PyTorchKR Papers

최신 논문

총 5,357편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2402.02320

Spin: An Efficient Secure Computation Framework with GPU Acceleration

Accuracy and efficiency remain challenges for multi-party computation (MPC) frameworks. Spin is a GPU-accelerated MPC framework that supports multiple computation parties and a dishonest majority adversarial setup. We propose optimized protocols for non-linear functions that are critical for machine learning, as well as several novel optimizations specific to attention that is the fundamental unit of Transformer models, allowing Spin to perform non-trivial CNNs training and Transformer inference without sacrificing security. At the backend level, Spin leverages GPU, CPU, and RDMA-enabled smart network cards for acceleration. Comprehensive evaluations demonstrate that Spin can be up to $2\times$ faster than the state-of-the-art for deep neural network training. For inference on a Transformer model with 18.9 million parameters, our attention-specific optimizations enable Spin to achieve better efficiency, less communication, and better accuracy.

arXiv:2402.03307공식 코드 · Papers with Code

4D-Rotor Gaussian Splatting: Towards Efficient Novel View Synthesis for Dynamic Scenes

We consider the problem of novel-view synthesis (NVS) for dynamic scenes. Recent neural approaches have accomplished exceptional NVS results for static 3D scenes, but extensions to 4D time-varying scenes remain non-trivial. Prior efforts often encode dynamics by learning a canonical space plus implicit or explicit deformation fields, which struggle in challenging scenarios like sudden movements or generating high-fidelity renderings. In this paper, we introduce 4D Gaussian Splatting (4DRotorGS), a novel method that represents dynamic scenes with anisotropic 4D XYZT Gaussians, inspired by the success of 3D Gaussian Splatting in static scenes. We model dynamics at each timestamp by temporally slicing the 4D Gaussians, which naturally compose dynamic 3D Gaussians and can be seamlessly projected into images. As an explicit spatial-temporal representation, 4DRotorGS demonstrates powerful capabilities for modeling complicated dynamics and fine details--especially for scenes with abrupt motions. We further implement our temporal slicing and splatting techniques in a highly optimized CUDA acceleration framework, achieving real-time inference rendering speeds of up to 277 FPS on an RTX 3090 GPU and 583 FPS on an RTX 4090 GPU. Rigorous evaluations on scenes with diverse motions showcase the superior efficiency and effectiveness of 4DRotorGS, which consistently outperforms existing methods both quantitatively and qualitatively.

arXiv:2402.03135

GPU-Accelerated 3D Polygon Visibility Volumes for Synergistic Perception and Navigation

UAV missions often require specific geometric constraints to be satisfied between ground locations and the vehicle location. Such requirements are typical for contexts where line-of-sight must be maintained between the vehicle location and the ground control location and are also important in surveillance applications where the UAV wishes to be able to sense, e.g., with a camera sensor, a specific region within a complex geometric environment. This problem is further complicated when the ground location is generalized to a convex 2D polygonal region. This article describes the theory and implementation of a system which can quickly calculate the 3D volume that encloses all 3D coordinates from which a 2D convex planar region can be entirely viewed; referred to as a visibility volume. The proposed approach computes visibility volumes using a combination of depth map computation using GPU-acceleration and geometric boolean operations. Solutions to this problem require complex 3D geometric analysis techniques that must execute using arbitrary precision arithmetic on a collection of discontinuous and non-analytic surfaces. Post-processing steps incorporate navigational constraints to further restrict the enclosed coordinates to include both visibility and navigation constraints. Integration of sensing visibility constraints with navigational constraints yields a range of navigable space where a vehicle will satisfy both perceptual sensing and navigational needs of the mission. This algorithm then provides a synergistic perception and navigation sensitive solution yielding a volume of coordinates in 3D that satisfy both the mission path and sensing needs.

