Parliamentary speech generation presents specific challenges for large language models beyond standard text generation tasks. Unlike general text generation, parliamentary speeches require not only linguistic quality but also political authenticity and ideological consistency. Current language models lack specialized training for parliamentary contexts, and existing evaluation methods focus on standard NLP metrics rather than political authenticity. To address this, we present ParliaBench, a benchmark for parliamentary speech generation. We constructed a dataset of speeches from UK Parliament to enable systematic model training. We introduce an evaluation framework combining computational metrics with LLM-as-a-judge assessments for measuring generation quality across three dimensions: linguistic quality, semantic coherence, and political authenticity. We propose two novel embedding-based metrics, Political Spectrum Alignment and Party Alignment, to quantify ideological positioning. We fine-tuned five large language models (LLMs), generated 28k speeches, and evaluated them using our framework, comparing baseline and fine-tuned models. Results show that fine-tuning produces statistically significant improvements across the majority of metrics and our novel metrics demonstrate strong discriminative power for political dimensions.

We introduce LOREN, a curvature-aware zeroth-order (ZO) optimization method for fine-tuning large language models (LLMs). Existing ZO methods, which estimate gradients via finite differences using random perturbations, often suffer from high variance and suboptimal search directions. Our approach addresses these challenges by: (i) reformulating the problem of gradient preconditioning as that of adaptively estimating an anisotropic perturbation distribution for gradient estimation, (ii) capturing curvature through a low-rank block diagonal preconditioner using the framework of natural evolution strategies, and (iii) applying a REINFORCE leave-one-out (RLOO) gradient estimator to reduce variance. Experiments on standard LLM benchmarks show that our method outperforms state-of-the-art ZO methods by achieving higher accuracy and faster convergence, while cutting peak memory usage by up to 27.3% compared with MeZO-Adam.

Second-order optimization methods for training neural networks, such as KFAC, exhibit superior convergence by utilizing curvature information of loss landscape. However, it comes at the expense of high computational burden. In this work, we analyze the two components that constitute the layer-wise Fisher information matrix (FIM) used in KFAC: the Kronecker factors related to activations and pre-activation gradients. Based on empirical observations on their eigenspectra, we propose efficient approximations for them, resulting in a computationally efficient optimization method called MAC. To the best of our knowledge, MAC is the first algorithm to apply the Kronecker factorization to the FIM of attention layers used in transformers and explicitly integrate attention scores into the preconditioning. We also study the convergence property of MAC on nonlinear neural networks and provide two conditions under which it converges to global minima. Our extensive evaluations on various network architectures and datasets show that the proposed method outperforms KFAC and other state-of-the-art methods in terms of accuracy, end-to-end training time, and memory usage.

Backpropagation (BP) remains the dominant and most successful method for training parameters of deep neural network models. However, BP relies on two computationally distinct phases, does not provide a satisfactory explanation of biological learning, and can be challenging to apply for training of networks with discontinuities or noisy node dynamics. By comparison, node perturbation (NP), also known as activity-perturbed forward gradients, proposes learning by the injection of noise into network activations, and subsequent measurement of the induced loss change. NP relies on two forward (inference) passes, does not make use of network derivatives, and has been proposed as a model for learning in biological systems. However, standard NP is highly data inefficient and can be unstable due to its unguided noise-based search process. In this work, we develop a modern perspective on NP by relating it to the directional derivative and incorporating input decorrelation. We find that a closer alignment with directional derivatives together with input decorrelation at every layer theoretically and practically enhances performance of NP learning with large improvements in parameter convergence and much higher performance on the test data, approaching that of BP. Furthermore, our novel formulation allows for application to noisy systems in which the noise process itself is inaccessible, which is of particular interest for on-chip learning in neuromorphic systems.

