Vision-Language Models (VLMs) offer promising capabilities for mobile devices, but their deployment faces significant challenges due to computational limitations and energy inefficiency, especially for real-time applications. This study provides a comprehensive survey of deployment frameworks for VLMs on mobile devices, evaluating llama.cpp, MLC-Imp, and mllm in the context of running LLaVA-1.5 7B, MobileVLM-3B, and Imp-v1.5 3B as representative workloads on a OnePlus 13R. Each deployment framework was evaluated on the OnePlus 13R while running VLMs, with measurements covering CPU, GPU, and NPU utilization, temperature, inference time, power consumption, and user experience. Benchmarking revealed critical performance bottlenecks across frameworks: CPU resources were consistently over-utilized during token generation, while GPU and NPU accelerators were largely unused. When the GPU was used, primarily for image feature extraction, it was saturated, leading to degraded device responsiveness. The study contributes framework-level benchmarks, practical profiling tools, and an in-depth analysis of hardware utilization bottlenecks, highlighting the consistent overuse of CPUs and the inef
The original OnePlus phone disrupted the mobile market 13 years ago with its affordable price and top-end specs。 Now, the company is exiting North America and Europe to focus on China
Low-precision neural network training has emerged as a promising direction for reducing computational costs and democratizing access to deep learning research. However, existing 4-bit quantization methods either rely on expensive GPU infrastructure or suffer from significant accuracy degradation. In this work, we present a practical method for training convolutional neural networks at true 4-bit precision using standard PyTorch operations on commodity CPUs. We introduce a novel tanh-based soft weight clipping technique that, combined with symmetric quantization, dynamic per-layer scaling, and straight-through estimators, achieves stable convergence and competitive accuracy. Training a VGG-style architecture with 3.25 million parameters from scratch on CIFAR-10, our method achieves 92.34% test accuracy on Google Colab's free CPU tier -- matching full-precision baseline performance (92.5%) with only a 0.16% gap. We further validate on CIFAR-100, achieving 70.94% test accuracy across 100 classes with the same architecture and training procedure, demonstrating that 4-bit training from scratch generalizes to harder classification tasks. Both experiments achieve 8x memory compression ove
Quantization is an effective technique to reduce the deployment cost of large language models (LLMs), and post-training quantization (PTQ) has been widely studied due to its efficiency. However, existing PTQ methods are limited by their inability to fine-tune model parameters and often suffer significant accuracy loss in low-bit scenarios. Quantization-aware training (QAT) provides a more principled solution, but its reliance on backpropagation incurs prohibitive memory costs, limiting its practicality for LLM deployment. To address these challenges, we propose ZeroQAT, a zeroth-order optimization-based QAT framework that supports both weight and activation quantization. ZeroQAT leverages forward-only gradient estimation to eliminate backpropagation, substantially reducing computational and memory overhead while retaining the benefits of end-to-end optimization. We further introduce a lightweight variant of ZeroQAT for quantized fine-tuning, which freezes and pre-quantizes most parameters to further cut memory usage. Experiments show that ZeroQAT consistently outperforms representative PTQ and QAT baselines while requiring significantly less memory. For example, ZeroQAT enables fin
Our research focuses on the analysis and improvement of the Graph-based Relation Inference Transformer (GRIT), which serves as an important benchmark in the field. We conduct a comprehensive ablation study using the PISC-fine dataset, to find and explore improvement in efficiency and performance of GRITv2. Our research has provided a new state-of-the-art relation recognition model on the PISC relation dataset. We introduce several features in the GRIT model and analyse our new benchmarks in two versions: GRITv2-L (large) and GRITv2-S (small). Our proposed GRITv2-L surpasses existing methods on relation recognition and the GRITv2-S is within 2% performance gap of GRITv2-L, which has only 0.0625x the model size and parameters of GRITv2-L. Furthermore, we also address the need for model compression, an area crucial for deploying efficient models on resource-constrained platforms. By applying quantization techniques, we efficiently reduced the GRITv2-S size to 22MB and deployed it on the flagship OnePlus 12 mobile which still surpasses the PISC-fine benchmarks in performance, highlighting the practical viability and improved efficiency of our model on mobile devices.
