SEARCH · Searcharxiv

Results for “transformer architecture”

Search indexed arXiv papers on artificial intelligence, large language models, computer vision and robotics. Read source abstracts and follow links to arXiv.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 325 records · Page 18Linked to original sources

arXiv2025

Apriel-H1: Towards Efficient Enterprise Reasoning Models

Large Language Models (LLMs) achieve remarkable reasoning capabilities through transformer architectures with attention mechanisms. However, transformers suffer from quadratic time and memory complexity in the attention module (MHA) and require caching key-value states during inference, which severely limits throughput and scalability. High inference throughput is critical for agentic tasks, long-context reasoning, efficient deployment under high request loads, and more efficient test-time compute scaling. State Space Models (SSMs) such as Mamba offer a promising alternative with linear inference complexity and a constant memory footprint via recurrent computation with fixed-size hidden states. In this technical report we introduce the Apriel-H1 family of hybrid LLMs that combine transformer attention and SSM sequence mixers for efficient reasoning at 15B model size. These models are obtained through incremental distillation from a pretrained reasoning transformer, Apriel-Nemotron-15B-Thinker, progressively replacing less critical attention layers with linear Mamba blocks. We release multiple post-distillation variants of Apriel-H1-15B-Thinker with different SSM-to-MHA ratios and analyse how reasoning performance degrades as more Mamba layers replace MHA. Additionally, we release a 30/50 hybrid variant of Apriel-H1, further fine-tuned on a supervised dataset of reasoning traces, achieving over 2x higher inference throughput when deployed in the production-ready vLLM environment, with minimal degradation in reasoning performance. This shows that distilled hybrid SSM-Transformer architectures can deliver substantial efficiency gains over the pretrained transformer equivalent without substantially compromising the reasoning quality.

cs.LG↗

arXiv2022

What Makes for Good Tokenizers in Vision Transformer?

The architecture of transformers, which recently witness booming applications in vision tasks, has pivoted against the widespread convolutional paradigm. Relying on the tokenization process that splits inputs into multiple tokens, transformers are capable of extracting their pairwise relationships using self-attention. While being the stemming building block of transformers, what makes for a good tokenizer has not been well understood in computer vision. In this work, we investigate this uncharted problem from an information trade-off perspective. In addition to unifying and understanding existing structural modifications, our derivation leads to better design strategies for vision tokenizers. The proposed Modulation across Tokens (MoTo) incorporates inter-token modeling capability through normalization. Furthermore, a regularization objective TokenProp is embraced in the standard training regime. Through extensive experiments on various transformer architectures, we observe both improved performance and intriguing properties of these two plug-and-play designs with negligible computational overhead. These observations further indicate the importance of the commonly-omitted designs of tokenizers in vision transformer.

cs.CV↗

arXiv2025

SparkAttention: High-Performance Multi-Head Attention for Large Models on Volta GPU Architecture

Transformer are widely used in various fields such as natural language processing and computer vision. However, the training time for large Transformer models can be challenging due to the Multi-Head Attention (MHA) mechanism. Especially as models become larger, training becomes more costly. So it is crucial to utilize various resources for efficient model training. Currently, NVIDIA Volta GPU is still widely used. However, because the computational shapes supported by Tensor Core Units (TCU) of Volta GPU differ from other GPU architectures, most efforts have not focused on using them to accelerate Transformer training. To address this issue, we propose SparkAttention, an acceleration library designed to speed up MHA training on the Volta GPU. SparkAttention leverages TCU and kernel fusion to reduce the number of high bandwidth memory (HBM) accesses and overhead. Our End-to-End experimental results on an NVIDIA V100 GPU show that SparkAttention achieves on average 1.80$\times$ (up to 2.46$\times$) speedup compared to using PyTorch.

