM involves three key iterative steps: pseudo-labeling unlabeled examples, selecting high-confidence examples through a confidence threshold, and supervised finetuning on the refined dataset. Across extensive experiments on various model configurations, we demonstrate that SSRM significantly improves reward models without incurring additional labeling costs. Notably, SSRM can achieve performance comparable to models trained entirely on labeled data of equivalent volumes. Overall, SSRM substantially reduces the dependency on large volumes of human-annotated data, thereby decreasing the overall cost and time involved in training effective reward models.

Peigen Liu, Rui Ding, Yuren Mao et al. · amazon-science

Large Language Model (LLM)-based Collective Intelligence (CI) presents a promising approach to overcoming the data wall and continuously boosting the capabilities of LLM agents. However, there is currently no dedicated arena for evolving and benchmarking LLM-based CI. To address this gap, we introduce OpenHospital, an interactive arena where physician agents can evolve CI through interactions with patient agents. This arena employs a data-in-agent-self paradigm that rapidly enhances agent capabilities and provides robust evaluation metrics for benchmarking both medical proficiency and system efficiency. Experiments demonstrate the effectiveness of OpenHospital in both fostering and quantifying CI.

Haoming Xu, Shuxun Wang, Yanqiu Zhao et al.

This paper presents the ZJUKLAB team's submission for SemEval-2025 Task 4: Unlearning Sensitive Content from Large Language Models. This task aims to selectively erase sensitive knowledge from large language models, avoiding both over-forgetting and under-forgetting issues. We propose an unlearning system that leverages Model Merging (specifically TIES-Merging), combining two specialized models into a more balanced unlearned model. Our system achieves competitive results, ranking second among 26 teams, with an online score of 0.944 for Task Aggregate and 0.487 for overall Aggregate. In this paper, we also conduct local experiments and perform a comprehensive analysis of the unlearning process, examining performance trajectories, loss dynamics, and weight perspectives, along with several supplementary experiments, to understand the effectiveness of our method. Furthermore, we analyze the shortcomings of our method and evaluation metrics, emphasizing that MIA scores and ROUGE-based metrics alone are insufficient to fully evaluate successful unlearning. Finally, we emphasize the need for more comprehensive evaluation methodologies and rethinking of unlearning objectives in future research. Code is available at https://github.com/zjunlp/unlearn/tree/main/semeval25.

Jizhan Fang, Xinle Deng, Haoming Xu et al. · amazon-science

Despite their remarkable capabilities, Large Language Models (LLMs) struggle to effectively leverage historical interaction information in dynamic and complex environments. Memory systems enable LLMs to move beyond stateless interactions by introducing persistent information storage, retrieval, and utilization mechanisms. However, existing memory systems often introduce substantial time and computational overhead. To this end, we introduce a new memory system called LightMem, which strikes a balance between the performance and efficiency of memory systems. Inspired by the Atkinson-Shiffrin model of human memory, LightMem organizes memory into three complementary stages. First, cognition-inspired sensory memory rapidly filters irrelevant information through lightweight compression and groups information according to their topics. Next, topic-aware short-term memory consolidates these topic-based groups, organizing and summarizing content for more structured access. Finally, long-term memory with sleep-time update employs an offline procedure that decouples consolidation from online inference. Experiments on LongMemEval with GPT and Qwen backbones show that LightMem outperforms strong baselines in accuracy (up to 10.9% gains) while reducing token usage by up to 117x, API calls by up to 159x, and runtime by over 12x. The code is available at https://github.com/zjunlp/LightMem.

Shigeki Karita, Nanxin Chen, Tomoki Hayashi et al.

Sequence-to-sequence models have been widely used in end-to-end speech processing, for example, automatic speech recognition (ASR), speech translation (ST), and text-to-speech (TTS). This paper focuses on an emergent sequence-to-sequence model called Transformer, which achieves state-of-the-art performance in neural machine translation and other natural language processing applications. We undertook intensive studies in which we experimentally compared and analyzed Transformer and conventional recurrent neural networks (RNN) in a total of 15 ASR, one multilingual ASR, one ST, and two TTS benchmarks. Our experiments revealed various training tips and significant performance benefits obtained with Transformer for each task including the surprising superiority of Transformer in 13/15 ASR benchmarks in comparison with RNN. We are preparing to release Kaldi-style reproducible recipes using open source and publicly available datasets for all the ASR, ST, and TTS tasks for the community to succeed our exciting outcomes.

