The paper investigates reducing memory usage of n‑gram embedding tables in open‑weight language models by storing them in low precision. Experiments on nanochat, Llama, and GPT‑2 families show that four‑bit tables cut table memory by about 75% while retaining over 99% of the validation‑loss improvement. The study also demonstrates that re‑allocating the saved memory to larger tables can surpass the performance of full‑precision tables, and that similar savings are achievable in large models like DeepSeek‑V4.1‑Flash and Qwen3.8‑Flash‑Next with minimal impact on perplexity and downstream tasks.
By Stephen Chung, Wenyu Du
The paper introduces Federated Graph Retrieval-Augmented Generation (FedGRAG), a framework that extends Graph Retrieval-Augmented Generation to federated settings where knowledge graphs are distributed across multiple clients. FedGRAG tackles the lack of a global graph by using a Federated Knowledge Routing mechanism that identifies relevant clients via lightweight indices, and it protects sensitive local graph data through a controlled-disclosure evidence construction method that abstracts retrieved subgraphs into evidence abstracts for server-side reconstruction. Experiments show that FedGRAG outperforms existing baselines in both retrieval and downstream generation tasks, establishing a new paradigm for deploying GraphRAG in federated environments.
By Xiaobao Wang, Renda Han, Zhiliang Xia, Meng Liu, Renxiang Guan, Di Jin
StyleStream 2.0 is a fast, controllable streaming voice style conversion system that reduces end‑to‑end latency to about 520 ms on a consumer GPU, halving the delay of its predecessor while preserving conversion quality. It replaces mel‑spectrogram context with a compact style embedding, enabling a unified text‑conditioned flow model that supports both text‑based voice design and instruction‑based voice editing. Experiments demonstrate that StyleStream 2.0 achieves superior target style fidelity, voice design, and voice editing performance while maintaining intelligibility and attribute preservation.
By Yisi Liu, Nicholas Lee, Gopala K. Anumanchipalli
The paper presents a lane‑change prediction model using knowledge‑graph embeddings and Bayesian inference to anticipate surrounding vehicles’ intentions 3–4 seconds before they occur. Three driving strategies—reactive, prediction‑based longitudinal, and prediction‑based lateral—are compared in CARLA simulations of risky merging scenarios. Prediction‑based approaches avoid collisions, reduce peak deceleration by up to 91 %, and are evaluated for ride comfort using ISO 2631‑1 vibration estimates.
By Mohamed Manzour, Aditya Kumar, Rub\'en Izquierdo, Miguel \'Angel Sotelo
The paper introduces a deep learning framework for regression with random objects in metric spaces, bypassing restrictive manifold or Euclidean embedding assumptions. It provides statistical consistency, convergence rates, and conditions that mitigate the curse of dimensionality for hierarchically compositional regression functions. Extensive simulations and two digital health case studies with distribution-valued outcomes demonstrate competitive predictive performance over existing methods.
By Marcos Matabuena, Xiaokai Luo, Oscar Hernan Madrid Padilla
The paper introduces Self-Assessment Reinforcement Learning (SARL), a framework that trains large language models to recognize their own search state during retrieval-augmented tasks. SARL uses a dual reward system to balance corrective assessment and efficiency, achieving significant performance gains over strong baselines on multi‑hop reasoning and deep research benchmarks. It also demonstrates improved training stability for longer search horizons, outperforming methods that rely on auxiliary supervision or multi‑agent coordination.
By Taewhoo Lee, Minju Song, Jungwoo Park, Chanwoong Yoon, Junha Jung, Hyeon Hwang, Sanghoon Lee, Minbyul Jeong, Jaewoo Kang
The study investigates whether speakers of different languages organize concepts similarly, a core question of the Sapir–Whorf hypothesis. Using word embeddings from two multilingual models, the authors compared relations among 196 basic concepts across 81 languages and found that cross‑language correspondence is stronger for sociocultural meanings than for nature and motion. They also observed that relational differences within the sociocultural domain grow with geographic separation, indicating that universality and relativity are intertwined across conceptual domains.
By Weiran Wang, Wenjuan Qin, Haitao Liu
arXiv:2610.31282v1 Announce Type: new
Abstract: Training-free personalization adapts a frozen vision-language model (VLM) to recognize user-defined concepts from only a few exemplars by retrieving co...