Action chunking is a widely adopted approach in Learning from Demonstration (LfD). By modeling multi-step action chunks rather than single-step actions, action chunking significantly enhances modeling capabilities for human expert policies. However, because action chunking makes a single decision only after a complete action block has been executed, the resulting reduction in decision frequency restricts the utilization of real-time observations, impairing reactivity in dynamic or noisy environments. Existing efforts to address this issue have primarily resorted to trading off reactivity against decision consistency, without achieving both. To address this limitation, we propose a novel algorithm, Temporal Action Selection (TAS), which caches predicted action chunks from multiple timesteps and dynamically selects the optimal action through a lightweight selector network. TAS achieves balanced optimization across both reactivity and decision consistency. Experiments across multiple tasks with diverse base policy architectures show that TAS significantly improves success rates. Furthermore, integrating TAS as a base policy with residual reinforcement learning (RL) improves both training efficiency and the performance ceiling. Experiments in both simulation and physical robots confirm the method's efficacy.

The adoption of agroecological practices in modern agriculture requires robotic systems capable of operating in highly diverse and complex field environments. Developing and evaluating such systems relies heavily on simulation, yet generating realistic and configurable 3D environments representative of agroecological diversity remains a major challenge. This paper presents CropCraft, an open-source procedural world generator built on Blender and Python, designed to produce 3D simulation environments tailored to agricultural robotics. CropCraft generates crop fields from a simple YAML configuration file, supporting a wide range of scenarios including intercropping, vineyards, and weed-infested fields. The tool includes a library of 3D plant models (crops, grasses, and weeds) at multiple growth stages, and uses stochastic placement algorithms to realistically reproduce the spatial variability observed in real fields. Generated worlds are directly importable into the Gazebo simulator and include ground-truth annotations for all placed elements, supporting both perception and navigation algorithm development. To demonstrate the practical utility of CropCraft, we apply it to the task of crop-weed semantic segmentation using deep learning. A dataset of 10,000 synthetic images of maize fields with varying weed densities, growth stages, and lighting conditions was generated and used to train several segmentation architectures. Models trained exclusively on synthetic data achieve a sim-to-real gap of approximately 10% mean Intersection over Union (mIoU) on real field images, outperforming previous state-of-the-art synthetic generation approaches. We further show that combining even a few real images with synthetic data improves generalization across domains, providing new insights into the effective use of synthetic data for agricultural perception tasks.

While several high profile video games have served as testbeds for Deep Reinforcement Learning (DRL), this technique has rarely been employed by the game industry for crafting authentic AI behaviors. Previous research focuses on training super-human agents with large models, which is impractical for game studios with limited resources aiming for human-like agents. This paper proposes a sample-efficient DRL method tailored for training and fine-tuning agents in industrial settings such as the video game industry. Our method improves sample efficiency of value-based DRL by leveraging pre-collected data and increasing network plasticity. We evaluate our method training a goalkeeper agent in EA SPORTS FC 25, one of the best-selling football simulations today. Our agent outperforms the game's built-in AI by 10% in ball saving rate. Ablation studies show that our method trains agents 50% faster compared to standard DRL methods. Finally, qualitative evaluation from domain experts indicates that our approach creates more human-like gameplay compared to hand-crafted agents. As a testament to the impact of the approach, the method has been adopted for use in the most recent release of the series.

Self-supervised pre-training on unlabeled graph data has become a common paradigm for Graph Neural Networks (GNNs). However, an objective gap often remains between pre-training objectives and downstream tasks. To bridge this gap, graph prompting methods adapt frozen pre-trained GNNs to specific downstream tasks through learnable prompts. Despite its effectiveness, most existing graph prompting methods primarily focus on improving model performance and largely overlook fairness concerns. As downstream graph data inherently contains biases in both node attributes and graph structures, pre-trained GNNs may produce representations that differ across demographic subgroups. To address this limitation, we propose Adaptive Dual Prompting (ADPrompt), a fairness-aware graph prompting framework for adapting pre-trained GNNs. ADPrompt incorporates two complementary components: Adaptive Feature Rectification, which learns personalized attribute prompts to suppress sensitive information at the input level, and Adaptive Message Calibration, which introduces layer-wise structure prompts to dynamically regulate information propagation from neighboring nodes. By jointly optimizing these two modules, ADPrompt adapts the pre-trained GNN while mitigating both attribute-level and structural bias. Experiments on four benchmark datasets with multiple pre-training strategies demonstrate that ADPrompt consistently outperforms seven competitive baselines in node classification tasks.

