arXiv subjects

Guorui Zhou

Publications and source records attributed to Guorui Zhou.

At least 73 records · Page 4Linked to original sources

arXiv2025

Leanabell-Prover-V2: Verifier-integrated Reasoning for Formal Theorem Proving via Reinforcement Learning

We introduce our Leanabell-Prover-V2, a 7B large language models (LLMs) that can produce formal theorem proofs in Lean 4, with verifier-integrated Long Chain-of-Thoughts (CoT). Following our previous work Leanabell-Prover-V1, we continual to choose to posttrain existing strong prover models for further performance improvement. In our V2 version, we mainly upgrade the Reinforcement Learning (RL) with feedback provided by the Lean 4 verifier. Crucially, verifier feedback, such as indicating success or detailing specific errors, allows the LLM to become ``self-aware'' of the correctness of its own reasoning process and learn to reflexively correct errors. Leanabell-Prover-V2 directly optimizes LLM reasoning trajectories with multi-turn verifier interactions, together with feedback token masking for stable RL training and a simple reward strategy. Experiments show that Leanabell-Prover-V2 improves performance by 3.2% (pass@128) with Kimina-Prover-Preview-Distill-7B and 2.0% (pass@128) with DeepSeek-Prover-V2-7B on the MiniF2F test set. The source codes, curated data and models are available at: https://github.com/Leanabell-LM/Leanabell-Prover-V2.

cs.AI↗

arXiv2025

RLEP: Reinforcement Learning with Experience Replay for LLM Reasoning

Reinforcement learning (RL) for large language models is an energy-intensive endeavor: training can be unstable, and the policy may gradually drift away from its pretrained weights. We present \emph{RLEP}\, -- \,Reinforcement Learning with Experience rePlay\, -- \,a two-phase framework that first collects verified trajectories and then replays them during subsequent training. At every update step, the policy is optimized on mini-batches that blend newly generated rollouts with these replayed successes. By replaying high-quality examples, RLEP steers the model away from fruitless exploration, focuses learning on promising reasoning paths, and delivers both faster convergence and stronger final performance. On the Qwen2.5-Math-7B base model, RLEP reaches baseline peak accuracy with substantially fewer updates and ultimately surpasses it, improving accuracy on AIME-2024 from 38.2% to 39.9%, on AIME-2025 from 19.8% to 22.3%, and on AMC-2023 from 77.0% to 82.2%. Our code, datasets, and checkpoints are publicly available at https://github.com/Kwai-Klear/RLEP to facilitate reproducibility and further research.

cs.CL↗

arXiv2025

Kwai Keye-VL Technical Report

While Multimodal Large Language Models (MLLMs) demonstrate remarkable capabilities on static images, they often fall short in comprehending dynamic, information-dense short-form videos, a dominant medium in today's digital landscape. To bridge this gap, we introduce \textbf{Kwai Keye-VL}, an 8-billion-parameter multimodal foundation model engineered for leading-edge performance in short-video understanding while maintaining robust general-purpose vision-language abilities. The development of Keye-VL rests on two core pillars: a massive, high-quality dataset exceeding 600 billion tokens with a strong emphasis on video, and an innovative training recipe. This recipe features a four-stage pre-training process for solid vision-language alignment, followed by a meticulous two-phase post-training process. The first post-training stage enhances foundational capabilities like instruction following, while the second phase focuses on stimulating advanced reasoning. In this second phase, a key innovation is our five-mode ``cold-start'' data mixture, which includes ``thinking'', ``non-thinking'', ``auto-think'', ``think with image'', and high-quality video data. This mixture teaches the model to decide when and how to reason. Subsequent reinforcement learning (RL) and alignment steps further enhance these reasoning capabilities and correct abnormal model behaviors, such as repetitive outputs. To validate our approach, we conduct extensive evaluations, showing that Keye-VL achieves state-of-the-art results on public video benchmarks and remains highly competitive on general image-based tasks (Figure 1). Furthermore, we develop and release the \textbf{KC-MMBench}, a new benchmark tailored for real-world short-video scenarios, where Keye-VL shows a significant advantage.

cs.CV↗

arXiv2025

Modality Curation: Building Universal Embeddings for Advanced Multimodal Information Retrieval