arXiv:2402.02887

Time-, Memory- and Parameter-Efficient Visual Adaptation

As foundation models become more popular, there is a growing need to efficiently finetune them for downstream tasks. Although numerous adaptation methods have been proposed, they are designed to be efficient only in terms of how many parameters are trained. They, however, typically still require backpropagating gradients throughout the model, meaning that their training-time and -memory cost does not reduce as significantly. We propose an adaptation method which does not backpropagate gradients through the backbone. We achieve this by designing a lightweight network in parallel that operates on features from the frozen, pretrained backbone. As a result, our method is efficient not only in terms of parameters, but also in training-time and memory usage. Our approach achieves state-of-the-art accuracy-parameter trade-offs on the popular VTAB benchmark, and we further show how we outperform prior works with respect to training-time and -memory usage too. We further demonstrate the training efficiency and scalability of our method by adapting a vision transformer backbone of 4 billion parameters for the computationally demanding task of video classification, without any intricate model parallelism. Here, we outperform a prior adaptor-based method which could only scale to a 1 billion parameter backbone, or fully-finetuning a smaller backbone, with the same GPU and less training time.

arXiv:2402.02554공식 코드 · Papers with Code

DeSparsify: Adversarial Attack Against Token Sparsification Mechanisms in Vision Transformers

Vision transformers have contributed greatly to advancements in the computer vision domain, demonstrating state-of-the-art performance in diverse tasks (e.g., image classification, object detection). However, their high computational requirements grow quadratically with the number of tokens used. Token sparsification mechanisms have been proposed to address this issue. These mechanisms employ an input-dependent strategy, in which uninformative tokens are discarded from the computation pipeline, improving the model's efficiency. However, their dynamism and average-case assumption makes them vulnerable to a new threat vector - carefully crafted adversarial examples capable of fooling the sparsification mechanism, resulting in worst-case performance. In this paper, we present DeSparsify, an attack targeting the availability of vision transformers that use token sparsification mechanisms. The attack aims to exhaust the operating system's resources, while maintaining its stealthiness. Our evaluation demonstrates the attack's effectiveness on three token sparsification mechanisms and examines the attack's transferability between them and its effect on the GPU resources. To mitigate the impact of the attack, we propose various countermeasures.

arXiv:2402.02361공식 코드 · Papers with Code

Pruner: A Speculative Exploration Mechanism to Accelerate Tensor Program Tuning

Tensor program tuning is essential for the efficient deployment of deep neural networks. Search-based approaches have demonstrated scalability and effectiveness in automatically finding high-performance programs for specific hardware. However, the search process is often inefficient, taking hours or even days to discover optimal programs due to the exploration mechanisms guided by an accurate but slow learned cost model. Meanwhile, the learned cost model trained on one platform cannot seamlessly adapt online to another, which we call cross-platform online unawareness. In this work, we propose Pruner and MoA-Pruner. Pruner is a speculative exploration mechanism that accelerates the search process using a "Draft-then-Verify" paradigm. Instead of applying the complex learned cost model to all explored candidates, Pruner drafts small-scale speculative candidates by introducing a naive symbol analyzer (draft model), then identifies the best candidates by the learned cost model. MoA-Pruner introduces Momentum online Adaptation to address the cross-platform online unawareness. We incorporate these techniques into the Ansor and conduct extensive experiments on three GPU-based platforms. Results show that in online cost model tuning scenarios, Pruner and MoA-Pruner can achieve an average speedup of $2.6 \times$ and $4.82 \times$ compared to Ansor. In offline tuning scenarios, Pruner can achieve an average speedup of $4.75 \times$ and $4.05\times$ compared to TenSet and TLP, respectively. The code is available at https://github.com/qiaolian9/Pruner.

arXiv:2402.01239공식 코드 · Papers with Code

PRIME: Protect Your Videos From Malicious Editing

With the development of generative models, the quality of generated content keeps increasing. Recently, open-source models have made it surprisingly easy to manipulate and edit photos and videos, with just a few simple prompts. While these cutting-edge technologies have gained popularity, they have also given rise to concerns regarding the privacy and portrait rights of individuals. Malicious users can exploit these tools for deceptive or illegal purposes. Although some previous works focus on protecting photos against generative models, we find there are still gaps between protecting videos and images in the aspects of efficiency and effectiveness. Therefore, we introduce our protection method, PRIME, to significantly reduce the time cost and improve the protection performance. Moreover, to evaluate our proposed protection method, we consider both objective metrics and human subjective metrics. Our evaluation results indicate that PRIME only costs 8.3% GPU hours of the cost of the previous state-of-the-art method and achieves better protection results on both human evaluation and objective metrics. Code can be found in https://github.com/GuanlinLee/prime.

