PyTorchKR Papers

최신 논문

총 5,576편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2305.13484공식 코드 · Papers with Code

Flover: A Temporal Fusion Framework for Efficient Autoregressive Model Parallel Inference

Autoregressive models, despite their commendable performance in a myriad of generative tasks, face challenges stemming from their inherently sequential structure. Inference on these models, by design, harnesses a temporal dependency, where the current token's probability distribution is conditioned on preceding tokens. This inherent characteristic severely impedes computational efficiency during inference as a typical inference request can require more than thousands of tokens, where generating each token requires a load of entire model weights, making the inference more memory-bound. The large overhead becomes profound in real deployment where requests arrive randomly, necessitating various generation lengths. Existing solutions, such as dynamic batching and concurrent instances, introduce significant response delays and bandwidth contention, falling short of achieving optimal latency and throughput. To address these shortcomings, we propose Flover -- a temporal fusion framework for efficiently inferring multiple requests in parallel. We deconstruct the general generation pipeline into pre-processing and token generation, and equip the framework with a dedicated work scheduler for fusing the generation process temporally across all requests. By orchestrating the token-level parallelism, Flover exhibits optimal hardware efficiency and significantly spares the system resources. By further employing a fast buffer reordering algorithm that allows memory eviction of finished tasks, it brings over 11x inference speedup on GPT and 16x on LLAMA compared to the cutting-edge solutions provided by NVIDIA FasterTransformer. Crucially, by leveraging the advanced tensor parallel technique, Flover proves efficacious across diverse computational landscapes, from single-GPU setups to distributed scenarios, thereby offering robust performance optimization that adapts to variable use cases.

arXiv:2305.14201공식 코드 · Papers with Code

Goat: Fine-tuned LLaMA Outperforms GPT-4 on Arithmetic Tasks

We introduce Goat, a fine-tuned LLaMA model that significantly outperforms GPT-4 on a range of arithmetic tasks. Fine-tuned on a synthetically generated dataset, Goat achieves state-of-the-art performance on BIG-bench arithmetic sub-task. In particular, the zero-shot Goat-7B matches or even surpasses the accuracy achieved by the few-shot PaLM-540B. Surprisingly, Goat can achieve near-perfect accuracy on large-number addition and subtraction through supervised fine-tuning only, which is almost impossible with previous pretrained language models, such as Bloom, OPT, GPT-NeoX, etc. We attribute Goat's exceptional performance to LLaMA's consistent tokenization of numbers. To tackle more challenging tasks like large-number multiplication and division, we propose an approach that classifies tasks based on their learnability, and subsequently decomposes unlearnable tasks, such as multi-digit multiplication and division, into a series of learnable tasks by leveraging basic arithmetic principles. We thoroughly examine the performance of our model, offering a comprehensive evaluation of the effectiveness of our proposed decomposition steps. Additionally, Goat-7B can be easily trained using LoRA on a 24GB VRAM GPU, facilitating reproducibility for other researchers. We release our model, dataset, and the Python script for dataset generation.

arXiv:2305.12213

Taming Resource Heterogeneity In Distributed ML Training With Dynamic Batching

Current techniques and systems for distributed model training mostly assume that clusters are comprised of homogeneous servers with a constant resource availability. However, cluster heterogeneity is pervasive in computing infrastructure, and is a fundamental characteristic of low-cost transient resources (such as EC2 spot instances). In this paper, we develop a dynamic batching technique for distributed data-parallel training that adjusts the mini-batch sizes on each worker based on its resource availability and throughput. Our mini-batch controller seeks to equalize iteration times on all workers, and facilitates training on clusters comprised of servers with different amounts of CPU and GPU resources. This variable mini-batch technique uses proportional control and ideas from PID controllers to find stable mini-batch sizes. Our empirical evaluation shows that dynamic batching can reduce model training times by more than 4x on heterogeneous clusters.