Multi-hop reasoning for question answering (QA) plays a critical role in retrieval-augmented generation (RAG) for modern large language models (LLMs). The accurate answer can be obtained through retrieving relational structure of entities from knowledge graph (KG). Regarding the inherent relation-dependency and reasoning pattern, multi-hop reasoning can be in general classified into two categories: i) parallel fact-verification multi-hop reasoning question, i.e., requiring simultaneous verifications of multiple independent sub-questions; and ii) chained multi-hop reasoning questions, i.e., demanding sequential multi-step inference with intermediate conclusions serving as essential premises for subsequent reasoning. Currently, the multi-hop reasoning approaches singly employ one of two techniques: LLM response-based fact verification and KG path-based chain construction. Nevertheless, the former excels at parallel fact-verification but underperforms on chained reasoning tasks, while the latter demonstrates proficiency in chained multi-hop reasoning but suffers from redundant path retrieval when handling parallel fact-verification reasoning. These limitations deteriorate the efficiency and accuracy for multi-hop QA tasks. To address this challenge, we propose a novel dual-track KG verification and reasoning framework DTKG, which is inspired by the Dual Process Theory in cognitive science. Specifically, DTKG comprises two main stages: the Classification Stage and the Branch Processing Stage.

Personalized large language models (LLMs) tailor content to individual preferences using user profiles or histories. However, existing parameter-efficient fine-tuning (PEFT) methods, such as the ``One-PEFT-Per-User'' (OPPU) paradigm, require training a separate adapter for each user, making them computationally expensive and impractical for real-time updates. We introduce Profile-to-PEFT, a scalable framework that employs a hypernetwork, trained end-to-end, to map a user's encoded profile directly to a full set of adapter parameters (e.g., LoRA), eliminating per-user training at deployment. This design enables instant adaptation, generalization to unseen users, and privacy-preserving local deployment. Experimental results demonstrate that our method outperforms both prompt-based personalization and OPPU while using substantially fewer computational resources at deployment. The framework exhibits strong generalization to out-of-distribution users and maintains robustness across varying user activity levels and different embedding backbones. The proposed Profile-to-PEFT framework enables efficient, scalable, and adaptive LLM personalization suitable for large-scale applications.

Autonomous vehicles (AVs) offer a cost-effective solution for scientific missions such as underwater tracking. Reinforcement learning (RL) has emerged as a powerful method for controlling AVs, but scaling to fleets (essential for multi-target tracking or rapidly moving targets) is challenging. Multi-Agent RL (MARL) is notoriously sample-inefficient, and while high-fidelity simulators like Gazebo's LRAUV provide up to 100x faster-than-real-time single-robot simulations, they offer little speedup in multi-vehicle scenarios, making MARL training impractical. Yet, high-fidelity simulation is crucial to test complex policies and close the sim-to-real gap. To address these limitations, we develop a GPU-accelerated environment that achieves up to 30,000x speedup over Gazebo while preserving its dynamics. This enables fast, end-to-end GPU training and seamless transfer to Gazebo for evaluation. We also introduce a Transformer-based architecture (TransfMAPPO) that learns policies invariant to fleet size and number of targets, enabling curriculum learning to train larger fleets on increasingly complex scenarios. After large-scale GPU training, we perform extensive evaluations in Gazebo, showing our method maintains tracking errors below 5m even with multiple fast-moving targets.