Multimodal information retrieval (MIR) faces inherent challenges due to the heterogeneity of data sources and the complexity of cross-modal alignment. While previous studies have identified modal gaps in feature spaces, a systematic approach to address these challenges remains unexplored. In this work, we introduce UNITE, a universal framework that tackles these challenges through two critical yet underexplored aspects: data curation and modality-aware training configurations. Our work provides the first comprehensive analysis of how modality-specific data properties influence downstream task performance across diverse scenarios. Moreover, we propose Modal-Aware Masked Contrastive Learning (MAMCL) to mitigate the competitive relationships among the instances of different modalities. Our framework achieves state-of-the-art results on multiple multimodal retrieval benchmarks, outperforming existing methods by notable margins. Through extensive experiments, we demonstrate that strategic modality curation and tailored training protocols are pivotal for robust cross-modal representation learning. This work not only advances MIR performance but also provides a foundational blueprint for future research in multimodal systems. Our project is available at https://friedrichor.github.io/projects/UNITE.

cs.CV↗

arXiv2025

Why Distillation can Outperform Zero-RL: The Role of Flexible Reasoning

Reinforcement learning (RL) has played an important role in improving the reasoning ability of large language models (LLMs). Some studies apply RL directly to \textit{smaller} base models (known as zero-RL) and also achieve notable progress. However, in this paper, we show that using only 920 examples, a simple distillation method based on the base model can clearly outperform zero-RL, which typically requires much more data and computational cost. By analyzing the token frequency in model outputs, we find that the distilled model shows more flexible reasoning. It uses anthropomorphic tokens and logical connectors much more often than the zero-RL model. Further analysis reveals that distillation enhances the presence of two advanced cognitive behaviors: Multi-Perspective Thinking or Attempting and Metacognitive Awareness. Frequent occurrences of these two advanced cognitive behaviors give rise to flexible reasoning, which is essential for solving complex reasoning problems, while zero-RL fails to significantly boost the frequency of these behaviors.

cs.AI↗

arXiv2025

Tool-Star: Empowering LLM-Brained Multi-Tool Reasoner via Reinforcement Learning

Recently, large language models (LLMs) have shown remarkable reasoning capabilities via large-scale reinforcement learning (RL). However, leveraging the RL algorithm to empower effective multi-tool collaborative reasoning in LLMs remains an open challenge. In this paper, we introduce Tool-Star, an RL-based framework designed to empower LLMs to autonomously invoke multiple external tools during stepwise reasoning. Tool-Star integrates six types of tools and incorporates systematic designs in both data synthesis and training. To address the scarcity of tool-use data, we propose a general tool-integrated reasoning data synthesis pipeline, which combines tool-integrated prompting with hint-based sampling to automatically and scalably generate tool-use trajectories. A subsequent quality normalization and difficulty-aware classification process filters out low-quality samples and organizes the dataset from easy to hard. Furthermore, we propose a two-stage training framework to enhance multi-tool collaborative reasoning by: (1) cold-start fine-tuning, which guides LLMs to explore reasoning patterns via tool-invocation feedback; and (2) a multi-tool self-critic RL algorithm with hierarchical reward design, which reinforces reward understanding and promotes effective tool collaboration. Experimental analyses on over 10 challenging reasoning benchmarks highlight the effectiveness and efficiency of Tool-Star. The code is available at https://github.com/dongguanting/Tool-Star.

cs.CL↗

arXiv2025

Pantheon: Personalized Multi-objective Ensemble Sort via Iterative Pareto Policy Optimization

In this paper, we provide our milestone ensemble sort work and the first-hand practical experience, Pantheon, which transforms ensemble sorting from a "human-curated art" to a "machine-optimized science". Compared with formulation-based ensemble sort, our Pantheon has the following advantages: (1) Personalized Joint Training: our Pantheon is jointly trained with the real-time ranking model, which could capture ever-changing user personalized interests accurately. (2) Representation inheritance: instead of the highly compressed Pxtrs, our Pantheon utilizes the fine-grained hidden-states as model input, which could benefit from the Ranking model to enhance our model complexity. Meanwhile, to reach a balanced multi-objective ensemble sort, we further devise an \textbf{iterative Pareto policy optimization} (IPPO) strategy to consider the multiple objectives at the same time. To our knowledge, this paper is the first work to replace the entire formulation-based ensemble sort in industry RecSys, which was fully deployed at Kuaishou live-streaming services, serving 400 Million users daily.

cs.SI↗

arXiv2025

Short Video Segment-level User Dynamic Interests Modeling in Personalized Recommendation

