arXiv subjects

Bolun Wang

Publications and source records attributed to Bolun Wang.

At least 19 recordsLinked to original sources

arXiv2026

Unified Lookup-Table Inference with Signed-Digit K/V Caches for Ternary LLMs

Ternary LLMs make their weight-dominated projections compact and efficient, but attention remains a mismatch: its K/V cache is created online and is typically processed by a separate higher-precision engine. Compressing this cache alone does not resolve the mismatch. To execute attention with the same lookup-table machinery as ternary projections, values accumulated in one reduction must retain a compatible representation and scale. This requirement also differs for keys and values during causal decoding, because newly generated values may belong to an unfinished cache block. This work develops a unified lookup-table inference approach for ternary LLMs. It stores runtime K/V states as scaled multi-plane signed digits organized around the reduction structure of attention. The resulting digit planes are consumed directly by activation-derived tables, avoiding dense K/V materialization between cache storage and attention computation. The design combines online K/V formation, bounded handling of incomplete value blocks, and a shared multi-stream datapath for Linear projections and attention. A constraint-guided search selects the representation and execution policy for a target quality--efficiency trade-off. Experiments on native and post-training ternary models validate the approach across cache capacity, model quality, and hardware efficiency.

cs.AR↗

arXiv2026

Nexusformer: Nonlinear Attention Expansion for Stable and Inheritable Transformer Scaling

Scaling Transformers typically necessitates training larger models from scratch, as standard architectures struggle to expand without discarding learned representations. We identify the primary bottleneck in the attention mechanism's linear projections, which strictly confine feature extraction to fixed-dimensional subspaces, limiting both expressivity and incremental capacity. To address this, we introduce Nexusformer, which replaces linear $Q/K/V$ projections with a Nexus-Rank layer, a three-stage nonlinear mapping driven by dual activations in progressively higher dimensional spaces. This design overcomes the linearity constraint and enables lossless structured growth: new capacity can be injected along two axes via zero-initialized blocks that preserve pretrained knowledge. Experiments on language modeling and reasoning benchmarks demonstrate that Nexusformer matches Tokenformer's perplexity using up to 41.5\% less training compute during progressive scaling (240M to 440M). Furthermore, our analysis of growth dynamics reveals that zero initialization induces a stable convergence trajectory, allowing us to derive a geometric scaling law that accurately predicts performance across expansion scales.

cs.LG↗

arXiv2026

InternVLA-A1: Unifying Understanding, Generation and Action for Robotic Manipulation

Prevalent Vision-Language-Action (VLA) models are typically built upon Multimodal Large Language Models (MLLMs) and demonstrate exceptional proficiency in semantic understanding, but they inherently lack the capability to deduce physical world dynamics. Consequently, recent approaches have shifted toward World Models, typically formulated via video prediction; however, these methods often suffer from a lack of semantic grounding and exhibit brittleness in the presence of video prediction errors. To synergize semantic understanding with dynamic predictive capabilities, we present InternVLA-A1. This model employs a unified Mixture-of-Transformers architecture, coordinating three experts for scene understanding, visual foresight generation, and action execution. These components interact seamlessly through a unified masked self attention mechanism. Building upon InternVL3 and Qwen3-VL, we instantiate InternVLA-A1 at 2B and 3B parameter scales. We pre-train these models on heterogeneous data sources over real-world robot data, synthetic simulation data, and human videos, covering over 692M frames. This hybrid training strategy effectively harnesses the diversity of synthetic simulation data while minimizing the sim-to-real gap. We evaluated InternVLA-A1 on 12 real-world robotic tasks and a simulation benchmark. The results show that InternVLA-A1 consistently outperforms prior leading models: compared with pi0.5, it achieves +4.4\% on static manipulation tasks and +2.6\% on the RoboTwin 2.0 simulation benchmark, and delivers a +26.7\% boost on dynamic manipulation tasks.

cs.RO↗

arXiv2025

InternVLA-M1: A Spatially Guided Vision-Language-Action Framework for Generalist Robot Policy

