109 citations · 425 across the 59 of their papers we have counts for
63 papers
MIRA: Reinforcing Multimodal Reasoning via Deceptive Contextual Augmentation
Zhihan Yin, Jianxin Liang, Yifeng Yao +5
While Reinforcement Learning with Verifiable Rewards (RLVR) has significantly advanced multimodal reasoning, existing frameworks suffer from a critical "perception-reasoning gap."…
Learning from Next-Frame Prediction: Towards Unified and Scalable Visual World Modeling
Yifeng Yao, Jinghan Li, Zhihao Zhang +5
Video provides a natural source of supervision: what happens next serves as a learning signal for visual world modeling. Yet masked video pretraining focuses on recovering missing…
Boosting LLM Exploration via Weak-Model Guidance in RLVR
Xingyu Shen, Huishuai Zhang, Peng Li +2
Reinforcement Learning with Verifiable Rewards (RLVR) significantly improves LLM reasoning but often causes a drop in policy entropy, leading to narrowed reasoning coverage and deg…
Token-Operations-Oriented Inference Optimization Techniques for Large Models
Shiguo Lian, Kai Wang, Zhaoxiang Liu +23
Large model inference optimization serves as a key foundation for supporting the scalable, low-cost, and highly stable operation of large model services. Centered on token-oriented…
BitNet Text Embeddings
Zhen Li, Xin Huang, Liang Wang +8
LLM-based text embedders have substantially improved retrieval and semantic representation quality, but their deployment remains costly: large backbone models slow down embedding i…
DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation
Xin Cheng, Xingkai Yu, Chenze Shao +30
Speculative decoding accelerates Large Language Model (LLM) inference by decoupling draft generation from target verification. While recent parallel drafters efficiently propose lo…
Qwen-Image-Agent: Bridging the Context Gap in Real-World Image Generation
Zekai Zhang, Jiahao Li, Jie Zhang +18
While text-to-image (T2I) models have achieved remarkable progress, they struggle with real-world requests that are often underspecified, implicit, or dependent on up-to-date knowl…
Beyond Fixed Benchmarks and Worst-Case Attacks: Dynamic Boundary Evaluation for Language Models
Haoxiang Wang, Da Yu, Huishuai Zhang
Evaluating large language models (LLMs) today rests on fixed benchmarks that apply the same set of items to any model, producing ceiling and floor effects that mask capability gaps…
GeoBuildBench: A Benchmark for Interactive and Executable Geometry Construction from Natural Language
Jinwoong Kim, Rui Yang, Huishuai Zhang
We introduce GeoBuildBench, a benchmark designed to evaluate whether large language models and multimodal agents can ground informal natural-language plane geometry problems into e…
SWE-MiniSandbox: Container-Free Reinforcement Learning for Building Software Engineering Agents
Danlong Yuan, Wei Wu, Enhan Zhao +4
Reinforcement learning (RL) has become a key paradigm for training software engineering (SWE) agents, but existing pipelines typically rely on per-task containers for isolation. At…
Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models
Xin Cheng, Rui Tian, Wangding Zeng +18
While Mixture-of-Experts (MoE) scales capacity via conditional computation, Transformers lack a native primitive for knowledge lookup, forcing them to inefficiently simulate retrie…
ReMamba: Equip Mamba with Effective Long-Sequence Modeling
Danlong Yuan, Jiahao Liu, Bei Li +4
While the Mamba architecture demonstrates superior inference efficiency and competitive performance on short-context natural language processing (NLP) tasks, empirical evidence sug…
Mixture-of-Modules: Reinventing Transformers as Dynamic Assemblies of Modules
Zhuocheng Gong, Ang Lv, Jian Guan +6
Is it always necessary to compute tokens from shallow to deep layers in Transformers? The continued success of vanilla Transformers and their variants suggests an undoubted "yes".…
Efficient Continual Pre-training by Mitigating the Stability Gap
Yiduo Guo, Jie Fu, Huishuai Zhang +2
Continual pre-training has increasingly become the predominant approach for adapting Large Language Models (LLMs) to new domains. This process involves updating the pre-trained LLM…
Automatic Jailbreaking of the Text-to-Image Generative AI Systems
Minseon Kim, Hyomin Lee, Boqing Gong +2
Recent AI systems have shown extremely powerful performance, even surpassing human performance, on various tasks such as information retrieval, language generation, and image gener…
xRAG: Extreme Context Compression for Retrieval-augmented Generation with One Token
Xin Cheng, Xun Wang, Xingxing Zhang +5
This paper introduces xRAG, an innovative context compression method tailored for retrieval-augmented generation. xRAG reinterprets document embeddings in dense retrieval--traditio…
©Plug-in Authorization for Human Content Copyright Protection in Text-to-Image Model
Chao Zhou, Huishuai Zhang, Jiang Bian +2
This paper addresses the contentious issue of copyright infringement in images generated by text-to-image models, sparking debates among AI developers, content creators, and legal…
On the Convergence of Adam under Non-uniform Smoothness: Separability from SGDM and Beyond
Bohan Wang, Huishuai Zhang, Qi Meng +3
This paper aims to clearly distinguish between Stochastic Gradient Descent with Momentum (SGDM) and Adam in terms of their convergence rates. We demonstrate that Adam achieves a fa…
Differentially Private Synthetic Data via Foundation Model APIs 2: Text
Chulin Xie, Zinan Lin, Arturs Backurs +9
Text data has become extremely valuable due to the emergence of machine learning algorithms that learn from it. A lot of high-quality text data generated in the real world is priva…
Exploring Transferability for Randomized Smoothing
Kai Qiu, Huishuai Zhang, Zhirong Wu +1
Training foundation models on extensive datasets and then finetuning them on specific tasks has emerged as the mainstream approach in artificial intelligence. However, the model ro…
FD-Align: Feature Discrimination Alignment for Fine-tuning Pre-Trained Models in Few-Shot Learning
Kun Song, Huimin Ma, Bochao Zou +2
Due to the limited availability of data, existing few-shot learning methods trained from scratch fail to achieve satisfactory performance. In contrast, large-scale pre-trained mode…
Gradient Descent with Polyak's Momentum Finds Flatter Minima via Large Catapults
Prin Phunyaphibarn, Junghyun Lee, Bohan Wang +2
Although gradient descent with Polyak's momentum is widely used in modern machine and deep learning, a concrete understanding of its effects on the training trajectory remains elus…
On the Generalization Properties of Diffusion Models
Puheng Li, Zhong Li, Huishuai Zhang +1
Diffusion models are a class of generative models that serve to establish a stochastic transport map between an empirically observed, yet unknown, target distribution and a known p…
Closing the Gap Between the Upper Bound and the Lower Bound of Adam's Iteration Complexity
Bohan Wang, Jingwen Fu, Huishuai Zhang +2
Recently, Arjevani et al. [1] established a lower bound of iteration complexity for the first-order optimization under an -smooth condition and a bounded noise variance assumpti…