The rapid growth of short videos has necessitated effective recommender systems to match users with content tailored to their evolving preferences. Current video recommendation models primarily treat each video as a whole, overlooking the dynamic nature of user preferences with specific video segments. In contrast, our research focuses on segment-level user interest modeling, which is crucial for understanding how users' preferences evolve during video browsing. To capture users' dynamic segment interests, we propose an innovative model that integrates a hybrid representation module, a multi-modal user-video encoder, and a segment interest decoder. Our model addresses the challenges of capturing dynamic interest patterns, missing segment-level labels, and fusing different modalities, achieving precise segment-level interest prediction. We present two downstream tasks to evaluate the effectiveness of our segment interest modeling approach: video-skip prediction and short video recommendation. Our experiments on real-world short video datasets with diverse modalities show promising results on both tasks. It demonstrates that segment-level interest modeling brings a deep understanding of user engagement and enhances video recommendations. We also release a unique dataset that includes segment-level video data and diverse user behaviors, enabling further research in segment-level interest modeling. This work pioneers a novel perspective on understanding user segment-level preference, offering the potential for more personalized and engaging short video experiences.

cs.IR↗

arXiv2025

LLM-Alignment Live-Streaming Recommendation

In recent years, integrated short-video and live-streaming platforms have gained massive global adoption, offering dynamic content creation and consumption. Unlike pre-recorded short videos, live-streaming enables real-time interaction between authors and users, fostering deeper engagement. However, this dynamic nature introduces a critical challenge for recommendation systems (RecSys): the same live-streaming vastly different experiences depending on when a user watching. To optimize recommendations, a RecSys must accurately interpret the real-time semantics of live content and align them with user preferences.

cs.IR↗

arXiv2025

OneRec: Unifying Retrieve and Rank with Generative Recommender and Iterative Preference Alignment

Recently, generative retrieval-based recommendation systems have emerged as a promising paradigm. However, most modern recommender systems adopt a retrieve-and-rank strategy, where the generative model functions only as a selector during the retrieval stage. In this paper, we propose OneRec, which replaces the cascaded learning framework with a unified generative model. To the best of our knowledge, this is the first end-to-end generative model that significantly surpasses current complex and well-designed recommender systems in real-world scenarios. Specifically, OneRec includes: 1) an encoder-decoder structure, which encodes the user's historical behavior sequences and gradually decodes the videos that the user may be interested in. We adopt sparse Mixture-of-Experts (MoE) to scale model capacity without proportionally increasing computational FLOPs. 2) a session-wise generation approach. In contrast to traditional next-item prediction, we propose a session-wise generation, which is more elegant and contextually coherent than point-by-point generation that relies on hand-crafted rules to properly combine the generated results. 3) an Iterative Preference Alignment module combined with Direct Preference Optimization (DPO) to enhance the quality of the generated results. Unlike DPO in NLP, a recommendation system typically has only one opportunity to display results for each user's browsing request, making it impossible to obtain positive and negative samples simultaneously. To address this limitation, We design a reward model to simulate user generation and customize the sampling strategy. Extensive experiments have demonstrated that a limited number of DPO samples can align user interest preferences and significantly improve the quality of generated results. We deployed OneRec in the main scene of Kuaishou, achieving a 1.6\% increase in watch-time, which is a substantial improvement.

cs.IR↗

arXiv2025

ChorusCVR: Chorus Supervision for Entire Space Post-Click Conversion Rate Modeling

Post-click conversion rate (CVR) estimation is a vital task in many recommender systems of revenue businesses, e.g., e-commerce and advertising. In a perspective of sample, a typical CVR positive sample usually goes through a funnel of exposure to click to conversion. For lack of post-event labels for un-clicked samples, CVR learning task commonly only utilizes clicked samples, rather than all exposed samples as for click-through rate (CTR) learning task. However, during online inference, CVR and CTR are estimated on the same assumed exposure space, which leads to a inconsistency of sample space between training and inference, i.e., sample selection bias (SSB). To alleviate SSB, previous wisdom proposes to design novel auxiliary tasks to enable the CVR learning on un-click training samples, such as CTCVR and counterfactual CVR, etc. Although alleviating SSB to some extent, none of them pay attention to the discrimination between ambiguous negative samples (un-clicked) and factual negative samples (clicked but un-converted) during modelling, which makes CVR model lacks robustness. To full this gap, we propose a novel ChorusCVR model to realize debiased CVR learning in entire-space.