arXiv:2402.01169

Faster Inference of Integer SWIN Transformer by Removing the GELU Activation

SWIN transformer is a prominent vision transformer model that has state-of-the-art accuracy in image classification tasks. Despite this success, its unique architecture causes slower inference compared with similar deep neural networks. Integer quantization of the model is one of the methods used to improve its inference latency. However, state-of-the-art has not been able to fully quantize the model. In this work, we improve upon the inference latency of the state-of-the-art methods by removing the floating-point operations, which are associated with the GELU activation in Swin Transformer. While previous work proposed to replace the non-integer operations with linear approximation functions, we propose to replace GELU with ReLU activation. The advantage of ReLU over previous methods is its low memory and computation complexity. We use iterative knowledge distillation to compensate for the lost accuracy due to replacing GELU with ReLU. We quantize our GELU-less SWIN transformer and show that on an RTX 4090 NVIDIA GPU we can improve the inference latency of the quantized SWIN transformer by at least $11\%$ while maintaining an accuracy drop of under $0.5\%$ on the ImageNet evaluation dataset.

arXiv:2402.00789

Graph-Mamba: Towards Long-Range Graph Sequence Modeling with Selective State Spaces

Attention mechanisms have been widely used to capture long-range dependencies among nodes in Graph Transformers. Bottlenecked by the quadratic computational cost, attention mechanisms fail to scale in large graphs. Recent improvements in computational efficiency are mainly achieved by attention sparsification with random or heuristic-based graph subsampling, which falls short in data-dependent context reasoning. State space models (SSMs), such as Mamba, have gained prominence for their effectiveness and efficiency in modeling long-range dependencies in sequential data. However, adapting SSMs to non-sequential graph data presents a notable challenge. In this work, we introduce Graph-Mamba, the first attempt to enhance long-range context modeling in graph networks by integrating a Mamba block with the input-dependent node selection mechanism. Specifically, we formulate graph-centric node prioritization and permutation strategies to enhance context-aware reasoning, leading to a substantial improvement in predictive performance. Extensive experiments on ten benchmark datasets demonstrate that Graph-Mamba outperforms state-of-the-art methods in long-range graph prediction tasks, with a fraction of the computational cost in both FLOPs and GPU memory consumption. The code and models are publicly available at https://github.com/bowang-lab/Graph-Mamba.

arXiv:2402.00306

An Accurate and Low-Parameter Machine Learning Architecture for Next Location Prediction

Next location prediction is a discipline that involves predicting a users next location. Its applications include resource allocation, quality of service, energy efficiency, and traffic management. This paper proposes an energy-efficient, small, and low parameter machine learning (ML) architecture for accurate next location prediction, deployable on modest base stations and edge devices. To accomplish this we ran a hundred hyperparameter experiments on the full human mobility patterns of an entire city, to determine an exact ML architecture that reached a plateau of accuracy with the least amount of model parameters. We successfully achieved a reduction in the number of model parameters within published ML architectures from 202 million down to 2 million. This reduced the total size of the model parameters from 791 MB down to 8 MB. Additionally, this decreased the training time by a factor of four, the amount of graphics processing unit (GPU) memory needed for training by a factor of twenty, and the overall accuracy was increased from 80.16% to 82.54%. This improvement allows for modest base stations and edge devices which do not have a large amount of memory or storage, to deploy and utilize the proposed ML architecture for next location prediction.