cs.DC↗

arXiv2026

WaveFormer: Wavelet Embedding Transformer for Biomedical Signals

Biomedical signal classification presents unique challenges due to long sequences, complex temporal dynamics, and multi-scale frequency patterns that are poorly captured by standard transformer architectures. We propose WaveFormer, a transformer architecture that integrates wavelet decomposition at two critical stages: embedding construction, where multi-channel Discrete Wavelet Transform (DWT) extracts frequency features to create tokens containing both time-domain and frequency-domain information, and positional encoding, where Dynamic Wavelet Positional Encoding (DyWPE) adapts position embeddings to signal-specific temporal structure through mono-channel DWT analysis. We evaluate WaveFormer on eight diverse datasets spanning human activity recognition and brain signal analysis, with sequence lengths ranging from 50 to 3000 timesteps and channel counts from 1 to 144. Experimental results demonstrate that WaveFormer achieves competitive performance through comprehensive frequency-aware processing. Our approach provides a principled framework for incorporating frequency-domain knowledge into transformer-based time series classification.

cs.LG↗

arXiv2023

Self-Supervised Learning from Non-Object Centric Images with a Geometric Transformation Sensitive Architecture

Most invariance-based self-supervised methods rely on single object-centric images (e.g., ImageNet images) for pretraining, learning features that invariant to geometric transformation. However, when images are not object-centric, the semantics of the image can be significantly altered due to cropping. Furthermore, as the model becomes insensitive to geometric transformations, it may struggle to capture location information. For this reason, we propose a Geometric Transformation Sensitive Architecture designed to be sensitive to geometric transformations, specifically focusing on four-fold rotation, random crop, and multi-crop. Our method encourages the student to be sensitive by predicting rotation and using targets that vary with those transformations through pooling and rotating the teacher feature map. Additionally, we use patch correspondence loss to encourage correspondence between patches with similar features. This approach allows us to capture long-term dependencies in a more appropriate way than capturing long-term dependencies by encouraging local-to-global correspondence, which occurs when learning to be insensitive to multi-crop. Our approach demonstrates improved performance when using non-object-centric images as pretraining data compared to other methods that train the model to be insensitive to geometric transformation. We surpass DINO[Caron et al.[2021b]] baseline in tasks including image classification, semantic segmentation, detection, and instance segmentation with improvements of 4.9 $Top-1 Acc$, 3.3 $mIoU$, 3.4 $AP^b$, and 2.7 $AP^m$. Code and pretrained models are publicly available at: https://github.com/bok3948/GTSA

cs.CV↗

arXiv2025

MSVIT: Improving Spiking Vision Transformer Using Multi-scale Attention Fusion

The combination of Spiking Neural Networks (SNNs) with Vision Transformer architectures has garnered significant attention due to their potential for energy-efficient and high-performance computing paradigms. However, a substantial performance gap still exists between SNN-based and ANN-based transformer architectures. While existing methods propose spiking self-attention mechanisms that are successfully combined with SNNs, the overall architectures proposed by these methods suffer from a bottleneck in effectively extracting features from different image scales. In this paper, we address this issue and propose MSVIT. This novel spike-driven Transformer architecture firstly uses multi-scale spiking attention (MSSA) to enhance the capabilities of spiking attention blocks. We validate our approach across various main datasets. The experimental results show that MSVIT outperforms existing SNN-based models, positioning itself as a state-of-the-art solution among SNN-transformer architectures. The codes are available at https://github.com/Nanhu-AI-Lab/MSViT.

