Field robots operating in mines, tunnels, and disaster areas require reliable localization despite GNSS denial and degraded camera and LiDAR measurements caused by dust, smoke, or poor illumination. Missions involving both ground and aerial robots further motivate complementary sensing modalities that do not depend on visibility, preinstalled infrastructure, or wheel contact. This letter presents MISLAM, a tightly coupled magnetic-inertial simultaneous localization and mapping system using an IMU and a nine-magnetometer array, without relying on external odometry or platform-specific motion assumptions. The system combines the inertial navigation model with local polynomial magnetic-field models for short-term motion estimation and a global Gaussian process model for drift correction using an error-state Kalman filter. Offline evaluation on three indoor datasets collected with a handheld rig shows reductions in total endpoint closure error of 45.7% - 78.0% relative to magnetic-inertial odometry. Under poor illumination, MISLAM maintains tracking after visual-inertial odometry fails, achieving a horizontal closure error of 0.332 m, although vertical drift remains a limitation. These results support magnetic-inertial SLAM as a complementary localization modality for robotic sensor suites operating under degraded visual conditions.
Reinforcement learning (RL) has become a cornerstone for improving the reasoning capabilities of large language models (LLMs), but the need for on-policy data substantially limits training efficiency. Reusing off-policy data through importance sampling (IS) can improve efficiency but introduce considerable instability. Hence, algorithms such as PPO and GRPO widely adopt IS-ratio clipping to stabilize training. However, training stability and gradient estimate are mainly determined by the product of IS ratio and advantage. To further stabilize training, we propose ACPO, which clips the product of the IS ratio and the advantage, leading to more stable gradient estimates. We also establish a connection between ACPO and gradient clipping in policy mirror descent (PMD), which is a standard technique to stabilize optimization process, and prove the convergence of clipped-PMD under the standard RL setting. Experiments on widely used mathematical reasoning benchmarks show that ACPO consistently outperforms PPO and GRPO in both accuracy and training efficiency, delivering 4-6 percentage points gains on standard math benchmarks, with Qwen3-8B+PPO. Hence, ACPO is a practical and effective alternative to conventional IS-ratio clipping for RL post-training of LLMs.
Baichuan Huang, Ananth Balashankar Google DeepMind and Google, Amir Aminifar Lund University / Lund Institute of Technology
原文摘要
Fine-tuning the bias terms of large language models (LLMs) has the potential to achieve unprecedented parameter efficiency while maintaining competitive performance, particularly in low-data regimes. However, the link between fine-tuning different bias terms (i.e., \boldsymbol{b}_q, \boldsymbol{b}_k, \boldsymbol{b}_v in the query, key, or value projections) and downstream performance remains largely unclear to date. In this paper, we investigate the link between fine-tuning \boldsymbol{b}_q, \boldsymbol{b}_k, \boldsymbol{b}_v with the performance of the downstream task. Our key finding is that directly fine-tuning \boldsymbol{b}_v generally leads to higher downstream performance in low-data regimes, in comparison to \boldsymbol{b}_q and \boldsymbol{b}_k. We extensively evaluate this unique property across a wide range of LLMs spanning encoder-only and decoder-only architectures up to 6.7B parameters (including bias-free LLMs). Our results provide strong evidence for the effectiveness of directly fine-tuning \boldsymbol{b}_v across various downstream tasks. The implementation code is available at https://github.com/whubaichuan/BEFT.
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表