Jiacheng Chen, Tianhao Liang, Sherman Siu et al. · amazon-science

We present MEGA-Bench, an evaluation suite that scales multimodal evaluation to over 500 real-world tasks, to address the highly heterogeneous daily use cases of end users. Our objective is to optimize for a set of high-quality data samples that cover a highly diverse and rich set of multimodal tasks, while enabling cost-effective and accurate model evaluation. In particular, we collected 505 realistic tasks encompassing over 8,000 samples from 16 expert annotators to extensively cover the multimodal task space. Instead of unifying these problems into standard multi-choice questions (like MMMU, MMBench, and MMT-Bench), we embrace a wide range of output formats like numbers, phrases, code, \LaTeX, coordinates, JSON, free-form, etc. To accommodate these formats, we developed over 40 metrics to evaluate these tasks. Unlike existing benchmarks, MEGA-Bench offers a fine-grained capability report across multiple dimensions (e.g., application, input type, output format, skill), allowing users to interact with and visualize model capabilities in depth. We evaluate a wide variety of frontier vision-language models on MEGA-Bench to understand their capabilities across these dimensions.

Rui Meng, Ziyan Jiang, Ye Liu et al. · amazon-science

Multimodal embedding models have been crucial in enabling various downstream tasks such as semantic similarity, information retrieval, and clustering over different modalities. However, existing multimodal embeddings like VLM2Vec, E5-V, GME are predominantly focused on natural images, with limited support for other visual forms such as videos and visual documents. This restricts their applicability in real-world scenarios, including AI agents, multi-modal search and recommendation, and retrieval-augmented generation (RAG). To close this gap, we propose VLM2Vec-V2, a unified framework for learning embeddings across diverse visual forms. First, we introduce MMEB-V2, a comprehensive benchmark that extends MMEB with five new task types: visual document retrieval, video retrieval, temporal grounding, video classification and video question answering - spanning text, image, video, and visual document inputs. Next, we train VLM2Vec-V2, a general-purpose embedding model that supports text, image, video, and visual document inputs. Extensive experiments show that VLM2Vec-V2 achieves strong performance not only on the newly introduced video and document retrieval tasks, but also improves over prior baselines on the original image benchmarks. Through extensive evaluation, our study offers insights into the generalizability of various multimodal embedding models and highlights effective strategies for unified embedding learning, laying the groundwork for more scalable and adaptable representation learning in both research and real-world settings.

Haohang Huang, Xuan Lu, Mingyi Su et al.

Multimodal embedding models aim to map heterogeneous inputs, such as text, images, videos, and audio, into a shared semantic space. However, existing methods and benchmarks remain largely limited to partial modality coverage, making it difficult to systematically evaluate full-modality representation learning. In this work, we take a step toward the full-modality setting. We introduce MMEB-V3, a comprehensive benchmark that evaluates embeddings across text, image, video, audio, as well as agent-centric scenarios. To enable more fine-grained diagnosis, we further construct OmniSET (Omni-modality Semantic Equivalence Tuples), where semantically equivalent instances are represented across modalities, allowing us to disentangle semantic similarity from modality effects. Through experiments on MMEB-V3, we conduct a systematic analysis of full-modality embeddings and identify three key findings: (1) models often fail to retrieve the intended target modality; (2) cross-modal retrieval is highly asymmetric and dominated by query-modality bias; and (3) instruction-induced shifts are either insufficient or misaligned with the target modality, and therefore do not reliably improve retrieval. These results indicate that current multimodal embeddings are not yet capable of reliably enforcing modality constraints specified by instructions, and consequently fail to exhibit consistent modality-aware retrieval behavior. We hope MMEB-V3 provides a useful benchmark for understanding and diagnosing these limitations, and for guiding future research on full-modality embeddings.