cs.CV↗

arXiv2025

Transformers to SSMs: Distilling Quadratic Knowledge to Subquadratic Models

Transformer architectures have become a dominant paradigm for domains like language modeling but suffer in many inference settings due to their quadratic-time self-attention. Recently proposed subquadratic architectures, such as Mamba, have shown promise, but have been pretrained with substantially less computational resources than the strongest Transformer models. In this work, we present a method that is able to distill a pretrained Transformer architecture into alternative architectures such as state space models (SSMs). The key idea to our approach is that we can view both Transformers and SSMs as applying different forms of mixing matrices over the token sequences. We can thus progressively distill the Transformer architecture by matching different degrees of granularity in the SSM: first matching the mixing matrices themselves, then the hidden units at each block, and finally the end-to-end predictions. Our method, called MOHAWK, is able to distill a Mamba-2 variant based on the Phi-1.5 architecture (Phi-Mamba) using only 3B tokens and a hybrid version (Hybrid Phi-Mamba) using 5B tokens. Despite using less than 1% of the training data typically used to train models from scratch, Phi-Mamba boasts substantially stronger performance compared to all past open-source non-Transformer models. MOHAWK allows models like SSMs to leverage computational resources invested in training Transformer-based architectures, highlighting a new avenue for building such models.

cs.LG↗

arXiv2025

Activator: GLU Activation Function as the Core Component of a Vision Transformer

The transformer architecture has driven many successes in a variety of tasks within the field of deep learning, in particular the recent advances in natural language processing (NLP) culminating with large language models (LLM). Adding to that success, transformer architecture has found widespread interest from computer vision (CV) researchers and practitioners, allowing for many advancements in vision-related tasks and opening the door for multitask and multi-modal deep learning architectures that share the same principle of operation. One drawback to these architectures is their reliance on the scaled dot product attention mechanism with the softmax activation function, which is computationally expensive and requires large compute capabilities for both training and inference. This paper investigates substituting the MLP and attention mechanism usually adopted for transformer architecture with an architecture based on incorporating a gated linear unit (GLU) activation function structure with the aim of reducing the computational cost. The equalized experimental assessments conducted in this work show that the proposed modification with the targeted reductions in computational complexity offers competitive performance compared to the selected baseline architectures. The results are significantly in support of the aims of this work, in which the focus was to extensively utilize GLU-based MLPs, establishing a more efficient but capable alternative to the traditional MLP and the attention mechanism as the core component in the design of transformer architectures.

cs.CV↗

arXiv2023

A Systematic Review for Transformer-based Long-term Series Forecasting

The emergence of deep learning has yielded noteworthy advancements in time series forecasting (TSF). Transformer architectures, in particular, have witnessed broad utilization and adoption in TSF tasks. Transformers have proven to be the most successful solution to extract the semantic correlations among the elements within a long sequence. Various variants have enabled transformer architecture to effectively handle long-term time series forecasting (LTSF) tasks. In this article, we first present a comprehensive overview of transformer architectures and their subsequent enhancements developed to address various LTSF tasks. Then, we summarize the publicly available LTSF datasets and relevant evaluation metrics. Furthermore, we provide valuable insights into the best practices and techniques for effectively training transformers in the context of time-series analysis. Lastly, we propose potential research directions in this rapidly evolving field.

cs.LG↗

arXiv2023

Sharingan: A Transformer-based Architecture for Gaze Following

Gaze is a powerful form of non-verbal communication and social interaction that humans develop from an early age. As such, modeling this behavior is an important task that can benefit a broad set of application domains ranging from robotics to sociology. In particular, Gaze Following is defined as the prediction of the pixel-wise 2D location where a person in the image is looking. Prior efforts in this direction have focused primarily on CNN-based architectures to perform the task. In this paper, we introduce a novel transformer-based architecture for 2D gaze prediction. We experiment with 2 variants: the first one retains the same task formulation of predicting a gaze heatmap for one person at a time, while the second one casts the problem as a 2D point regression and allows us to perform multi-person gaze prediction with a single forward pass. This new architecture achieves state-of-the-art results on the GazeFollow and VideoAttentionTarget datasets. The code for this paper will be made publicly available.