arXiv:2305.12356

Integer or Floating Point? New Outlooks for Low-Bit Quantization on Large Language Models

Efficient deployment of large language models (LLMs) necessitates low-bit quantization to minimize model size and inference cost. While low-bit integer formats (e.g., INT8/INT4) have been the conventional choice, emerging low-bit floating-point formats (e.g., FP8/FP4) offer a compelling alternative and are gaining support from cutting-edge hardware, such as NVIDIA's H100 GPU. However, the superiority of low-bit INT versus FP formats for quantization on LLMs remains unclear. In this study, we conduct a comparative analysis of INT and FP quantization with the same bit-width, revealing that the optimal quantization format varies across different layers due to the complexity and diversity of tensor distribution. Consequently, we advocate the Mixture of Formats Quantization (MoFQ), which selects the optimal format on a layer-wise basis. This simple yet effective approach achieves state-of-the-art results in both weight-only (W-only) and weight-activation (WA) post-training quantization scenarios when tested on LLaMA across various tasks. In 4-bit W-only quantization, MoFQ surpasses GPTQ without complex hyperparameter tuning and with an order of magnitude faster quantization speed. While in 8-bit WA quantization, MoFQ significantly outperforms INT/FP-only methods, achieving performance close to the full precision model. Notably, MoFQ incurs no hardware overhead compared to INT/FP-only quantization, as the bit-width remains unchanged.

arXiv:2305.12127

DexPBT: Scaling up Dexterous Manipulation for Hand-Arm Systems with Population Based Training

In this work, we propose algorithms and methods that enable learning dexterous object manipulation using simulated one- or two-armed robots equipped with multi-fingered hand end-effectors. Using a parallel GPU-accelerated physics simulator (Isaac Gym), we implement challenging tasks for these robots, including regrasping, grasp-and-throw, and object reorientation. To solve these problems we introduce a decentralized Population-Based Training (PBT) algorithm that allows us to massively amplify the exploration capabilities of deep reinforcement learning. We find that this method significantly outperforms regular end-to-end learning and is able to discover robust control policies in challenging tasks. Video demonstrations of learned behaviors and the code can be found at https://sites.google.com/view/dexpbt

arXiv:2305.12031공식 코드 · Papers with Code

Clinical Camel: An Open Expert-Level Medical Language Model with Dialogue-Based Knowledge Encoding

We present Clinical Camel, an open large language model (LLM) explicitly tailored for clinical research. Fine-tuned from LLaMA-2 using QLoRA, Clinical Camel achieves state-of-the-art performance across medical benchmarks among openly available medical LLMs. Leveraging efficient single-GPU training, Clinical Camel surpasses GPT-3.5 in five-shot evaluations on all assessed benchmarks, including 64.3% on the USMLE Sample Exam (compared to 58.5% for GPT-3.5), 77.9% on PubMedQA (compared to 60.2%), 60.7% on MedQA (compared to 53.6%), and 54.2% on MedMCQA (compared to 51.0%). In addition to these benchmarks, Clinical Camel demonstrates its broader capabilities, such as synthesizing plausible clinical notes. This work introduces dialogue-based knowledge encoding, a novel method to synthesize conversational data from dense medical texts. While benchmark results are encouraging, extensive and rigorous human evaluation across diverse clinical scenarios is imperative to ascertain safety before implementation. By openly sharing Clinical Camel, we hope to foster transparent and collaborative research, working towards the safe integration of LLMs within the healthcare domain. Significant challenges concerning reliability, bias, and the potential for outdated knowledge persist. Nonetheless, the transparency provided by an open approach reinforces the scientific rigor essential for future clinical applications.

arXiv:2305.11186

Compress, Then Prompt: Improving Accuracy-Efficiency Trade-off of LLM Inference with Transferable Prompt