We introduce InternVLA-M1, a unified framework for spatial grounding and robot control that advances instruction-following robots toward scalable, general-purpose intelligence. Its core idea is spatially guided vision-language-action training, where spatial grounding serves as the critical link between instructions and robot actions. InternVLA-M1 employs a two-stage pipeline: (i) spatial grounding pre-training on over 2.3M spatial reasoning data to determine ``where to act'' by aligning instructions with visual, embodiment-agnostic positions, and (ii) spatially guided action post-training to decide ``how to act'' by generating embodiment-aware actions through plug-and-play spatial prompting. This spatially guided training recipe yields consistent gains: InternVLA-M1 outperforms its variant without spatial guidance by +14.6% on SimplerEnv Google Robot, +17% on WidowX, and +4.3% on LIBERO Franka, while demonstrating stronger spatial reasoning capability in box, point, and trace prediction. To further scale instruction following, we built a simulation engine to collect 244K generalizable pick-and-place episodes, enabling a 6.2% average improvement across 200 tasks and 3K+ objects. In real-world clustered pick-and-place, InternVLA-M1 improved by 7.3%, and with synthetic co-training, achieved +20.6% on unseen objects and novel configurations. Moreover, in long-horizon reasoning-intensive scenarios, it surpassed existing works by over 10%. These results highlight spatially guided training as a unifying principle for scalable and resilient generalist robots. Code and models are available at https://github.com/InternRobotics/InternVLA-M1.

cs.RO↗

arXiv2025

Unlocking Euclidean Problems with Isotropic Initialization

Many problems in Euclidean geometry, arising in computational design and fabrication, amount to a system of constraints, which is challenging to solve. We suggest a new general approach to the solution, which is to start with analogous problems in isotropic geometry. Isotropic geometry can be viewed as a structure-preserving simplification of Euclidean geometry. The solutions found in the isotropic case give insight and can initialize optimization algorithms to solve the original Euclidean problems. We illustrate this general approach with three examples: quad-mesh mechanisms, composite asymptotic-geodesic gridshells, and asymptotic gridshells with constant node angle.

cs.CG↗

arXiv2025

Re$^3$Sim: Generating High-Fidelity Simulation Data via 3D-Photorealistic Real-to-Sim for Robotic Manipulation

Real-world data collection for robotics is costly and resource-intensive, requiring skilled operators and expensive hardware. Simulations offer a scalable alternative but often fail to achieve sim-to-real generalization due to geometric and visual gaps. To address these challenges, we propose a 3D-photorealistic real-to-sim system, namely, RE$^3$SIM, addressing geometric and visual sim-to-real gaps. RE$^3$SIM employs advanced 3D reconstruction and neural rendering techniques to faithfully recreate real-world scenarios, enabling real-time rendering of simulated cross-view cameras within a physics-based simulator. By utilizing privileged information to collect expert demonstrations efficiently in simulation, and train robot policies with imitation learning, we validate the effectiveness of the real-to-sim-to-real pipeline across various manipulation task scenarios. Notably, with only simulated data, we can achieve zero-shot sim-to-real transfer with an average success rate exceeding 58%. To push the limit of real-to-sim, we further generate a large-scale simulation dataset, demonstrating how a robust policy can be built from simulation data that generalizes across various objects. Codes and demos are available at: https://re3sim.github.io/.

cs.RO↗

arXiv2024

Gated Slot Attention for Efficient Linear-Time Sequence Modeling

Linear attention Transformers and their gated variants, celebrated for enabling parallel training and efficient recurrent inference, still fall short in recall-intensive tasks compared to traditional Transformers and demand significant resources for training from scratch. This paper introduces Gated Slot Attention (GSA), which enhances Attention with Bounded-memory-Control (ABC) by incorporating a gating mechanism inspired by Gated Linear Attention (GLA). Essentially, GSA comprises a two-layer GLA linked via $\operatorname{softmax}$, utilizing context-aware memory reading and adaptive forgetting to improve memory capacity while maintaining compact recurrent state size. This design greatly enhances both training and inference efficiency through GLA's hardware-efficient training algorithm and reduced state size. Additionally, retaining the $\operatorname{softmax}$ operation is particularly beneficial in "finetuning pretrained Transformers to RNNs" (T2R) settings, reducing the need for extensive training from scratch. Extensive experiments confirm GSA's superior performance in scenarios requiring in-context recall and in T2R settings.