cs.CV↗

arXiv2023

Transformers in Time-series Analysis: A Tutorial

Transformer architecture has widespread applications, particularly in Natural Language Processing and computer vision. Recently Transformers have been employed in various aspects of time-series analysis. This tutorial provides an overview of the Transformer architecture, its applications, and a collection of examples from recent research papers in time-series analysis. We delve into an explanation of the core components of the Transformer, including the self-attention mechanism, positional encoding, multi-head, and encoder/decoder. Several enhancements to the initial, Transformer architecture are highlighted to tackle time-series tasks. The tutorial also provides best practices and techniques to overcome the challenge of effectively training Transformers for time-series analysis.

cs.LG↗

arXiv2022

KVT: k-NN Attention for Boosting Vision Transformers

Convolutional Neural Networks (CNNs) have dominated computer vision for years, due to its ability in capturing locality and translation invariance. Recently, many vision transformer architectures have been proposed and they show promising performance. A key component in vision transformers is the fully-connected self-attention which is more powerful than CNNs in modelling long range dependencies. However, since the current dense self-attention uses all image patches (tokens) to compute attention matrix, it may neglect locality of images patches and involve noisy tokens (e.g., clutter background and occlusion), leading to a slow training process and potential degradation of performance. To address these problems, we propose the $k$-NN attention for boosting vision transformers. Specifically, instead of involving all the tokens for attention matrix calculation, we only select the top-$k$ similar tokens from the keys for each query to compute the attention map. The proposed $k$-NN attention naturally inherits the local bias of CNNs without introducing convolutional operations, as nearby tokens tend to be more similar than others. In addition, the $k$-NN attention allows for the exploration of long range correlation and at the same time filters out irrelevant tokens by choosing the most similar tokens from the entire image. Despite its simplicity, we verify, both theoretically and empirically, that $k$-NN attention is powerful in speeding up training and distilling noise from input tokens. Extensive experiments are conducted by using 11 different vision transformer architectures to verify that the proposed $k$-NN attention can work with any existing transformer architectures to improve its prediction performance. The codes are available at \url{https://github.com/damo-cv/KVT}.

cs.CV↗

arXiv2020

Improvement of a dedicated model for open domain persona-aware dialogue generation

This paper analyzes some speed and performance improvement methods of Transformer architecture in recent years, mainly its application in dedicated model training. The dedicated model studied here refers to the open domain persona-aware dialogue generation model, and the dataset is multi turn short dialogue, The total length of a single input sequence is no more than 105 tokens. Therefore, many improvements in the architecture and attention mechanism of transformer architecture for long sequence processing are not discussed in this paper. The source code of the experiments has been open sourced: https://github.com/ghosthamlet/persona

cs.CL↗

arXiv2024

Transformers-based architectures for stroke segmentation: A review

Stroke remains a significant global health concern, necessitating precise and efficient diagnostic tools for timely intervention and improved patient outcomes. The emergence of deep learning methodologies has transformed the landscape of medical image analysis. Recently, Transformers, initially designed for natural language processing, have exhibited remarkable capabilities in various computer vision applications, including medical image analysis. This comprehensive review aims to provide an in-depth exploration of the cutting-edge Transformer-based architectures applied in the context of stroke segmentation. It commences with an exploration of stroke pathology, imaging modalities, and the challenges associated with accurate diagnosis and segmentation. Subsequently, the review delves into the fundamental ideas of Transformers, offering detailed insights into their architectural intricacies and the underlying mechanisms that empower them to effectively capture complex spatial information within medical images. The existing literature is systematically categorized and analyzed, discussing various approaches that leverage Transformers for stroke segmentation. A critical assessment is provided, highlighting the strengths and limitations of these methods, including considerations of performance and computational efficiency. Additionally, this review explores potential avenues for future research and development