While the numerous parameters in Large Language Models (LLMs) contribute to their superior performance, this massive scale makes them inefficient and memory-hungry. Thus, they are hard to deploy on commodity hardware, such as one single GPU. Given the memory and power constraints of such devices, model compression methods are widely employed to reduce both the model size and inference latency, which essentially trades off model quality in return for improved efficiency. Thus, optimizing this accuracy-efficiency trade-off is crucial for the LLM deployment on commodity hardware. In this paper, we introduce a new perspective to optimize this trade-off by prompting compressed models. Specifically, we first observe that for certain questions, the generation quality of a compressed LLM can be significantly improved by adding carefully designed hard prompts, though this isn't the case for all questions. Based on this observation, we propose a soft prompt learning method where we expose the compressed model to the prompt learning process, aiming to enhance the performance of prompts. Our experimental analysis suggests our soft prompt strategy greatly improves the performance of the 8x compressed LLaMA-7B model (with a joint 4-bit quantization and 50% weight pruning compression), allowing them to match their uncompressed counterparts on popular benchmarks. Also, we demonstrate that these learned prompts can be transferred across various datasets, tasks, and compression levels. Hence with this transferability, we can stitch the soft prompt to a newly compressed model to improve the test-time accuracy in an ``in-situ'' way.

arXiv:2305.11805공식 코드 · Papers with Code

PANNA 2.0: Efficient neural network interatomic potentials and new architectures

We present the latest release of PANNA 2.0 (Properties from Artificial Neural Network Architectures), a code for the generation of neural network interatomic potentials based on local atomic descriptors and multilayer perceptrons. Built on a new back end, this new release of PANNA features improved tools for customizing and monitoring network training, better GPU support including a fast descriptor calculator, new plugins for external codes and a new architecture for the inclusion of long-range electrostatic interactions through a variational charge equilibration scheme. We present an overview of the main features of the new code, and several benchmarks comparing the accuracy of PANNA models to the state of the art, on commonly used benchmarks as well as richer datasets.

arXiv:2305.11419공식 코드 · Papers with Code

JetSeg: Efficient Real-Time Semantic Segmentation Model for Low-Power GPU-Embedded Systems

Real-time semantic segmentation is a challenging task that requires high-accuracy models with low-inference times. Implementing these models on embedded systems is limited by hardware capability and memory usage, which produces bottlenecks. We propose an efficient model for real-time semantic segmentation called JetSeg, consisting of an encoder called JetNet, and an improved RegSeg decoder. The JetNet is designed for GPU-Embedded Systems and includes two main components: a new light-weight efficient block called JetBlock, that reduces the number of parameters minimizing memory usage and inference time without sacrificing accuracy; a new strategy that involves the combination of asymmetric and non-asymmetric convolutions with depthwise-dilated convolutions called JetConv, a channel shuffle operation, light-weight activation functions, and a convenient number of group convolutions for embedded systems, and an innovative loss function named JetLoss, which integrates the Precision, Recall, and IoUB losses to improve semantic segmentation and reduce computational complexity. Experiments demonstrate that JetSeg is much faster on workstation devices and more suitable for Low-Power GPU-Embedded Systems than existing state-of-the-art models for real-time semantic segmentation. Our approach outperforms state-of-the-art real-time encoder-decoder models by reducing 46.70M parameters and 5.14% GFLOPs, which makes JetSeg up to 2x faster on the NVIDIA Titan RTX GPU and the Jetson Xavier than other models. The JetSeg code is available at https://github.com/mmontielpz/jetseg.