Filesystem vulnerabilities persist as a significant threat to Android systems, despite various proposed defenses and testing techniques. The complexity of program behaviors and access control mechanisms in Android systems makes it challenging to effectively identify these vulnerabilities. In this paper, we present PathSentinel, which overcomes the limitations of previous techniques by combining static program analysis and access control policy analysis to detect three types of filesystem vulnerabilities: path traversals, hijacking vulnerabilities, and luring vulnerabilities. By unifying program and access control policy analysis, PathSentinel identifies attack surfaces accurately and prunes many impractical attacks to generate input payloads for vulnerability testing. To streamline vulnerability validation, PathSentinel leverages large language models (LLMs) to generate targeted exploit code based on the identified vulnerabilities and generated input payloads. The LLMs serve as a tool to reduce the engineering effort required for writing test applications, demonstrating the potential of combining static analysis with LLMs to enhance the efficiency of exploit generation and vulnerab
This study focuses on assessing smartphone camera characteristics for developing an economic smartphone-based Particle Image Velocimetry (PIV) system. In the investigation, flow around a cylinder was visualized using two commercially-available smartphones (OnePlus 5T and iPhone X) cameras and low-intensity laser diodes. Hydrogen bubbles generated from electrolysis (termed Bubble Image Velocimetry) of aluminum electrodes were used as seeding medium. OpenPIV, an open-source toolbox, was used for processing captured images and obtaining the flow fields. A parametric analysis of the two smartphones was conducted across varying camera characteristics such as ISO, exposure compensation and frame rate. The results obtained through experimentation were compared with the results of a validated computational fluid dynamics (CFD) study with the same flow conditions and were found to be in good agreement, with deviation ranging from 1% to 3.5% for iPhone X and 1% to 7% for OnePlus 5T. It was observed that a higher frame rate results in greater accuracy of the measurement. Further, an exposure compensation of -1 EV and an ISO of 400 was found to produce results with the least error as compared
A new study suggests the brain begins making decisions much earlier than scientists previously thought。 Researchers found that even primary sensory regions are influenced by higher brain areas through rapid feedback loops, rather than simply passing information forward。 This more dynamic view of brain function could help engineers design future AI
Scientists have developed a new framework that could finally apply the laws of thermodynamics to real, ever-changing black holes instead of only perfectly stable ones。 The advance may improve our understanding of black hole mergers, evaporation, and the powerful gravitational wave events detected by observatories like LIGO
Hubble has captured a spectacular view of LH 95, where about 2,500 young stars are still on their journey to becoming full-fledged stars。 Scientists discovered these growing stars can keep pulling in gas and dust for millions of years, extending an important stage of stellar development。 The region also contains multiple generations of stars living
"I pretty much, at that point in time, gave up on being an astronaut
Scientists at Nanyang Technological University in Singapore have discovered a surprisingly simple way to create exotic light structures called optical skyrmions using a 200-year-old optical effect known as the Poisson spot。 Instead of relying on expensive, highly engineered materials, they simply shine a laser at a tiny circular disc, producing sta
Celebrating the United States' 250th anniversary, NASA released a stunning Hubble portrait of Messier 3, an ancient globular cluster with more than 500,000 stars。 The remarkable cluster is helping scientists unravel the Milky Way's past thanks to its rare stars and possible origins in a long ago cosmic merger
Scientists have uncovered new evidence that fireworks can pollute both the air and water in ways that extend beyond the visible smoke。 The findings show that leftover debris, fine particles, and airborne chemicals may affect ecosystems and increase people's exposure to air pollution during major celebrations
NASA is ramping up its lunar ambitions by awarding nearly $600 million for four commercial Moon landings planned for late 2028。 Each mission will carry the same trio of science instruments to improve lunar navigation, study dangerous dust kicked up during landings, and map the Moon's radiation environment。 The agency also revealed plans for new rov