Zheng Zhang, Fan Yang, Ziyan Jiang et al. · amazon-science

Recent advances in large language models (LLMs) have enhanced their ability to process long input contexts. This development is particularly crucial for tasks that involve retrieving knowledge from an external datastore, which can result in long inputs. However, recent studies show a positional bias in LLMs, demonstrating varying performance depending on the location of useful information within the input sequence. In this study, we conduct extensive experiments to investigate the root causes of positional bias. Our findings indicate that the primary contributor to LLM positional bias stems from the inherent positional preferences of different models. We demonstrate that merely employing prompt-based solutions is inadequate for overcoming the positional preferences. To address this positional bias issue of a pre-trained LLM, we developed a Position-Aware Parameter Efficient Fine-Tuning (PAPEFT) approach which is composed of a data augmentation technique and a parameter efficient adapter, enhancing a uniform attention distribution across the input context. Our experiments demonstrate that the proposed approach effectively reduces positional bias, improving LLMs' effectiveness in handling long context sequences for various tasks that require externally retrieved knowledge.

Xuan Zhang, Ziyan Jiang, Rui Meng et al. · amazon-science

Trajectory data, capturing human actions and environmental states across various modalities, holds significant potential for enhancing AI agent capabilities, particularly in GUI environments. However, how to model the representation of trajectory-level data presents a significant challenge that has not been systematically addressed amid explosive trajectory data growth. In this work, we introduce Multimodal Trajectory Retrieval, bridging the gap between universal retrieval and agent-centric trajectory modeling. We construct the Unified Agent Trajectory Dataset (UATD) from annotated demonstrations and states across diverse real-world scenarios. Based on this, we present GAE-Bench, a benchmark containing a large number of trajectory-based retrieval pairs. In addition, we propose GAE-Retriever, a multimodal retrieval framework that adopts vision-language models and incorporates optimized contrastive learning through a token selection and the GradCache mechanism. Comprehensive evaluations across multiple datasets show that GAE-Retriever consistently outperforms strong baselines in retrieval recall, highlighting its effectiveness in advancing multimodal trajectory retrieval.

Ziyan Jiang, Xueguang Ma, Wenhu Chen

In traditional RAG framework, the basic retrieval units are normally short. The common retrievers like DPR normally work with 100-word Wikipedia paragraphs. Such a design forces the retriever to search over a large corpus to find the `needle' unit. In contrast, the readers only need to generate answers from the short retrieved units. The imbalanced `heavy' retriever and `light' reader design can lead to sub-optimal performance. The loss of contextual information in the short, chunked units may increase the likelihood of introducing hard negatives during the retrieval stage. Additionally, the reader might not fully leverage the capabilities of recent advancements in LLMs. In order to alleviate the imbalance, we propose a new framework LongRAG, consisting of a `long retriever' and a `long reader'. In the two Wikipedia-based datasets, NQ and HotpotQA, LongRAG processes the entire Wikipedia corpus into 4K-token units by grouping related documents. By increasing the unit size, we significantly reduce the total number of units. This greatly reduces the burden on the retriever, resulting in strong retrieval performance with only a few (less than 8) top units. Without requiring any training, LongRAG achieves an EM of 62.7% on NQ and 64.3% on HotpotQA, which are on par with the (fully-trained) SoTA model. Furthermore, we test on two non-Wikipedia-based datasets, Qasper and MultiFieldQA-en. LongRAG processes each individual document as a single (long) unit rather than chunking them into smaller units. By doing so, we achieve an F1 score of 25.9% on Qasper and 57.5% on MultiFieldQA-en. Our study offers insights into the future roadmap for combining RAG with long-context LLMs.

Xuan He, Dongfu Jiang, Ge Zhang et al.

The recent years have witnessed great advances in video generation. However, the development of automatic video metrics is lagging significantly behind. None of the existing metric is able to provide reliable scores over generated videos. The main barrier is the lack of large-scale human-annotated dataset. In this paper, we release VideoFeedback, the first large-scale dataset containing human-provided multi-aspect score over 37.6K synthesized videos from 11 existing video generative models. We train VideoScore (initialized from Mantis) based on VideoFeedback to enable automatic video quality assessment. Experiments show that the Spearman correlation between VideoScore and humans can reach 77.1 on VideoFeedback-test, beating the prior best metrics by about 50 points. Further result on other held-out EvalCrafter, GenAI-Bench, and VBench show that VideoScore has consistently much higher correlation with human judges than other metrics. Due to these results, we believe VideoScore can serve as a great proxy for human raters to (1) rate different video models to track progress (2) simulate fine-grained human feedback in Reinforcement Learning with Human Feedback (RLHF) to improve current video generation models.