cs.IR↗

arXiv2025

ELASTIC: Efficient Linear Attention for Sequential Interest Compression

State-of-the-art sequential recommendation models heavily rely on transformer's attention mechanism. However, the quadratic computational and memory complexities of self attention have limited its scalability for modeling users' long range behaviour sequences. To address this problem, we propose ELASTIC, an Efficient Linear Attention for SequenTial Interest Compression, requiring only linear time complexity and decoupling model capacity from computational cost. Specifically, ELASTIC introduces a fixed length interest experts with linear dispatcher attention mechanism which compresses the long-term behaviour sequences to a significantly more compact representation which reduces up to 90% GPU memory usage with x2.7 inference speed up. The proposed linear dispatcher attention mechanism significantly reduces the quadratic complexity and makes the model feasible for adequately modeling extremely long sequences. Moreover, in order to retain the capacity for modeling various user interests, ELASTIC initializes a vast learnable interest memory bank and sparsely retrieves compressed user's interests from the memory with a negligible computational overhead. The proposed interest memory retrieval technique significantly expands the cardinality of available interest space while keeping the same computational cost, thereby striking a trade-off between recommendation accuracy and efficiency. To validate the effectiveness of our proposed ELASTIC, we conduct extensive experiments on various public datasets and compare it with several strong sequential recommenders. Experimental results demonstrate that ELASTIC consistently outperforms baselines by a significant margin and also highlight the computational efficiency of ELASTIC when modeling long sequences. We will make our implementation code publicly available.

cs.AI↗

arXiv2024

Multifaceted User Modeling in Recommendation: A Federated Foundation Models Approach

Multifaceted user modeling aims to uncover fine-grained patterns and learn representations from user data, revealing their diverse interests and characteristics, such as profile, preference, and personality. Recent studies on foundation model-based recommendation have emphasized the Transformer architecture's remarkable ability to capture complex, non-linear user-item interaction relationships. This paper aims to advance foundation model-based recommendersystems by introducing enhancements to multifaceted user modeling capabilities. We propose a novel Transformer layer designed specifically for recommendation, using the self-attention mechanism to capture sequential user-item interaction patterns. Specifically, we design a group gating network to identify user groups, enabling hierarchical discovery across different layers, thereby capturing the multifaceted nature of user interests through multiple Transformer layers. Furthermore, to broaden the data scope and further enhance multifaceted user modeling, we extend the framework to a federated setting, enabling the use of private datasets while ensuring privacy. Experimental validations on benchmark datasets demonstrate the superior performance of our proposed method. Code is available.

cs.IR↗

arXiv2024

QARM: Quantitative Alignment Multi-Modal Recommendation at Kuaishou

In recent years, with the significant evolution of multi-modal large models, many recommender researchers realized the potential of multi-modal information for user interest modeling. In industry, a wide-used modeling architecture is a cascading paradigm: (1) first pre-training a multi-modal model to provide omnipotent representations for downstream services; (2) The downstream recommendation model takes the multi-modal representation as additional input to fit real user-item behaviours. Although such paradigm achieves remarkable improvements, however, there still exist two problems that limit model performance: (1) Representation Unmatching: The pre-trained multi-modal model is always supervised by the classic NLP/CV tasks, while the recommendation models are supervised by real user-item interaction. As a result, the two fundamentally different tasks' goals were relatively separate, and there was a lack of consistent objective on their representations; (2) Representation Unlearning: The generated multi-modal representations are always stored in cache store and serve as extra fixed input of recommendation model, thus could not be updated by recommendation model gradient, further unfriendly for downstream training. Inspired by the two difficulties challenges in downstream tasks usage, we introduce a quantitative multi-modal framework to customize the specialized and trainable multi-modal information for different downstream models.

cs.IR↗

arXiv2024

KuaiFormer: Transformer-Based Retrieval at Kuaishou

In large-scale content recommendation systems, retrieval serves as the initial stage in the pipeline, responsible for selecting thousands of candidate items from billions of options to pass on to ranking modules. Traditionally, the dominant retrieval method has been Embedding-Based Retrieval (EBR) using a Deep Neural Network (DNN) dual-tower structure. However, applying transformer in retrieval tasks has been the focus of recent research, though real-world industrial deployment still presents significant challenges. In this paper, we introduce KuaiFormer, a novel transformer-based retrieval framework deployed in a large-scale content recommendation system. KuaiFormer fundamentally redefines the retrieval process by shifting from conventional score estimation tasks (such as click-through rate estimate) to a transformer-driven Next Action Prediction paradigm. This shift enables more effective real-time interest acquisition and multi-interest extraction, significantly enhancing retrieval performance. KuaiFormer has been successfully integrated into Kuaishou App's short-video recommendation system since May 2024, serving over 400 million daily active users and resulting in a marked increase in average daily usage time of Kuaishou users. We provide insights into both the technical and business aspects of deploying transformer in large-scale recommendation systems, addressing practical challenges encountered during industrial implementation. Our findings offer valuable guidance for engineers and researchers aiming to leverage transformer models to optimize large-scale content recommendation systems.