arXiv:2401.18079공식 코드 · Papers with Code

KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization

LLMs are seeing growing use for applications which require large context windows, and with these large context windows KV cache activations surface as the dominant contributor to memory consumption during inference. Quantization is a promising approach for compressing KV cache activations; however, existing solutions fail to represent activations accurately in sub-4-bit precision. Our work, KVQuant, facilitates low precision KV cache quantization by incorporating several novel methods: (i) Per-Channel Key Quantization, where we adjust the dimension along which we quantize the Key activations to better match the distribution; (ii) Pre-RoPE Key Quantization, where we quantize Key activations before the rotary positional embedding to mitigate its impact on quantization; (iii) Non-Uniform KV Cache Quantization, where we derive per-layer sensitivity-weighted non-uniform datatypes that better represent the distributions; and (iv) Per-Vector Dense-and-Sparse Quantization, where we isolate outliers separately for each vector to minimize skews in quantization ranges. By applying our method to the LLaMA, Llama-2, Llama-3, and Mistral models, we achieve < 0.1 perplexity degradation with 3-bit quantization on both Wikitext-2 and C4, outperforming existing approaches. Our method enables serving LLaMA-7B with a context length of up to 1 million on a single A100-80GB GPU and up to 10 million on an 8-GPU system. We develop custom CUDA kernels for KVQuant, showing that we can achieve up to ~1.7x speedups, compared to baseline fp16 matrix-vector multiplications, for the LLaMA-7B model.

arXiv:2401.18034

Paramanu: A Family of Novel Efficient Generative Foundation Language Models for Indian Languages

We present "Paramanu", a family of novel language models (LM) for Indian languages, consisting of auto-regressive monolingual, bilingual, and multilingual models pretrained from scratch. Currently, it covers 10 languages (Assamese, Bangla, Hindi, Konkani, Maithili, Marathi, Odia, Sanskrit, Tamil, Telugu) across 5 scripts (Bangla, Devanagari, Odia, Tamil, Telugu). The models are pretrained on a single GPU with context size of 1024 and vary in size from 13.29 million (M) to 367.5 M parameters. We proposed a RoPE embedding scaling method that enables us to pretrain language models from scratch at larger sequence length context size than typical GPU memory permits. We also introduced a novel efficient Indic tokenizer, "mBharat", using a combination of BPE and Unigram, achieving the least fertility score and the ability to tokenize unseen languages in both the same script & Roman script. We also proposed and performed language-specific tokenization for multilingual models & domain-specific tokenization for monolingual models. To address the "curse of multilinguality" in our mParamanu model, we pretrained on comparable corpora based on typological grouping within the same script. Our findings show a language transfer phenomenon from low-resource to high-resource languages within languages of the same script & typology. Human evaluations for open-ended text generation demonstrated that Paramanu models outperformed several LLMs, despite being 20 to 64 times smaller. We created instruction-tuning datasets & instruction-tuned our models on 23,000 instructions in respective languages. Comparisons with multilingual LLMs across various benchmarks for natural language (NL) understanding, NL inference, & reading comprehension highlight the advantages of our models; leads to the conclusion that high quality generative LM are possible without high amount of compute power & enormous number of parameters.

arXiv:2401.17870

Efficient Subseasonal Weather Forecast using Teleconnection-informed Transformers

Subseasonal forecasting, which is pivotal for agriculture, water resource management, and early warning of disasters, faces challenges due to the chaotic nature of the atmosphere. Recent advances in machine learning (ML) have revolutionized weather forecasting by achieving competitive predictive skills to numerical models. However, training such foundation models requires thousands of GPU days, which causes substantial carbon emissions and limits their broader applicability. Moreover, ML models tend to fool the pixel-wise error scores by producing smoothed results which lack physical consistency and meteorological meaning. To deal with the aforementioned problems, we propose a teleconnection-informed transformer. Our architecture leverages the pretrained Pangu model to achieve good initial weights and integrates a teleconnection-informed temporal module to improve predictability in an extended temporal range. Remarkably, by adjusting 1.1% of the Pangu model's parameters, our method enhances predictability on four surface and five upper-level atmospheric variables at a two-week lead time. Furthermore, the teleconnection-filtered features improve the spatial granularity of outputs significantly, indicating their potential physical consistency. Our research underscores the importance of atmospheric and oceanic teleconnections in driving future weather conditions. Besides, it presents a resource-efficient pathway for researchers to leverage existing foundation models on versatile downstream tasks.

