The paper introduces the Offline AI Modules workstream, which provides a voice‑first offline architecture, a low‑cost hardware reference bill of materials, and a reproducible quantization and benchmarking pipeline for instruction‑tuned language models in the 2‑5B parameter range. It evaluates three models across four quantization formats on two hardware tiers—NVIDIA Jetson Orin NX and Raspberry Pi5—measuring deployment metrics and multilingual quality. The key result is that Q4_K_M quantization offers the best size‑to‑quality trade‑off, enabling high decode throughput and strong topic classification accuracy while staying within memory limits on both tiers.
By Sunday Afariogun, Odunolaoluwa Jenrola, Zeinab Nezami
The paper introduces two closed‑form projection operators that enforce physical validity in biomolecular diffusion models during inference. One operator resolves inter‑chain van der Waals clashes, while the other restores ligand geometry, including bond lengths, angles, planarity, and chirality. These lightweight, model‑agnostic corrections preserve structural accuracy and ligand placement across multiple benchmarks without adding runtime or memory overhead.
By Qurat-ul-ain, Yee Whye Teh, Charlotte M. Deane, Matteo Cagiada
The paper introduces Diffusion On-Policy Context Distillation (D-OPCD), a method that internalizes the benefits of an agentic harness—such as memory, skills, and iterative prompt refinement—into the weights of a diffusion model. By treating the agent‑improved prompt as privileged context, D-OPCD distills harness knowledge into the model, raising direct‑generation scores from 60.52 to 65.09 across four benchmarks. This internalization allows the harness to shed saturated skills and continue evolving, enabling a second round of skill evolution that further improves performance by 1.83 points.
By Wenxuan Wang, Zekai Liu, Weinan Zhang, Yu Cheng, Yang Yang
The paper introduces NP‑Bench, a benchmark and a proactive scheduling planner that coordinates parallel large‑language‑model coding agents. By partitioning work scopes and ordering merges ahead of time, the planner improves clean‑integration rates from 1/9 to 9/9 and eliminates merge conflicts, outperforming both no‑coordination and reactive‑detection baselines. It also demonstrates that cross‑session memory can eliminate repeated mistakes and that routing facts to agents does not improve long‑context accuracy at scale.
By Sumanyu Muku
The paper presents a retrieval‑admissibility verification framework for long‑term memory agents, classifying each memory‑query pair as admissible, inadmissible, or unresolved. It evaluates the framework on public benchmarks (RHELM and MemOps), showing improved anchor recall and reduced exact similarity errors, while also revealing that existing verifiers miss certain inadmissible exposures. The study highlights the need for separate checks on candidate support, admissibility, prompt exposure, and answer disclosure to ensure safe memory retrieval.
By Zi Wang, Xingqiao Wang, Emmanuel Addai, Devika Ambekar, Xiaowei Xu
MemCo introduces a memory‑centric collaboration framework that equips large language model agents with both local and global memory spaces. Local memory preserves environment‑specific details, while global memory captures transferable workflows from local trajectories. During online interaction, MemCo selectively routes relevant memories based on the agent’s state and decision phase, enabling reuse of other agents’ experiences without over‑generalizing environment‑specific information.
By Xinting Liao, Siyan Liu, Rabab K. Ward, Holger R. Roth, Xiaoxiao Li
The paper introduces COMPASS, an inference-time steering technique that enhances large language model performance on mathematical reasoning tasks. By using the correctness of a model’s direct answer attempts as a latent signal, COMPASS identifies a small subset of attention heads whose activations can be steered to elicit reasoning. Experiments across three model families and multiple math benchmarks show that COMPASS improves GSM8K accuracy by an average of 16 percentage points and achieves near Chain‑of‑Thought performance while generating 20–70% fewer tokens.
By Pratyay Dutta, Kowshik Thopalli, Vivek Narayanaswamy
arXiv:2610.08101v1 Announce Type: new
Abstract: Shared memory coordinates agents' actions, but correct records do not establish that those actions satisfy task requirements. Memory governance and fai...