cs.IR↗

arXiv2024

RecFlow: An Industrial Full Flow Recommendation Dataset

Industrial recommendation systems (RS) rely on the multi-stage pipeline to balance effectiveness and efficiency when delivering items from a vast corpus to users. Existing RS benchmark datasets primarily focus on the exposure space, where novel RS algorithms are trained and evaluated. However, when these algorithms transition to real world industrial RS, they face a critical challenge of handling unexposed items which are a significantly larger space than the exposed one. This discrepancy profoundly impacts their practical performance. Additionally, these algorithms often overlook the intricate interplay between multiple RS stages, resulting in suboptimal overall system performance. To address this issue, we introduce RecFlow, an industrial full flow recommendation dataset designed to bridge the gap between offline RS benchmarks and the real online environment. Unlike existing datasets, RecFlow includes samples not only from the exposure space but also unexposed items filtered at each stage of the RS funnel. Our dataset comprises 38M interactions from 42K users across nearly 9M items with additional 1.9B stage samples collected from 9.3M online requests over 37 days and spanning 6 stages. Leveraging the RecFlow dataset, we conduct courageous exploration experiments, showcasing its potential in designing new algorithms to enhance effectiveness by incorporating stage-specific samples. Some of these algorithms have already been deployed online, consistently yielding significant gains. We propose RecFlow as the first comprehensive benchmark dataset for the RS community, supporting research on designing algorithms at any stage, study of selection bias, debiased algorithms, multi-stage consistency and optimality, multi-task recommendation, and user behavior modeling. The RecFlow dataset, along with the corresponding source code, is available at https://github.com/RecFlow-ICLR/RecFlow.

cs.IR↗

arXiv2024

Chinese Tiny LLM: Pretraining a Chinese-Centric Large Language Model

In this study, we introduce CT-LLM, a 2B large language model (LLM) that illustrates a pivotal shift towards prioritizing the Chinese language in developing LLMs. Uniquely initiated from scratch, CT-LLM diverges from the conventional methodology by primarily incorporating Chinese textual data, utilizing an extensive corpus of 1,200 billion tokens, including 800 billion Chinese tokens, 300 billion English tokens, and 100 billion code tokens. This strategic composition facilitates the model's exceptional proficiency in understanding and processing Chinese, a capability further enhanced through alignment techniques. Demonstrating remarkable performance on the CHC-Bench, CT-LLM excels in Chinese language tasks, and showcases its adeptness in English through SFT. This research challenges the prevailing paradigm of training LLMs predominantly on English corpora and then adapting them to other languages, broadening the horizons for LLM training methodologies. By open-sourcing the full process of training a Chinese LLM, including a detailed data processing procedure with the obtained Massive Appropriate Pretraining Chinese Corpus (MAP-CC), a well-chosen multidisciplinary Chinese Hard Case Benchmark (CHC-Bench), and the 2B-size Chinese Tiny LLM (CT-LLM), we aim to foster further exploration and innovation in both academia and industry, paving the way for more inclusive and versatile language models.

cs.CL↗

arXiv2024

A Unified Framework for Cross-Domain Recommendation

In addressing the persistent challenges of data-sparsity and cold-start issues in domain-expert recommender systems, Cross-Domain Recommendation (CDR) emerges as a promising methodology. CDR aims at enhancing prediction performance in the target domain by leveraging interaction knowledge from related source domains, particularly through users or items that span across multiple domains (e.g., Short-Video and Living-Room). For academic research purposes, there are a number of distinct aspects to guide CDR method designing, including the auxiliary domain number, domain-overlapped element, user-item interaction types, and downstream tasks. With so many different CDR combination scenario settings, the proposed scenario-expert approaches are tailored to address a specific vertical CDR scenario, and often lack the capacity to adapt to multiple horizontal scenarios. In an effect to coherently adapt to various scenarios, and drawing inspiration from the concept of domain-invariant transfer learning, we extend the former SOTA model UniCDR in five different aspects, named as UniCDR+. Our work was successfully deployed on the Kuaishou Living-Room RecSys.

cs.IR↗