cs.CL↗

arXiv2023

RWKV: Reinventing RNNs for the Transformer Era

Transformers have revolutionized almost all natural language processing (NLP) tasks but suffer from memory and computational complexity that scales quadratically with sequence length. In contrast, recurrent neural networks (RNNs) exhibit linear scaling in memory and computational requirements but struggle to match the same performance as Transformers due to limitations in parallelization and scalability. We propose a novel model architecture, Receptance Weighted Key Value (RWKV), that combines the efficient parallelizable training of transformers with the efficient inference of RNNs. Our approach leverages a linear attention mechanism and allows us to formulate the model as either a Transformer or an RNN, thus parallelizing computations during training and maintains constant computational and memory complexity during inference. We scale our models as large as 14 billion parameters, by far the largest dense RNN ever trained, and find RWKV performs on par with similarly sized Transformers, suggesting future work can leverage this architecture to create more efficient models. This work presents a significant step towards reconciling trade-offs between computational efficiency and model performance in sequence processing tasks.

cs.CL↗

arXiv2021

Time of Impact Dataset for Continuous Collision Detection and a Scalable Conservative Algorithm

We introduce a large-scale benchmark for broad- and narrow-phase continuous collision detection (CCD) over linearized trajectories with exact time of impacts and use it to evaluate the accuracy, correctness, and efficiency of 13 state-of-the-art CCD algorithms. Our analysis shows that several methods exhibit problems either in efficiency or accuracy. To overcome these limitations, we introduce an algorithm for CCD designed to be scalable on modern parallel architectures and provably correct when implemented using floating point arithmetic. We integrate our algorithm within the Incremental Potential Contact solver [24] and evaluate its impact on various simulation scenarios. Our approach includes a broad-phase CCD to quickly filter out primitives having disjoint bounding boxes and a narrow-phase CCD that establishes whether the remaining primitive pairs indeed collide. Our broad-phase algorithm is efficient and scalable thanks to the experimental observation that sweeping along a coordinate axis performs surprisingly well on modern parallel architectures. For narrow-phase CCD, we re-design the recently proposed interval-based algorithm of Wang et al. [45] to work on massively parallel hardware. To foster the adoption and development of future linear CCD algorithms, and to evaluate their correctness, scalability, and overall performance, we release the dataset with analytic ground truth, the implementation of all the algorithms tested, and our testing framework.

cs.GR↗

arXiv2021

Key-Point Interpolation: A Sparse Data Interpolation Algorithm based on B-splines

B-splines are widely used in the fields of reverse engineering and computer-aided design, due to their superior properties. Traditional B-spline surface interpolation algorithms usually assume regularity of the data distribution. In this paper, we introduce a novel B-spline surface interpolation algorithm: KPI, which can interpolate sparsely and non-uniformly distributed data points. As a two-stage algorithm, our method generates the dataset out of the sparse data using Kriging, and uses the proposed KPI (Key-Point Interpolation) method to generate the control points. Our algorithm can be extended to higher dimensional data interpolation, such as reconstructing dynamic surfaces. We apply the method to interpolating the temperature of Shanxi Province. The generated dynamic surface accurately interpolates the temperature data provided by the weather stations, and the preserved dynamic characteristics can be useful for meteorology studies.

cs.CG↗

arXiv2021

A Large Scale Benchmark and an Inclusion-Based Algorithm for Continuous Collision Detection