arXiv:2305.11624공식 코드 · Papers with Code

Efficient ConvBN Blocks for Transfer Learning and Beyond

Convolution-BatchNorm (ConvBN) blocks are integral components in various computer vision tasks and other domains. A ConvBN block can operate in three modes: Train, Eval, and Deploy. While the Train mode is indispensable for training models from scratch, the Eval mode is suitable for transfer learning and beyond, and the Deploy mode is designed for the deployment of models. This paper focuses on the trade-off between stability and efficiency in ConvBN blocks: Deploy mode is efficient but suffers from training instability; Eval mode is widely used in transfer learning but lacks efficiency. To solve the dilemma, we theoretically reveal the reason behind the diminished training stability observed in the Deploy mode. Subsequently, we propose a novel Tune mode to bridge the gap between Eval mode and Deploy mode. The proposed Tune mode is as stable as Eval mode for transfer learning, and its computational efficiency closely matches that of the Deploy mode. Through extensive experiments in object detection, classification, and adversarial example generation across $5$ datasets and $12$ model architectures, we demonstrate that the proposed Tune mode retains the performance while significantly reducing GPU memory footprint and training time, thereby contributing efficient ConvBN blocks for transfer learning and beyond. Our method has been integrated into both PyTorch (general machine learning framework) and MMCV/MMEngine (computer vision framework). Practitioners just need one line of code to enjoy our efficient ConvBN blocks thanks to PyTorch's builtin machine learning compilers.

arXiv:2305.11072공식 코드 · Papers with Code

Self-supervised Fine-tuning for Improved Content Representations by Speaker-invariant Clustering

Self-supervised speech representation models have succeeded in various tasks, but improving them for content-related problems using unlabeled data is challenging. We propose speaker-invariant clustering (Spin), a novel self-supervised learning method that clusters speech representations and performs swapped prediction between the original and speaker-perturbed utterances. Spin disentangles speaker information and preserves content representations with just 45 minutes of fine-tuning on a single GPU. Spin improves pre-trained networks and outperforms prior methods in speech recognition and acoustic unit discovery.

arXiv:2305.10611

ACRoBat: Optimizing Auto-batching of Dynamic Deep Learning at Compile Time

Dynamic control flow is an important technique often used to design expressive and efficient deep learning computations for applications such as text parsing, machine translation, exiting early out of deep models and so on. The control flow divergence resulting from dynamic control flow makes batching, an important optimization enabling high throughput and hardware utilization, difficult to perform manually. In this paper, we present ACRoBat, a framework that enables efficient automatic batching for dynamic deep learning computations by performing hybrid static+dynamic compiler optimizations and end-to-end tensor code generation. ACRoBat performs up to 8.5X better than DyNet, a state-of-the-art framework for automatic batching, on an Nvidia GeForce GPU.

arXiv:2305.10727

Boost Vision Transformer with GPU-Friendly Sparsity and Quantization

The transformer extends its success from the language to the vision domain. Because of the stacked self-attention and cross-attention blocks, the acceleration deployment of vision transformer on GPU hardware is challenging and also rarely studied. This paper thoroughly designs a compression scheme to maximally utilize the GPU-friendly 2:4 fine-grained structured sparsity and quantization. Specially, an original large model with dense weight parameters is first pruned into a sparse one by 2:4 structured pruning, which considers the GPU's acceleration of 2:4 structured sparse pattern with FP16 data type, then the floating-point sparse model is further quantized into a fixed-point one by sparse-distillation-aware quantization aware training, which considers GPU can provide an extra speedup of 2:4 sparse calculation with integer tensors. A mixed-strategy knowledge distillation is used during the pruning and quantization process. The proposed compression scheme is flexible to support supervised and unsupervised learning styles. Experiment results show GPUSQ-ViT scheme achieves state-of-the-art compression by reducing vision transformer models 6.4-12.7 times on model size and 30.3-62 times on FLOPs with negligible accuracy degradation on ImageNet classification, COCO detection and ADE20K segmentation benchmarking tasks. Moreover, GPUSQ-ViT can boost actual deployment performance by 1.39-1.79 times and 3.22-3.43 times of latency and throughput on A100 GPU, and 1.57-1.69 times and 2.11-2.51 times improvement of latency and throughput on AGX Orin.

arXiv:2305.10863공식 코드 · Papers with Code

Quiver: Supporting GPUs for Low-Latency, High-Throughput GNN Serving with Workload Awareness