Yubo Wang, Xueguang Ma, Ge Zhang et al.

In the age of large-scale language models, benchmarks like the Massive Multitask Language Understanding (MMLU) have been pivotal in pushing the boundaries of what AI can achieve in language comprehension and reasoning across diverse domains. However, as models continue to improve, their performance on these benchmarks has begun to plateau, making it increasingly difficult to discern differences in model capabilities. This paper introduces MMLU-Pro, an enhanced dataset designed to extend the mostly knowledge-driven MMLU benchmark by integrating more challenging, reasoning-focused questions and expanding the choice set from four to ten options. Additionally, MMLU-Pro eliminates the trivial and noisy questions in MMLU. Our experimental results show that MMLU-Pro not only raises the challenge, causing a significant drop in accuracy by 16% to 33% compared to MMLU but also demonstrates greater stability under varying prompts. With 24 different prompt styles tested, the sensitivity of model scores to prompt variations decreased from 4-5% in MMLU to just 2% in MMLU-Pro. Additionally, we found that models utilizing Chain of Thought (CoT) reasoning achieved better performance on MMLU-Pro compared to direct answering, which is in stark contrast to the findings on the original MMLU, indicating that MMLU-Pro includes more complex reasoning questions. Our assessments confirm that MMLU-Pro is a more discriminative benchmark to better track progress in the field.

Zheng Chen, Ziyan Jiang, Fan Yang et al.

Conversational AI systems such as Alexa need to understand defective queries to ensure robust conversational understanding and reduce user friction. These defective queries often arise from user ambiguities, mistakes, or errors in automatic speech recognition (ASR) and natural language understanding (NLU). Personalized query rewriting is an approach that focuses on reducing defects in queries by taking into account the user's individual behavior and preferences. It typically relies on an index of past successful user interactions with the conversational AI. However, unseen interactions within the user's history present additional challenges for personalized query rewriting. This paper presents our "Collaborative Query Rewriting" approach, which specifically addresses the task of rewriting new user interactions that have not been previously observed in the user's history. This approach builds a "User Feedback Interaction Graph" (FIG) of historical user-entity interactions and leverages multi-hop graph traversal to enrich each user's index to cover future unseen defective queries. The enriched user index is called a Collaborative User Index and contains hundreds of additional entries. To counteract precision degradation from the enlarged index, we add additional transformer layers to the L1 retrieval model and incorporate graph-based and guardrail features into the L2 ranking model. Since the user index can be pre-computed, we further investigate the utilization of a Large Language Model (LLM) to enhance the FIG for user-entity link prediction in the Video/Music domains. Specifically, this paper investigates the Dolly-V2 7B model. We found that the user index augmented by the fine-tuned Dolly-V2 generation significantly enhanced the coverage of future unseen user interactions, thereby boosting QR performance on unseen queries compared with the graph traversal only approach.

Fan Yang, Zheng Chen, Ziyan Jiang et al.

Large language models (LLMs) have recently received significant attention for their exceptional capabilities. Despite extensive efforts in developing general-purpose LLMs that can be utilized in various natural language processing (NLP) tasks, there has been less research exploring their potential in recommender systems. In this paper, we propose a novel framework, named PALR, which aiming to combine user history behaviors (such as clicks, purchases, ratings, etc.) with LLMs to generate user preferred items. Specifically, we first use user/item interactions as guidance for candidate retrieval. Then we adopt a LLM-based ranking model to generate recommended items. Unlike existing approaches that typically adopt general-purpose LLMs for zero/few-shot recommendation testing or training on small-sized language models (with less than 1 billion parameters), which cannot fully elicit LLMs' reasoning abilities and leverage rich item side parametric knowledge, we fine-tune a 7 billion parameters LLM for the ranking purpose. This model takes retrieval candidates in natural language format as input, with instruction which explicitly asking to select results from input candidates during inference. Our experimental results demonstrate that our solution outperforms state-of-the-art models on various sequential recommendation tasks.