By Changseon Yu, Hyeokjun Kweon
arXiv:2610.28686v1 Announce Type: new
Abstract: Query clarification is essential for retrieval-augmented generation (RAG) systems that face ambiguous user queries. We consider multi-turn grounded con...
By Lihong He, Lucian Popa, Marina Danilevsky
arXiv:2610.30344v1 Announce Type: cross
Abstract: Training Arabic large language models faces challenges from noisy web data, dialectal variation, and the high cost of continual pretraining (CPT). We...
By Houssam Eddine Boukhalfa, Mourad Bouache, Mourad Dahmane
arXiv:2610.30941v1 Announce Type: cross
Abstract: General-purpose embeddings trained on profile similarity often misrank candidates in professional networks, where relevance is inherently complementa...
By Zhongkai Shangguan, Yanbin Kang, Yihan Cao, Ruijie Jiang, Qiannan Yin, Chang Yue, Zijian Zhou, Girish Kathalagiri, Changshuai Wei, Zhengming Xing
arXiv:2610.30705v1 Announce Type: new
Abstract: Text embedding and text generation are two of the most useful capabilities of modern large language models, yet they are typically provided by separate...
By Yang Liu, Zhiqiang Yuan, Shu-Dong Huang, Jiancheng Lv, Liwei Yang, Qiwei Yan, Kailin Lv, Jinchao Zhang
arXiv:2507.00401v2 Announce Type: replace-cross
Abstract: We investigate cross-domain few-shot learning under the constraint that fine-tuning of backbones (i.e., feature extractors) is impossible or...
By Xin Xu, Eibe Frank, Geoffrey Holmes
arXiv:2610.31467v1 Announce Type: new
Abstract: As wearable and digital health systems increasingly generate longitudinal physiological records, communicating these observations in patient-accessible...
By Bayram Ayadi, Florian Geissler, Johanna Schmidhuber
arXiv:2510.18795v4 Announce Type: replace
Abstract: The original CLIP text encoder is limited by a maximum input length of 77 tokens, which hampers its ability to effectively process long texts and p...
By Xiaoxing Hu, Kaicheng Yang, Ziyang Gong, Qi Ming, Zonghao Guo, Yu Tian, Xiang An, Ziyong Feng, Xue Yang
arXiv:2610.30114v1 Announce Type: cross
Abstract: Consensus is a fundamental problem in distributed computing and at the core of state machine replication (SMR). Despite decades of active human resea...
By Cl\'ement Burgelin, Rachid Guerraoui, Zicong Hong, Yinan Hu, Anne-Marie Kermarrec, Rafael Pires, Rasmus Moorits Veski, Martijn de Vos
arXiv:2610.30847v1 Announce Type: new
Abstract: Trained network weights are increasingly treated as data, but there is rarely ground truth about what a population of models should encode, which hampe...
By Tobias G\"obel, Miranda C. N. Cheng
arXiv:2610.31194v1 Announce Type: new
Abstract: Joint-embedding predictive architectures (JEPAs) learn world models in latent space, enabling efficient planning but leaving their states opaque. We in...
By Hossein Resani, Javen Qinfeng Shi
arXiv:2610.31309v1 Announce Type: new
Abstract: Scaling embedding capacity alongside Transformer computation is an emerging trend in language models. This trend makes high compression rates increasin...
By Jiacheng Liu, Yaxin Luo, Xiaohan Zhao, Peng Sun, Xinyi Shang, Jiacheng Cui, Yonina C. Eldar, Zhiqiang Shen
NEST is an unsupervised framework that learns a single embedding encoding coarse‑to‑fine cluster structure, optimizing clustering and separation across nested prefix subspaces. The method uses rank and within‑cluster regularization to prevent collapse, and distances in the embedding align more closely with the true hierarchy than the input distances, even though the hierarchy is not provided during training. Experiments on synthetic, vision, text, and single‑cell data show that NEST can recover multiscale structure given only the number of clusters per level, demonstrating its ability to discover multiscale organization in real‑world representations.
By Zhanqi Zhang, Mikio Aoi, Gal Mishne