arXiv:2401.16492

GPU Cluster Scheduling for Network-Sensitive Deep Learning

We propose a novel GPU-cluster scheduler for distributed DL (DDL) workloads that enables proximity based consolidation of GPU resources based on the DDL jobs' sensitivities to the anticipated communication-network delays. Our scheduler consists of three major components: (i) a classical delay scheduling algorithm to facilitate job placement and consolidation; (ii) a network-sensitive job preemption strategy; and (iii) an "auto-tuner" mechanism to optimize delay timers for effective delay scheduling. Additionally, to enable a cost-effective methodology for large-scale experiments, we develop a data-driven DDL cluster simulation platform. Employing the simulation platform we compare against several state-of-the-art alternatives on real-world workload traces to demonstrate the benefits of our design. Our scheduler can provide improvement of up to 69% in end-to-end Makespan for training all jobs compared to the prevailing consolidation-based scheduling methods, while reducing the average job completion time by up to 83% and minimizing the communication overheads by up to 98% under congested networking conditions.

arXiv:2401.16456공식 코드 · Papers with Code

SHViT: Single-Head Vision Transformer with Memory Efficient Macro Design

Recently, efficient Vision Transformers have shown great performance with low latency on resource-constrained devices. Conventionally, they use 4x4 patch embeddings and a 4-stage structure at the macro level, while utilizing sophisticated attention with multi-head configuration at the micro level. This paper aims to address computational redundancy at all design levels in a memory-efficient manner. We discover that using larger-stride patchify stem not only reduces memory access costs but also achieves competitive performance by leveraging token representations with reduced spatial redundancy from the early stages. Furthermore, our preliminary analyses suggest that attention layers in the early stages can be substituted with convolutions, and several attention heads in the latter stages are computationally redundant. To handle this, we introduce a single-head attention module that inherently prevents head redundancy and simultaneously boosts accuracy by parallelly combining global and local information. Building upon our solutions, we introduce SHViT, a Single-Head Vision Transformer that obtains the state-of-the-art speed-accuracy tradeoff. For example, on ImageNet-1k, our SHViT-S4 is 3.3x, 8.1x, and 2.4x faster than MobileViTv2 x1.0 on GPU, CPU, and iPhone12 mobile device, respectively, while being 1.3% more accurate. For object detection and instance segmentation on MS COCO using Mask-RCNN head, our model achieves performance comparable to FastViT-SA12 while exhibiting 3.8x and 2.0x lower backbone latency on GPU and mobile device, respectively.

arXiv:2401.16757

SwapNet: Efficient Swapping for DNN Inference on Edge AI Devices Beyond the Memory Budget

Executing deep neural networks (DNNs) on edge artificial intelligence (AI) devices enables various autonomous mobile computing applications. However, the memory budget of edge AI devices restricts the number and complexity of DNNs allowed in such applications. Existing solutions, such as model compression or cloud offloading, reduce the memory footprint of DNN inference at the cost of decreased model accuracy or autonomy. To avoid these drawbacks, we divide DNN into blocks and swap them in and out in order, such that large DNNs can execute within a small memory budget. Nevertheless, naive swapping on edge AI devices induces significant delays due to the redundant memory operations in the DNN development ecosystem for edge AI devices. To this end, we develop SwapNet, an efficient DNN block swapping middleware for edge AI devices. We systematically eliminate the unnecessary memory operations during block swapping while retaining compatible with the deep learning frameworks, GPU backends, and hardware architectures of edge AI devices. We further showcase the utility of SwapNet via a multi-DNN scheduling scheme. Evaluations on eleven DNN inference tasks in three applications demonstrate that SwapNet achieves almost the same latency as the case with sufficient memory even when DNNs demand 2.32x to 5.81x memory beyond the available budget. The design of SwapNet also provides novel and feasible insights for deploying large language models (LLMs) on edge AI devices in the future.