Depth estimation remains central to autonomous driving, and radar-camera fusion offers robustness in adverse conditions by providing complementary geometric cues. In this paper, we present XD-RCDepth, a lightweight architecture that reduces the parameters by 29.7% relative to the state-of-the-art lightweight baseline while maintaining comparable accuracy. To preserve performance under compression and enhance interpretability, we introduce two knowledge-distillation strategies: an explainability-aligned distillation that transfers the teacher's saliency structure to the student, and a depth-distribution distillation that recasts depth regression as soft classification over discretized bins. Together, these components reduce the MAE compared with direct training with 7.97% and deliver competitive accuracy with real-time efficiency on nuScenes and ZJU-4DRadarCam datasets. Code: https://github.com/harborsarah/XD_RCDepth

We consider the task of predicting Hamiltonian matrices to accelerate electronic structure calculations, which plays an important role in physics, chemistry, and materials science. Motivated by the inherent relationship between the off-diagonal blocks of the Hamiltonian matrix and the SO(2) local frame, we propose a novel and efficient network, called QHNetV2, that achieves global SO(3) equivariance without the costly SO(3) Clebsch-Gordan tensor products. This is achieved by introducing a set of new efficient and powerful SO(2)-equivariant operations and performing all off-diagonal feature updates and message passing within SO(2) local frames, thereby eliminating the need of SO(3) tensor products. Moreover, a continuous SO(2) tensor product is performed within the SO(2) local frame at each node to fuse node features, mimicking the symmetric contraction operation. Extensive experiments on the large QH9 and MD17 datasets demonstrate that our model achieves superior performance across a wide range of molecular structures and trajectories, highlighting its strong generalization capability. The proposed SO(2) operations on SO(2) local frames offer a promising direction for scalable and symmetry-aware learning of electronic structures. Our code will be released as part of the AIRS library https://github.com/divelab/AIRS.

Large Language Models (LLMs) have recently emerged as a powerful paradigm for Knowledge Graph Completion (KGC), offering strong reasoning and generalization capabilities beyond traditional embedding-based approaches. However, existing LLM-based methods often struggle to fully exploit structured semantic representations, as the continuous embedding space of pretrained KG models is fundamentally misaligned with the discrete token space of LLMs. This discrepancy hinders effective semantic transfer and limits their performance. To address this challenge, we propose ReaLM, a novel and effective framework that bridges the gap between KG embeddings and LLM tokenization through the mechanism of residual vector quantization. ReaLM discretizes pretrained KG embeddings into compact code sequences and integrates them as learnable tokens within the LLM vocabulary, enabling seamless fusion of symbolic and contextual knowledge. Furthermore, we incorporate ontology-guided class constraints to enforce semantic consistency, refining entity predictions based on class-level compatibility. Extensive experiments on two widely used benchmark datasets demonstrate that ReaLM achieves state-of-the-art performance, confirming its effectiveness in aligning structured knowledge with large-scale language models.

Reinforcement learning with verifiable rewards (RLVR) efficiently scales the reasoning ability of large language models (LLMs) but is bottlenecked by scarce labeled data. Reinforcement learning with intrinsic rewards (RLIR) offers a scalable alternative via self-rewarding, yet often suffers from instability and inferior performance. We trace this gap to a systemic bias in confidence-coupled self-rewarding: the model tends to over-reward high-confidence mistakes, forming a self-confirming loop. We quantify this feedback-loop bias with three metrics: reward noise magnitude (rho_noise), policy-reward coupling (rho_selfbias), and over-/under-reward skew (rho_symbias). Our analyses show a compounding effect where strong coupling amplifies confidence-conditioned errors and drives a drift toward over-reward, leading to instability and a lower performance ceiling. To mitigate this, we propose reinforcement learning with ensembled rewards (RLER), which aggregates diverse models with adaptive reward interpolation and disagreement-aware rollout selection to reduce coupling and suppress over-reward drift. Extensive experiments show that RLER improves by 6.2% over the best RLIR baseline and is within 3.6% of RLVR, while exhibiting stable scaling on unlabeled samples.