By Zhe Yu, Zixuan Wang, Peidong Wang, Hehai Lin, Ruochen Zhao, Chengwei Qin
arXiv:2610.08258v1 Announce Type: new
Abstract: Auditing the claimed outcomes of large language model (LLM) training is challenging when model weights and training data are private, while cryptograph...
By Junkai Liang, Zhanpeng Guo, Pengfei Wu, Qingni Shen, Jiaheng Zhang, Zhonghai Wu, Haiyang Xue, Shengfang Zhai
arXiv:2610.08327v1 Announce Type: new
Abstract: Large language models (LLMs) have shown promise in medical question answering and clinical reasoning, yet their improvement remains constrained by stat...
By Xilin Dang, Weilin Ruan, Xue Yang, Jinghao Wang, Xiaowei Hu, Jinpeng Li, Pheng-Ann Heng
arXiv:2610.07043v1 Announce Type: cross
Abstract: Low-precision execution can substantially accelerate reinforcement learning (RL) for large language models, but discrepancies between learner and sam...
By Zhen Li, Shuai Zhang, Yanggan Gu, Yiming Zhang, Yang Yu, Mingfa Feng, Congkai Xie, Shuang Yu, Junjie Lai, Hongxia Yang
arXiv:2610.08563v1 Announce Type: new
Abstract: Sequence-level power sampling has recently emerged as a training-free approach to reasoning by sampling from a sharpened output distribution of a base...
By Bingnan Xiao, Chenhao Yang, Bingcong Li, Wei Ni, Xin Wang
arXiv:2610.08691v1 Announce Type: new
Abstract: Large language model agents are accelerating scientific automation, yet verified executions rarely become persistent program-level improvements, and ex...
By Mingda Zhang, Wenjin Liu, Tiesunlong Shen, Zikai Xiao, Zhenghong Lin, Qing Xu, Erik Cambria, Xiaoying Tang, Haoran Luo
arXiv:2610.06994v1 Announce Type: cross
Abstract: Backdoor-defense leaderboards print a clean-accuracy drop and read it as removal cost. Measured on the poisoned victim alone, the drop cannot separat...
By Ruizhi Xu, Wei Xu, Sibo Zhu
arXiv:2610.07032v1 Announce Type: cross
Abstract: Large-scale pretrained transformer models have achieved state-of-the-art performance across diverse machine translation tasks, including multilingual...
By Maria Zafar, Souhail Bakkali, Rejwanul Haque
arXiv:2610.07125v1 Announce Type: cross
Abstract: While open-weight models have enjoyed steady progress in capabilities and wide adoption across multiple domains, their safety remains an important co...
By Abhinav Sudhakar Dubey (University of California Santa Cruz), Scott Sirri (University of California Santa Cruz), Vaggos Chatziafratis (University of California Santa Cruz), C. Seshadhri (University of California Santa Cruz)
arXiv:2610.07217v1 Announce Type: cross
Abstract: Despite its promise for scaling robot learning, egocentric manipulation data is still scarce today. Collection at scale requires vertically integrati...
By Grounded Superintelligence, BitRobot
arXiv:2610.07338v1 Announce Type: cross
Abstract: Audio benchmarks are built around short, pre-segmented clips, limiting model design to brief inputs or fixed vocabularies. To close this gap, we intr...
By Kwanghee Choi, Suwon Shon, Dmitriy Serdyuk, Guitang Lan, Chao-Wei Huang, Mohammad Sadegh Rasooli, Sangeeta Srivastava, Zhaojiang Lin, Saurabh Adya, Ming Sun
arXiv:2610.07348v1 Announce Type: cross
Abstract: Training large language models (LLMs) is resource-intensive, and adapting them for diverse deployment scenarios with varying computational constraint...
By Arnav Kundu, Zhaoyang Xu, Bairu Hou, Chang Gao, Reed Li, Tao Lei
arXiv:2610.07470v1 Announce Type: cross
Abstract: Combinatorial Thompson sampling (CTS) draws independent posterior samples for every arm, so its exploration dynamics ignore any relation among arms....
By Vikram Kakaria, Anish Kataria, Anany Kotawala