arXiv:2401.15207공식 코드 · Papers with Code

HiFT: A Hierarchical Full Parameter Fine-Tuning Strategy

Full-parameter fine-tuning has become the go-to choice for adapting language models (LMs) to downstream tasks due to its excellent performance. As LMs grow in size, fine-tuning the full parameters of LMs requires a prohibitively large amount of GPU memory. Existing approaches utilize zeroth-order optimizer to conserve GPU memory, which can potentially compromise the performance of LMs as non-zero order optimizers tend to converge more readily on most downstream tasks. In this paper, we propose a novel optimizer-independent end-to-end hierarchical fine-tuning strategy, HiFT, which only updates a subset of parameters at each training step. HiFT can significantly reduce the amount of gradients and optimizer state parameters residing in GPU memory at the same time, thereby reducing GPU memory usage. Our results demonstrate that: (1) HiFT achieves comparable performance to parameter-efficient fine-tuning and standard full parameter fine-tuning. (2) HiFT supports various optimizers including AdamW, AdaGrad, SGD, etc. (3) HiFT can save more than 60\% GPU memory compared with standard full-parameter fine-tuning for 7B model. (4) HiFT enables full-parameter fine-tuning of a 7B model on single 48G A6000 with a precision of 32 using the AdamW optimizer, without using any memory saving techniques.

arXiv:2401.15855공식 코드 · Papers with Code

Cross-Scale MAE: A Tale of Multi-Scale Exploitation in Remote Sensing

Remote sensing images present unique challenges to image analysis due to the extensive geographic coverage, hardware limitations, and misaligned multi-scale images. This paper revisits the classical multi-scale representation learning problem but under the general framework of self-supervised learning for remote sensing image understanding. We present Cross-Scale MAE, a self-supervised model built upon the Masked Auto-Encoder (MAE).During pre-training, Cross-Scale MAE employs scale augmentation techniques and enforces cross-scale consistency constraints through both contrastive and generative losses to ensure consistent and meaningful representations well-suited for a wide range of downstream tasks. Further, our implementation leverages the xFormers library to accelerate network pre-training on a single GPU while maintaining the quality of learned representations. Experimental evaluations demonstrate that Cross-Scale MAE exhibits superior performance compared to standard MAE and other state-of-the-art remote sensing MAE methods.

arXiv:2401.15896공식 코드 · Papers with Code

M2-Encoder: Advancing Bilingual Image-Text Understanding by Large-scale Efficient Pretraining

Vision-language foundation models like CLIP have revolutionized the field of artificial intelligence. Nevertheless, VLM models supporting multi-language, e.g., in both Chinese and English, have lagged due to the relative scarcity of large-scale pretraining datasets. Toward this end, we introduce a comprehensive bilingual (Chinese-English) dataset BM-6B with over 6 billion image-text pairs, aimed at enhancing multimodal foundation models to well understand images in both languages. To handle such a scale of dataset, we propose a novel grouped aggregation approach for image-text contrastive loss computation, which reduces the communication overhead and GPU memory demands significantly, facilitating a 60% increase in training speed. We pretrain a series of bilingual image-text foundation models with an enhanced fine-grained understanding ability on BM-6B, the resulting models, dubbed as $M^2$-Encoders (pronounced "M-Square"), set new benchmarks in both languages for multimodal retrieval and classification tasks. Notably, Our largest $M^2$-Encoder-10B model has achieved top-1 accuracies of 88.5% on ImageNet and 80.7% on ImageNet-CN under a zero-shot classification setting, surpassing previously reported SoTA methods by 2.2% and 21.1%, respectively. The $M^2$-Encoder series represents one of the most comprehensive bilingual image-text foundation models to date, so we are making it available to the research community for further exploration and development.

arXiv:2401.14489공식 코드 · Papers with Code

The Case for Co-Designing Model Architectures with Hardware

While GPUs are responsible for training the vast majority of state-of-the-art deep learning models, the implications of their architecture are often overlooked when designing new deep learning (DL) models. As a consequence, modifying a DL model to be more amenable to the target hardware can significantly improve the runtime performance of DL training and inference. In this paper, we provide a set of guidelines for users to maximize the runtime performance of their transformer models. These guidelines have been created by carefully considering the impact of various model hyperparameters controlling model shape on the efficiency of the underlying computation kernels executed on the GPU. We find the throughput of models with efficient model shapes is up to 39\% higher while preserving accuracy compared to models with a similar number of parameters but with unoptimized shapes.