Earth Observation Foundation Models (EOFMs) have exploded in prevalence as tools for processing the massive volumes of remotely sensed and other earth observation data, and for delivering impact on the many essential earth monitoring tasks. An emerging trend posits using the outputs of pre-trained models as 'embeddings' which summarize high dimensional data to be used for generic tasks such as similarity search and content-specific queries. However, most EOFM models are trained only on single modalities of data and then applied or benchmarked by matching bands across different modalities. It is not clear from existing work what impact diverse sensor architectures have on the internal representations of the present suite of EOFMs. We show in this work that the representation space of EOFMs is highly sensitive to sensor architecture and that understanding this difference gives a vital perspective on the pitfalls of current EOFM design and signals for how to move forward as model developers, users, and a community guided by robust remote-sensing science.

Alignment of large language models remains a central challenge in natural language processing. Preference optimization has emerged as a popular and effective method for improving alignment, typically through training-time or prompt-based interventions. In this paper, we introduce alignment-aware decoding (AAD), a method to enhance model alignment directly at inference. Theoretically, AAD can be interpreted as implicit reward optimization, yet it requires no specialized training beyond the standard DPO setup. Empirically, AAD consistently outperforms strong baselines across diverse alignment benchmarks and model scales. Moreover, in data-constrained settings, AAD can produce high-quality synthetic data to improve alignment under standard decoding, providing a practical solution when labeled data is limited.

Implicit in-context learning (ICL) has newly emerged as a promising paradigm that simulates ICL behaviors in the representation space of large language models (LLMs), aiming to attain few-shot performance at zero-shot cost. However, existing approaches largely rely on injecting shift vectors into residual flows, which are typically constructed from labeled demonstrations or task-specific alignment. Such designs fall short of utilizing the structural mechanisms underlying ICL and suffer from limited generalizability. To address this, we propose In-Context Routing (ICR), a novel implicit ICL method that captures and utilizes generalizable ICL patterns at the attention logits level. It extracts reusable structural directions that emerge during ICL and employs a learnable input-conditioned router to modulate attention logits accordingly, enabling an efficient train-once-and-reuse framework. We evaluate ICR on 12 real-world datasets spanning diverse domains and multiple LLMs. The results show that ICR consistently outperforms existing implicit ICL methods that require task-specific retrieval or training, while demonstrating robust generalization to out-of-domain tasks where they struggle. These findings position ICR to push the boundary of the practical value of ICL. The code is available at https://github.com/Lijiaqian1/In-Context-Routing.git.

High-quality molecular representations are essential for property prediction and molecular design, yet large labeled datasets remain scarce. While self-supervised pretraining on molecular graphs has shown promise, many existing approaches either depend on hand-crafted augmentations or complex generative objectives, and often rely solely on 2D topology, leaving valuable 3D structural information underutilized. To address this gap, we introduce C-FREE (Contrast-Free Representation learning on Ego-nets), a simple framework that integrates 2D graphs with ensembles of 3D conformers. C-FREE learns molecular representations by predicting subgraph embeddings from their complementary neighborhoods in the latent space, using fixed-radius ego-nets as modeling units across different conformers. This design allows us to integrate both geometric and topological information within a hybrid Graph Neural Network (GNN)-Transformer backbone, without negatives, positional encodings, or expensive pre-processing. Pretraining on the GEOM dataset, which provides rich 3D conformational diversity, C-FREE achieves state-of-the-art results on MoleculeNet, surpassing contrastive, generative, and other multimodal self-supervised methods. Fine-tuning across datasets with diverse sizes and molecule types further demonstrates that pretraining transfers effectively to new chemical domains, highlighting the importance of 3D-informed molecular representations.

