19 papers
UBTree: Parallel Tree Drafting via Unigram and Bigram Models for Speculative Decoding
Chumeng Liang, Linxuan Wang, Xinyu Peng +6
Speculative decoding accelerates language model inference by verifying multiple draft tokens in a single target-model pass. Recent parallel drafters have achieved breakthrough perf…
LLMRouter: Unified Infrastructure for Developing, Evaluating, and Deploying LLM Routers
Tao Feng, Fangxu Yu, Haozhen Zhang +9
No single large language model (LLM) is optimal across all queries and budget constraints, making model routing essential for cost-effective deployment. Existing routers adopt dive…
Reinforcement Learning with Evolving Rubrics as Rewards for Audio Reasoning
Fangxu Yu, Tao Feng, Dehai Min +6
Audio reasoning is essential for machine understanding of the acoustic world. Reinforcement learning with verifiable rewards can elicit such reasoning, yet existing reward designs…
SMART: LLM-Augmented Hybrid Retrieval for Dynamic Product Ads
Congfei Zhang, Jingxiao Ma, Xiaodong Liu +14
Dynamic Product Ads (DPA) require retrieving relevant items from multi-million product catalogs, balancing two competing objectives: retargeting (re-surfacing known interests) and…
TSRouter: Dynamic Modality-Model Selection for Time Series Reasoning
Fangxu Yu, Tao Feng, Dehai Min +3
Time series reasoning is essential for real-world problem-solving. While both Large Language Models (LLMs) and Vision-Language Models (VLMs) can reason about time-series data, thei…
EventConnector: Mining Social Event Relations through Temporal Graphs
Zijie Lei, Haofei Yu, Ge Liu +1
Understanding and retrieving related real-world events based on their temporal dynamics is a fundamental challenge in time-sensitive applications such as forecasting, information r…
ExpWeaver: LLM Agents Learn from Experience via Latent RAG
Tao Feng, Tianyang Luo, Jingjun Xu +5
Experience learning has achieved promising results in enhancing LLM agent planning and reasoning by integrating past interactions as reusable knowledge. However, existing methods r…
ExpHarness: Model-Agnostic Experience Learning through a Trainable Harness
Tao Feng, Chongrui Ye, Fangxu Yu +11
Large language model (LLM) agents increasingly operate within a harness, the scaffolding that determines what enters the executor's context, yet the experience they accumulate acro…
ElasticMem: Latent Memory as a Learnable Resource for LLM Agents
Tao Feng, Chongrui Ye, Fangxu Yu +8
Long-term memory is essential for LLM agents to reason coherently across extended interactions, personalize responses, and reuse past experience. However, existing memory-augmented…
LRanker: LLM Ranker for Massive Candidates
Tao Feng, Zijie Lei, Zhigang Hua +4
Large language models (LLMs) have recently shown strong potential for ranking by capturing semantic relevance and adapting across diverse domains, yet existing methods remain const…
RouteProfile: Graph-Based Profiling for Cold-Start LLM Routing
Jingjun Xu, Hongji Pu, Tao Feng +3
LLM routing is increasingly important for selecting suitable models under diverse user needs and deployment constraints, but its practical effectiveness depends on continual adapta…
Auto-Dreamer: Learning Offline Memory Consolidation for Language Agents
Chongrui Ye, Yuxiang Liu, Yu Wang +5
Language agents increasingly operate over streams of related tasks, yet existing memory systems struggle to convert accumulated experience into reusable knowledge. Retrieval-augmen…
LangFlow: Continuous Diffusion Rivals Discrete in Language Modeling
Yuxin Chen, Chumeng Liang, Hangke Sui +4
Continuous diffusion has been the foundation of high-fidelity, controllable, and few-step generation of many data modalities such as images. However, in language modeling, prior co…
Thought-Retriever: Don't Just Retrieve Raw Data, Retrieve Thoughts for Memory-Augmented Agentic Systems
Tao Feng, Pengrui Han, Guanyu Lin +2
Large language models (LLMs) have transformed AI research thanks to their powerful internal capabilities and knowledge. However, existing LLMs still fail to effectively incorporate…
MemReward: Graph-Based Experience Memory for LLM Reward Prediction with Limited Labels
Tianyang Luo, Tao Feng, Zhigang Hua +4
Reinforcement learning has emerged as a powerful paradigm for improving large language model (LLM) reasoning, where rollouts are sampled from the policy and reward signals computed…
Steer2Adapt: Dynamically Composing Steering Vectors Elicits Efficient Adaptation of LLMs
Pengrui Han, Xueqiang Xu, Keyang Xuan +12
Activation steering has emerged as a promising approach for efficiently adapting large language models (LLMs) to downstream behaviors. However, most existing steering methods rely…
Probing the Knowledge Boundary: An Interactive Agentic Framework for Deep Knowledge Extraction
Yuheng Yang, Siqi Zhu, Tao Feng +2
Large Language Models (LLMs) can be seen as compressed knowledge bases, but it remains unclear what knowledge they truly contain and how far their knowledge boundary extends. Exist…
UniRec: Unified Multimodal Encoding for LLM-Based Recommendations
Zijie Lei, Tao Feng, Zhigang Hua +5
Large language models have recently shown promise for multimodal recommendation, particularly with text and image inputs. Yet real-world recommendation signals extend far beyond th…
Paper Copilot: A Self-Evolving and Efficient LLM System for Personalized Academic Assistance
Guanyu Lin, Tao Feng, Pengrui Han +2
As scientific research proliferates, researchers face the daunting task of navigating and reading vast amounts of literature. Existing solutions, such as document QA, fail to provi…