We introduce a large scale benchmark for continuous collision detection (CCD) algorithms, composed of queries manually constructed to highlight challenging degenerate cases and automatically generated using existing simulators to cover common cases. We use the benchmark to evaluate the accuracy, correctness, and efficiency of state-of-the-art continuous collision detection algorithms, both with and without minimal separation. We discover that, despite the widespread use of CCD algorithms, existing algorithms are either: (1) correct but impractically slow, (2) efficient but incorrect, introducing false negatives which will lead to interpenetration, or (3) correct but over conservative, reporting a large number of false positives which might lead to inaccuracies when integrated in a simulator. By combining the seminal interval root finding algorithm introduced by Snyder in 1992 with modern predicate design techniques, we propose a simple and efficient CCD algorithm. This algorithm is competitive with state of the art methods in terms of runtime while conservatively reporting the time of impact and allowing explicit trade off between runtime efficiency and number of false positives reported.

cs.GR↗

arXiv2021

A novel feed rate scheduling method based on Sigmoid function with chord error and kinematics constraints

In high speed CNC (Compute Numerical Control) machining, the feed rate scheduling has played an important role to ensure machining quality and machining efficiency. In this paper, a novel feed rate scheduling method is proposed for generating smooth feed rate profile conveniently with the consideration of both geometric error and kinematic error. First, a relationship between feed rate value and chord error is applied to determine the feed rate curve. Then, breaking points, which can split whole curve into several blocks, can be found out using proposed two step screening method. For every block, a feed rate profile based on the Sigmoid function is generated. With the consideration of kinematic limitation and machining efficiency, a time-optimal feed rate adjustment algorithm is proposed to further adjust feed rate value at breaking points. After planning feed rate profile for each block, all blocks feed rate profile will be connected smoothly. The resulting feed rate profile is more concise compared with the polynomial profile and more efficient compared with the trigonometric profile. Finally, simulations with two free-form NURBS curves are conducted and comparison with the sine-curve method are carried out to verify the feasibility and applicability of the proposed method.

eess.SY↗

arXiv2021

A novel S-shape based NURBS interpolation with acc-jerk- Continuity and round-off error elimination

Feedrate scheduling is a key step in computer numerical control (CNC) machining, as it has a close relationship with machining time and surface quality, and has now become a hot issue in industry and academia. To reduce high chord errors and round-off errors, and generate continuous velocity, acceleration, and jerk profile of parametric interpolation, a novel and complete S-shape based feedrate scheduling algorithm is presented in this paper. The algorithm consists of three modules: bidirectional scanning module, velocity scheduling module and round-off error elimination module. The bidirectional scanning module with the limitations of chord error, normal acceleration/jerk and command feedrate aims to guarantee the continuity of the feed rate at the junctions between successive NURBS sub-curves. After the NURBS sub-curves have been classified into two cases, the velocity scheduling module firstly calculates the actual maximum federate, and then generates the feed rate profiles of all NURBS sub-curves according to our velocity scheduling function. Later, the round-off error elimination module is proposed to make the total interpolating time become an integer multiple of the interpolation period, which leads to the elimination of round-off errors. Finally, benchmarks are conducted to verify the applicability of the proposed method compared with some other methods.

eess.SY↗

arXiv2021

Deepfake Videos in the Wild: Analysis and Detection

AI-manipulated videos, commonly known as deepfakes, are an emerging problem. Recently, researchers in academia and industry have contributed several (self-created) benchmark deepfake datasets, and deepfake detection algorithms. However, little effort has gone towards understanding deepfake videos in the wild, leading to a limited understanding of the real-world applicability of research contributions in this space. Even if detection schemes are shown to perform well on existing datasets, it is unclear how well the methods generalize to real-world deepfakes. To bridge this gap in knowledge, we make the following contributions: First, we collect and present the largest dataset of deepfake videos in the wild, containing 1,869 videos from YouTube and Bilibili, and extract over 4.8M frames of content. Second, we present a comprehensive analysis of the growth patterns, popularity, creators, manipulation strategies, and production methods of deepfake content in the real-world. Third, we systematically evaluate existing defenses using our new dataset, and observe that they are not ready for deployment in the real-world. Fourth, we explore the potential for transfer learning schemes and competition-winning techniques to improve defenses.

cs.CR↗

arXiv2020

Gotta Catch 'Em All: Using Honeypots to Catch Adversarial Attacks on Neural Networks