Safe and feasible trajectory planning is critical for real-world autonomous driving systems. However, existing learning-based planners rely heavily on expert demonstrations, which not only lack explicit safety awareness but also risk inheriting undesirable behaviors such as speeding from suboptimal human driving data. Inspired by the success of large language models, we propose Plan-R1, a two-stage trajectory planning framework that decouples principle alignment from behavior learning. In the first stage, a general trajectory predictor is pre-trained on expert data to capture diverse, human-like driving behaviors. In the second stage, the model is fine-tuned with rule-based rewards using Group Relative Policy Optimization (GRPO), explicitly aligning ego planning with principles such as safety, comfort, and traffic rule compliance. This two-stage paradigm retains human-like behaviors while enhancing safety awareness and discarding undesirable patterns from demonstrations. Furthermore, we identify a key limitation of directly applying GRPO to planning: group-wise normalization erases cross-group scale differences, causing rare, high-variance safety-violation groups to have similar advantages as abundant low-variance safe groups, thereby suppressing optimization for safety-critical objectives. To address this, we propose Variance-Decoupled GRPO (VD-GRPO), which replaces normalization with centering and fixed scaling to preserve absolute reward magnitudes, ensuring that safety-critical objectives remain dominant throughout training. Experiments on the nuPlan benchmark demonstrate that Plan-R1 significantly improves planning safety and feasibility, achieving state-of-the-art performance, particularly in realistic reactive settings. Our code is available at https://github.com/XiaolongTang23/Plan-R1.

Predicting properties of crystals from their structures is a fundamental yet challenging task in materials science. Unlike molecules, crystal structures exhibit infinite periodic arrangements of atoms, requiring methods capable of capturing both local and global information effectively. However, current works fall short of capturing long-range interactions within periodic structures. To address this, we leverage \emph{reciprocal space}, the natural domain for periodic crystals, and construct a Fourier series representation from fractional coordinates and reciprocal lattice vectors with learnable filters. Building on this, we introduce the reciprocal space-based geometry network (\textbf{ReciNet}), a novel architecture that integrates geometric GNNs and reciprocal blocks to model short-range and long-range interactions. Experiments on comprehensive benchmarks JARVIS, Materials Project, and MatBench demonstrate that ReciNet achieves outstanding predictive accuracy across a range of crystal property prediction tasks. Additionally, we explore a model extension for multi-property prediction with the mixture-of-experts, which demonstrates high computational efficiency and reveals positive transfer between correlated properties. These findings highlight the potential of our model as a scalable and accurate solution for crystal property prediction.

Reinforcement learning has enabled significant progress in complex domains such as coordinating and navigating multiple quadrotors. However, even well-trained policies remain vulnerable to collisions in obstacle-rich environments. Addressing these infrequent but critical safety failures through retraining or fine-tuning is costly and risks degrading previously learned skills. Inspired by activation steering in large language models and latent editing in computer vision, we introduce a framework for inference-time Latent Activation Editing (LAE) that refines the behavior of pre-trained policies without modifying their weights or architecture. The framework operates in two stages: (i) an online classifier monitors intermediate activations to detect states associated with undesired behaviors, and (ii) an activation editing module that selectively modifies flagged activations to shift the policy towards safer regimes. In this work, we focus on improving safety in multi-quadrotor navigation. We hypothesize that amplifying a policy's internal perception of risk can induce safer behaviors. We instantiate this idea through a latent collision world model trained to predict future pre-collision activations, thereby prompting earlier and more cautious avoidance responses. Extensive simulations and real-world Crazyflie experiments demonstrate that LAE achieves statistically significant reduction in collisions (nearly 90% fewer cumulative collisions compared to the unedited baseline) and substantially increases the fraction of collision-free trajectories, while preserving task completion. More broadly, our results establish LAE as a lightweight paradigm, feasible on resource-constrained hardware, for post-deployment refinement of learned robot policies.

Diffusion probability models have shown significant promise in offline reinforcement learning by directly modeling trajectory sequences. However, existing approaches primarily focus on time-domain features while overlooking frequency-domain features, leading to frequency shift and degraded performance according to our observation. In this paper, we investigate the RL problem from a new perspective of the frequency domain. We first observe that time-domain-only approaches inadvertently introduce shifts in the low-frequency components of the frequency domain, which results in trajectory instability and degraded performance. To address this issue, we propose Wavelet Fourier Diffuser (WFDiffuser), a novel diffusion-based RL framework that integrates Discrete Wavelet Transform to decompose trajectories into low- and high-frequency components. To further enhance diffusion modeling for each component, WFDiffuser employs Short-Time Fourier Transform and cross attention mechanisms to extract frequency-domain features and facilitate cross-frequency interaction. Extensive experiment results on the D4RL benchmark demonstrate that WFDiffuser effectively mitigates frequency shift, leading to smoother, more stable trajectories and improved decision-making performance over existing methods.

