Title: LoRA on the Go: Instance-level Dynamic LoRA Selection and Merging Paper Link: https://arxiv.org/pdf/2511.07129 Innovations For the first time, a combination of Generative Mask and Discriminative Mask is used, where the generative mask is applied to video data reconstruction tasks, and the discriminative mask is used for video understanding tasks. Both share the same network … Read more
This article reviews concepts such as the attention mechanism. Self-Attention The concept of “attention” originated from efforts to improve Recurrent Neural Networks (RNNs) to handle longer sequences or sentences. For example, consider translating a sentence from one language to another. Word-for-word translation is often impractical as it ignores the complex grammatical structures and idiomatic expressions … Read more
On November 17, Arm Technology officially launched its next-generation NPU IP ‘Zhouyi’ X3 at a product launch event in Shanghai on the 13th of this month. It adopts a hybrid architecture of DSP + DSA and can be widely used in four core areas: infrastructure, smart vehicles, mobile terminals, and smart IoT. The X3 is … Read more
In the current era of large language models (LLMs), building an efficient inference framework from scratch allows us to gain a deeper understanding of the underlying logic of AI-generated content. C++, with its close-to-hardware and low-overhead characteristics, has become the preferred language for implementing lightweight LLM inference. This article will integrate the core design ideas … Read more
Your question is very critical! Indeed, Huawei’s Ascend series chips are based on NPU (Neural Processing Unit) architecture, which is fundamentally different from general-purpose accelerators like NVIDIA GPUs. However, it does not mean that it cannot perform high-end training; rather, there are differences in technical routes, ecosystem maturity, and absolute computing power. The following is … Read more
The GPU Revolution: Simplifying Computational Architecture for Large Model Training As the hundred billion parameter models roar in GPU clusters, a revolution in computational efficiency driven by architectural simplification is quietly reconstructing the physical laws and energy consumption boundaries of large model training. 1. The GPU Dilemma in Large Model Training: Challenges of Computational Power, … Read more
Title: A Synergistic CNN-Transformer Network with Pooling Attention Fusion for Hyperspectral Image Classification Paper Link:https://github.com/chenpeng052/SCT-Net Collaborative Architecture: The “Synergistic CNN-Transformer” framework is proposed, utilizing the local spatial representation strengths of CNNs and the global spectral dependency modeling capabilities of Transformers for hyperspectral pixel-level classification. TBFE Dual-Branch Feature Extraction: A parallel 3D convolution branch (1×1×3 kernel, … Read more
Title: A synergistic CNN-transformer network with pooling attention fusion for hyperspectral image classification Paper Link: https://github.com/chenpeng052/SCT-Net Twin-Branch Feature Extraction Module (TBFE): Parallel combination of3D Convolution (for spectral features) and2D Convolution (for spatial features), achieving spectral-spatial joint modeling early on. Hybrid Pooling Attention Module (HPA): Combines average pooling and max pooling to capture spatial dependencies through … Read more
Title: A synergistic CNN-transformer network with pooling attention fusion for hyperspectral image classification Paper link: https://github.com/chenpeng052/SCT-Net HPA (Hybrid Pooling Attention) Hybrid Pooling Attention: Divides channels into groups and performs global average pooling and global max pooling in both horizontal and vertical directions, followed by a 1×11 imes11×1 convolution and Sigmoid to obtain channel attention; simultaneously, … Read more
Title:A synergistic CNN-transformer network with pooling attention fusion for hyperspectral image classification Paper link: https://github.com/chenpeng052/SCT-Net Proposed a CNN-Transformer synergistic network, combining the local feature extraction capability of CNNs with the global modeling capability of Transformers. Designed a Twin-Branch Feature Extraction (TBFE) module, which uses 2D and 3D convolutions in parallel to extract spatial and spectral … Read more