Systems for serving inference requests on graph neural networks (GNN) must combine low latency with high throughout, but they face irregular computation due to skew in the number of sampled graph nodes and aggregated GNN features. This makes it challenging to exploit GPUs effectively: using GPUs to sample only a few graph nodes yields lower performance than CPU-based sampling; and aggregating many features exhibits high data movement costs between GPUs and CPUs. Therefore, current GNN serving systems use CPUs for graph sampling and feature aggregation, limiting throughput. We describe Quiver, a distributed GPU-based GNN serving system with low-latency and high-throughput. Quiver's key idea is to exploit workload metrics for predicting the irregular computation of GNN requests, and governing the use of GPUs for graph sampling and feature aggregation: (1) for graph sampling, Quiver calculates the probabilistic sampled graph size, a metric that predicts the degree of parallelism in graph sampling. Quiver uses this metric to assign sampling tasks to GPUs only when the performance gains surpass CPU-based sampling; and (2) for feature aggregation, Quiver relies on the feature access probability to decide which features to partition and replicate across a distributed GPU NUMA topology. We show that Quiver achieves up to 35 times lower latency with an 8 times higher throughput compared to state-of-the-art GNN approaches (DGL and PyG).

arXiv:2305.10006공식 코드 · Papers with Code

EfficientSCI: Densely Connected Network with Space-time Factorization for Large-scale Video Snapshot Compressive Imaging

Video snapshot compressive imaging (SCI) uses a two-dimensional detector to capture consecutive video frames during a single exposure time. Following this, an efficient reconstruction algorithm needs to be designed to reconstruct the desired video frames. Although recent deep learning-based state-of-the-art (SOTA) reconstruction algorithms have achieved good results in most tasks, they still face the following challenges due to excessive model complexity and GPU memory limitations: 1) these models need high computational cost, and 2) they are usually unable to reconstruct large-scale video frames at high compression ratios. To address these issues, we develop an efficient network for video SCI by using dense connections and space-time factorization mechanism within a single residual block, dubbed EfficientSCI. The EfficientSCI network can well establish spatial-temporal correlation by using convolution in the spatial domain and Transformer in the temporal domain, respectively. We are the first time to show that an UHD color video with high compression ratio can be reconstructed from a snapshot 2D measurement using a single end-to-end deep learning model with PSNR above 32 dB. Extensive results on both simulation and real data show that our method significantly outperforms all previous SOTA algorithms with better real-time performance. The code is at https://github.com/ucaswangls/EfficientSCI.git.

arXiv:2305.09559

Robust and lightweight audio fingerprint for Automatic Content Recognition

This research paper presents a novel audio fingerprinting system for Automatic Content Recognition (ACR). By using signal processing techniques and statistical transformations, our proposed method generates compact fingerprints of audio segments that are robust to noise degradations present in real-world audio. The system is designed to be highly scalable, with the ability to identify thousands of hours of content using fingerprints generated from millions of TVs. The fingerprint's high temporal correlation and utilization of existing GPU-compatible Approximate Nearest Neighbour (ANN) search algorithms make this possible. Furthermore, the fingerprint generation can run on low-power devices with limited compute, making it accessible to a wide range of applications. Experimental results show improvements in our proposed system compared to a min-hash based audio fingerprint on all evaluated metrics, including accuracy on proprietary ACR datasets, retrieval speed, memory usage, and robustness to various noises. For similar retrieval accuracy, our system is 30x faster and uses 6x fewer fingerprints than the min-hash method.