Millimeter-wave radar provides robust perception in fog, smoke, dust, and low light, making it attractive for size-, weight-, and power-constrained robotic platforms. Existing radar imaging methods typically rely on synthetic aperture or multi-frame aggregation to improve resolution, which is impractical for small aerial, inspection, or wearable systems. We present RadarSFD, a conditional latent diffusion framework that reconstructs dense LiDAR-like point clouds from a single radar frame without motion or SAR. Our approach transfers geometric priors from a pretrained monocular depth estimator into the diffusion backbone, anchors them to radar inputs via channel-wise latent concatenation, and regularizes outputs with a dual-space objective combining latent and pixel-space losses. On the RadarHD benchmark, RadarSFD achieves state-of-the-art performance against baseline models. Qualitative results show recovery of fine walls and narrow gaps, and experiments across new environments confirm strong generalization. Ablation studies highlight the importance of pretrained initialization, radar BEV conditioning, and the dual-space loss. Together, these results establish a practical single-frame, no-SAR mmWave radar pipeline for dense point cloud perception in compact robotic systems.

Scale-consistent ego-motion estimation is fundamental for autonomous ground robots. Bird's-Eye-View (BEV) representation naturally addresses the scale drift problem of monocular visual odometry (MVO) by providing a metric-scaled planar workspace, enabling the simplification of 6-DoF ego-motion to a more robust 3-DoF model. However, existing BEV-based methods suffer from two key limitations: sparse supervision signals from pose-only training, and information loss during perspective-to-BEV projection. We present BEV-ODOM2, an enhanced framework that addresses both limitations without requiring additional annotations. Our approach introduces (1) dense BEV optical flow supervision constructed directly from 3-DoF pose ground truth for pixel-level guidance, and (2) Perspective View (PV)-BEV fusion that computes correlation volumes before projection to preserve 6-DoF motion cues. An enhanced rotation sampling strategy further balances diverse motion patterns during training. We evaluate on four datasets with varied spatial scales: KITTI, Oxford, NCLT, and our newly collected ZJH-VO benchmark. BEV-ODOM2 achieves a 40\% RTE improvement over prior BEV-based methods, with real-time inference on an NVIDIA Jetson AGX Orin confirming edge deployment feasibility. The source code and the ZJH-VO dataset are publicly released to facilitate future research.

Earlier Sign Language Production (SLP) models typically relied on autoregressive decoding, which naturally preserves temporal causality but suffers from error accumulation at inference time. More recent diffusion-based approaches improve generation quality through iterative denoising, yet their sequence-level refinement process introduces substantial latency. To address this trade-off, we propose HybridSign, a hybrid autoregressive-diffusion model for low-latency sign language production that combines causal frame generation with flow-based diffusion refinement. A Multi-Scale Pose Representation module captures fine-grained articulator features, while a Confidence-Aware Causal Attention mechanism leverages joint-level confidence scores to improve robustness under noisy 2D pose observations. Experiments on PHOENIX14T and How2Sign show that HybridSign consistently achieves the best quality--efficiency trade-off among the compared baselines. On the How2Sign test split, it reaches BLEU-1/4 scores of 30.12/6.48 and DTW of 3.89, while reducing time-to-first-frame to 5.90s and increasing throughput to 10.17 FPS under a 60-frame evaluation protocol.