eess.IV↗

arXiv2023

POTTER: Pooling Attention Transformer for Efficient Human Mesh Recovery

Transformer architectures have achieved SOTA performance on the human mesh recovery (HMR) from monocular images. However, the performance gain has come at the cost of substantial memory and computational overhead. A lightweight and efficient model to reconstruct accurate human mesh is needed for real-world applications. In this paper, we propose a pure transformer architecture named POoling aTtention TransformER (POTTER) for the HMR task from single images. Observing that the conventional attention module is memory and computationally expensive, we propose an efficient pooling attention module, which significantly reduces the memory and computational cost without sacrificing performance. Furthermore, we design a new transformer architecture by integrating a High-Resolution (HR) stream for the HMR task. The high-resolution local and global features from the HR stream can be utilized for recovering more accurate human mesh. Our POTTER outperforms the SOTA method METRO by only requiring 7% of total parameters and 14% of the Multiply-Accumulate Operations on the Human3.6M (PA-MPJPE metric) and 3DPW (all three metrics) datasets. The project webpage is https://zczcwh.github.io/potter_page.

cs.CV↗

arXiv2026

FED-HARGPT: A Hybrid Centralized-Federated Approach of a Transformer-based Architecture for Human Context Recognition

The study explores a hybrid centralized-federated approach for Human Activity Recognition (HAR) using a Transformer-based architecture. With the increasing ubiquity of edge devices, such as smartphones and wearables, a significant amount of private data from wearable and inertial sensors is generated, facilitating discreet monitoring of human activities, including resting, sleeping, and walking. This research focuses on deploying HAR technologies using mobile sensor data and leveraging Federated Learning within the Flower framework to evaluate the training of a federated model derived from a centralized baseline. The experimental results demonstrate the effectiveness of the proposed hybrid approach in improving the accuracy and robustness of HAR models while preserving data privacy in a non-IID data scenario. The federated learning setup demonstrated comparable performance to centralized models, highlighting the potential of federated learning to strike a balance between data privacy and model performance in real-world applications.

eess.SP↗

arXiv2024

Synthesizing Proteins on the Graphics Card. Protein Folding and the Limits of Critical AI Studies

This paper investigates the application of the transformer architecture in protein folding, as exemplified by DeepMind's AlphaFold project, and its implications for the understanding of so-called large language models. The prevailing discourse often assumes a ready-made analogy between proteins, encoded as sequences of amino acids, and natural language, which we term the language paradigm of computational (structural) biology. Instead of assuming this analogy as given, we critically evaluate it to assess the kind of knowledge-making afforded by the transformer architecture. We first trace the analogy's emergence and historical development, carving out the influence of structural linguistics on structural biology beginning in the mid-20th century. We then examine three often overlooked preprocessing steps essential to the transformer architecture, including subword tokenization, word embedding, and positional encoding, to demonstrate its regime of representation based on continuous, high-dimensional vector spaces, which departs from the discrete nature of language. The successful deployment of transformers in protein folding, we argue, discloses what we consider a non-linguistic approach to token processing intrinsic to the architecture. We contend that through this non-linguistic processing, the transformer architecture carves out unique epistemological territory and produces a new class of knowledge, distinct from established domains. We contend that our search for intelligent machines has to begin with the shape, rather than the place, of intelligence. Consequently, the emerging field of critical AI studies should take methodological inspiration from the history of science in its quest to conceptualize the contributions of artificial intelligence to knowledge-making, within and beyond the domain-specific sciences.

cs.CY↗

arXiv2024

Looped Transformers are Better at Learning Learning Algorithms

Transformers have demonstrated effectiveness in in-context solving data-fitting problems from various (latent) models, as reported by Garg et al. However, the absence of an inherent iterative structure in the transformer architecture presents a challenge in emulating the iterative algorithms, which are commonly employed in traditional machine learning methods. To address this, we propose the utilization of looped transformer architecture and its associated training methodology, with the aim of incorporating iterative characteristics into the transformer architectures. Experimental results suggest that the looped transformer achieves performance comparable to the standard transformer in solving various data-fitting problems, while utilizing less than 10% of the parameter count.

cs.LG↗