arXiv:2305.09041공식 코드 · Papers with Code

What Matters in Reinforcement Learning for Tractography

Recently, deep reinforcement learning (RL) has been proposed to learn the tractography procedure and train agents to reconstruct the structure of the white matter without manually curated reference streamlines. While the performances reported were competitive, the proposed framework is complex, and little is still known about the role and impact of its multiple parts. In this work, we thoroughly explore the different components of the proposed framework, such as the choice of the RL algorithm, seeding strategy, the input signal and reward function, and shed light on their impact. Approximately 7,400 models were trained for this work, totalling nearly 41,000 hours of GPU time. Our goal is to guide researchers eager to explore the possibilities of deep RL for tractography by exposing what works and what does not work with the category of approach. As such, we ultimately propose a series of recommendations concerning the choice of RL algorithm, the input to the agents, the reward function and more to help future work using reinforcement learning for tractography. We also release the open source codebase, trained models, and datasets for users and researchers wanting to explore reinforcement learning for tractography.

arXiv:2305.08318

CMSG Cross-Media Semantic-Graph Feature Matching Algorithm for Autonomous Vehicle Relocalization

Relocalization is the basis of map-based localization algorithms. Camera and LiDAR map-based methods are pervasive since their robustness under different scenarios. Generally, mapping and localization using the same sensor have better accuracy since matching features between the same type of data is easier. However, due to the camera's lack of 3D information and the high cost of LiDAR, cross-media methods are developing, which combined live image data and Lidar map. Although matching features between different media is challenging, we believe cross-media is the tendency for AV relocalization since its low cost and accuracy can be comparable to the same-sensor-based methods. In this paper, we propose CMSG, a novel cross-media algorithm for AV relocalization tasks. Semantic features are utilized for better interpretation the correlation between point clouds and image features. What's more, abstracted semantic graph nodes are introduced, and a graph network architecture is integrated to better extract the similarity of semantic features. Validation experiments are conducted on the KITTI odometry dataset. Our results show that CMSG can have comparable or even better accuracy compared to current single-sensor-based methods at a speed of 25 FPS on NVIDIA 1080 Ti GPU.

arXiv:2305.08040공식 코드 · Papers with Code

Provable Multi-instance Deep AUC Maximization with Stochastic Pooling

This paper considers a novel application of deep AUC maximization (DAM) for multi-instance learning (MIL), in which a single class label is assigned to a bag of instances (e.g., multiple 2D slices of a CT scan for a patient). We address a neglected yet non-negligible computational challenge of MIL in the context of DAM, i.e., bag size is too large to be loaded into {GPU} memory for backpropagation, which is required by the standard pooling methods of MIL. To tackle this challenge, we propose variance-reduced stochastic pooling methods in the spirit of stochastic optimization by formulating the loss function over the pooled prediction as a multi-level compositional function. By synthesizing techniques from stochastic compositional optimization and non-convex min-max optimization, we propose a unified and provable muli-instance DAM (MIDAM) algorithm with stochastic smoothed-max pooling or stochastic attention-based pooling, which only samples a few instances for each bag to compute a stochastic gradient estimator and to update the model parameter. We establish a similar convergence rate of the proposed MIDAM algorithm as the state-of-the-art DAM algorithms. Our extensive experiments on conventional MIL datasets and medical datasets demonstrate the superiority of our MIDAM algorithm.

arXiv:2305.08285공식 코드 · Papers with Code

Parameter-Efficient Fine-Tuning with Layer Pruning on Free-Text Sequence-to-Sequence Modeling

The increasing size of language models raises great research interests in parameter-efficient fine-tuning such as LoRA that freezes the pre-trained model, and injects small-scale trainable parameters for multiple downstream tasks (e.g., summarization, question answering and translation). To further enhance the efficiency of fine-tuning, we propose a framework that integrates LoRA and structured layer pruning. The integrated framework is validated on two created deidentified medical report summarization datasets based on MIMIC-IV-Note and two public medical dialogue datasets. By tuning 0.6% parameters of the original model and pruning over 30% Transformer-layers, our framework can reduce 50% of GPU memory usage and speed up 100% of the training phase, while preserving over 92% generation qualities on free-text sequence-to-sequence tasks.