Large language models (LLMs) have demonstrated impressive capabilities across a wide range of applications, but demand substantial memory and compute resources during inference. Existing quantization methods expose a trade-off between efficiency and accuracy: weight-only quantization (WOQ) incurs costly dequantization overheads, while integer weight-and-activation quantization (INT-WAQ) reduces precision and degrades model quality. Non-uniform weight-and-activation quantization (NU-WAQ) can better capture the non-uniform distributions of LLM weights and activations, yet remains incompatible with conventional low-precision compute units. This paper presents OASIS, a lookup table (LUT)-based architecture that enables efficient general matrix multiplication (GEMM) between non-uniformly quantized weights and activations without requiring dequantization. OASIS employs pre-computed Cartesian Product LUTs, achieving a 64x reduction in LUT size and enabling a 1024x higher computational parallelism over existing LUT-based GEMM methods. To preserve accuracy under aggressive activation quantization, OASIS introduces an outlier-aware quantization scheme with concurrent LUT-based GEMM and error compensation for outliers. Furthermore, we design Orizuru, an efficient top-k detection engine for real-time activation outlier identification. According to extensive evaluations, OASIS incurs an average accuracy drop of only 1.98% compared to the FP16 baseline, which is 5.18% lower than Atom. On the hardware side, OASIS achieves an average 3.00x speedup and a 1.44x energy efficiency improvement compared to the FIGLUT accelerator.

Hyperspectral unmixing (HU) targets to decompose each mixed pixel in remote sensing images into a set of endmembers and their corresponding abundances. Despite significant progress in this field using deep learning, most methods fail to simultaneously characterize global dependencies and local consistency, making it difficult to preserve both long-range interactions and boundary details. This letter proposes a novel transformer-guided content-adaptive graph unmixing framework (T-CAGU), which overcomes these challenges by employing a transformer to capture global dependencies and introducing a content-adaptive graph neural network to enhance local relationships. Unlike previous work, T-CAGU integrates multiple propagation orders to dynamically learn the graph structure, ensuring robustness against noise. Furthermore, T-CAGU leverages a graph residual mechanism to preserve global information and stabilize training. Experimental results demonstrate its superiority over the state-of-the-art methods. Our code is available at https://github.com/xianchaoxiu/T-CAGU.

Formula alpha mining, which generates predictive signals from financial data, is critical for quantitative investment. Although various algorithmic approaches-such as genetic programming, reinforcement learning, and large language models-have significantly expanded the capacity for alpha discovery, systematic evaluation remains a key challenge. Existing evaluation metrics predominantly include backtesting and correlation-based measures. Backtesting is computationally intensive, inherently sequential, and sensitive to specific strategy parameters. Correlation-based metrics, though efficient, assess only predictive ability and overlook other crucial properties such as temporal stability, robustness, diversity, and interpretability. Additionally, the closed-source nature of most existing alpha mining models hinders reproducibility and slows progress in this field. To address these issues, we propose AlphaEval, a unified, parallelizable, and backtest-free evaluation framework for automated alpha mining models. AlphaEval assesses the overall quality of generated alphas along five complementary dimensions: predictive power, stability, robustness to market perturbations, financial logic, and diversity. Extensive experiments across representative alpha mining algorithms demonstrate that AlphaEval achieves evaluation consistency comparable to comprehensive backtesting, while providing more comprehensive insights and higher efficiency. Furthermore, AlphaEval effectively identifies superior alphas compared to traditional single-metric screening approaches. All implementations and evaluation tools are open-sourced to promote reproducibility and community engagement.

\textbf{BEAVR} is an open-source, bimanual, multi-embodiment Virtual Reality (VR) teleoperation system for robots, designed to unify real-time control, data recording, and policy learning across heterogeneous robotic platforms. BEAVR enables real-time, dexterous teleoperation using commodity VR hardware, supports modular integration with robots ranging from 7-DoF manipulators to full-body humanoids, and records synchronized multi-modal demonstrations directly in the LeRobot dataset schema. Our system features a zero-copy streaming architecture achieving $\leq$35\,ms latency, an asynchronous ``think--act'' control loop for scalable inference, and a flexible network API optimized for real-time, multi-robot operation. We benchmark BEAVR across diverse manipulation tasks and demonstrate its compatibility with leading visuomotor policies such as ACT, DiffusionPolicy, and SmolVLA. All code is publicly available, and datasets are released on Hugging Face\footnote{Code, datasets, and VR app available at https://github.com/ARCLab-MIT/BEAVR-Bot.