Deep neural networks (DNN) are known to be vulnerable to adversarial attacks. Numerous efforts either try to patch weaknesses in trained models, or try to make it difficult or costly to compute adversarial examples that exploit them. In our work, we explore a new "honeypot" approach to protect DNN models. We intentionally inject trapdoors, honeypot weaknesses in the classification manifold that attract attackers searching for adversarial examples. Attackers' optimization algorithms gravitate towards trapdoors, leading them to produce attacks similar to trapdoors in the feature space. Our defense then identifies attacks by comparing neuron activation signatures of inputs to those of trapdoors. In this paper, we introduce trapdoors and describe an implementation of a trapdoor-enabled defense. First, we analytically prove that trapdoors shape the computation of adversarial attacks so that attack inputs will have feature representations very similar to those of trapdoors. Second, we experimentally show that trapdoor-protected models can detect, with high accuracy, adversarial examples generated by state-of-the-art attacks (PGD, optimization-based CW, Elastic Net, BPDA), with negligible impact on normal classification. These results generalize across classification domains, including image, facial, and traffic-sign recognition. We also present significant results measuring trapdoors' robustness against customized adaptive attacks (countermeasures).

cs.LG↗

arXiv2020

Fast Tetrahedral Meshing in the Wild

We propose a new tetrahedral meshing method, fTetWild, to convert triangle soups into high-quality tetrahedral meshes. Our method builds on the TetWild algorithm, replacing the rational triangle insertion with a new incremental approach to construct and optimize the output mesh, interleaving triangle insertion and mesh optimization. Our approach makes it possible to maintain a valid floating-point tetrahedral mesh at all algorithmic stages, eliminating the need for costly constructions with rational numbers used by TetWild, while maintaining full robustness and similar output quality. This allows us to improve on TetWild in two ways. First, our algorithm is significantly faster, with running time comparable to less robust Delaunay-based tetrahedralization algorithms. Second, our algorithm is guaranteed to produce a valid tetrahedral mesh with floating-point vertex coordinates, while TetWild produces a valid mesh with rational coordinates which is not guaranteed to be valid after floating-point conversion. As a trade-off, our algorithm no longer guarantees that all input triangles are present in the output mesh, but in practice, as confirmed by our tests on the Thingi10k dataset, the algorithm always succeeds in inserting all input triangles.

cs.GR↗

arXiv2018

Wireless Side-Lobe Eavesdropping Attacks

Millimeter-wave wireless networks offer high throughput and can (ideally) prevent eavesdropping attacks using narrow, directional beams. Unfortunately, imperfections in physical hardware mean today's antenna arrays all exhibit side lobes, signals that carry the same sensitive data as the main lobe. Our work presents results of the first experimental study of the security properties of mmWave transmissions against side-lobe eavesdropping attacks. We show that these attacks on mmWave links are highly effective in both indoor and outdoor settings, and they cannot be eliminated by improved hardware or currently proposed defenses.

cs.CR↗

arXiv2016

Defending against Sybil Devices in Crowdsourced Mapping Services

Real-time crowdsourced maps such as Waze provide timely updates on traffic, congestion, accidents and points of interest. In this paper, we demonstrate how lack of strong location authentication allows creation of software-based {\em Sybil devices} that expose crowdsourced map systems to a variety of security and privacy attacks. Our experiments show that a single Sybil device with limited resources can cause havoc on Waze, reporting false congestion and accidents and automatically rerouting user traffic. More importantly, we describe techniques to generate Sybil devices at scale, creating armies of virtual vehicles capable of remotely tracking precise movements for large user populations while avoiding detection. We propose a new approach to defend against Sybil devices based on {\em co-location edges}, authenticated records that attest to the one-time physical co-location of a pair of devices. Over time, co-location edges combine to form large {\em proximity graphs} that attest to physical interactions between devices, allowing scalable detection of virtual vehicles. We demonstrate the efficacy of this approach using large-scale simulations, and discuss how they can be used to dramatically reduce the impact of attacks against crowdsourced mapping services.

cs.SI↗