PyTorchKR Papers

최신 논문

총 1,512편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2306.06625공식 코드 · Papers with Code

Are Intermediate Layers and Labels Really Necessary? A General Language Model Distillation Method

The large scale of pre-trained language models poses a challenge for their deployment on various devices, with a growing emphasis on methods to compress these models, particularly knowledge distillation. However, current knowledge distillation methods rely on the model's intermediate layer features and the golden labels (also called hard labels), which usually require aligned model architecture and enough labeled data respectively. Moreover, the parameters of vocabulary are usually neglected in existing methods. To address these problems, we propose a general language model distillation (GLMD) method that performs two-stage word prediction distillation and vocabulary compression, which is simple and surprisingly shows extremely strong performance. Specifically, GLMD supports more general application scenarios by eliminating the constraints of dimension and structure between models and the need for labeled datasets through the absence of intermediate layers and golden labels. Meanwhile, based on the long-tailed distribution of word frequencies in the data, GLMD designs a strategy of vocabulary compression through decreasing vocabulary size instead of dimensionality. Experimental results show that our method outperforms 25 state-of-the-art methods on the SuperGLUE benchmark, achieving an average score that surpasses the best method by 3%.

arXiv:2306.05808공식 코드 · Papers with Code

RankFormer: Listwise Learning-to-Rank Using Listwide Labels

Web applications where users are presented with a limited selection of items have long employed ranking models to put the most relevant results first. Any feedback received from users is typically assumed to reflect a relative judgement on the utility of items, e.g. a user clicking on an item only implies it is better than items not clicked in the same ranked list. Hence, the objectives optimized in Learning-to-Rank (LTR) tend to be pairwise or listwise. Yet, by only viewing feedback as relative, we neglect the user's absolute feedback on the list's overall quality, e.g. when no items in the selection are clicked. We thus reconsider the standard LTR paradigm and argue the benefits of learning from this listwide signal. To this end, we propose the RankFormer as an architecture that, with a Transformer at its core, can jointly optimize a novel listwide assessment objective and a traditional listwise LTR objective. We simulate implicit feedback on public datasets and observe that the RankFormer succeeds in benefitting from listwide signals. Additionally, we conduct experiments in e-commerce on Amazon Search data and find the RankFormer to be superior to all baselines offline. An online experiment shows that knowledge distillation can be used to find immediate practical use for the RankFormer.

arXiv:2306.04186공식 코드 · Papers with Code

Self-supervised Audio Teacher-Student Transformer for Both Clip-level and Frame-level Tasks

Self-supervised learning (SSL) has emerged as a popular approach for learning audio representations. One goal of audio self-supervised pre-training is to transfer knowledge to downstream audio tasks, generally including clip-level and frame-level tasks. While frame-level tasks are important for fine-grained acoustic scene/event understanding, prior studies primarily evaluate on clip-level downstream tasks. In order to tackle both clip-level and frame-level tasks, this paper proposes Audio Teacher-Student Transformer (ATST), with a clip-level version (named ATST-Clip) and a frame-level version (named ATST-Frame), responsible for learning clip-level and frame-level representations, respectively. Both methods use a Transformer encoder and a teacher-student training scheme. We have carefully designed the view creation strategy for ATST-Clip and ATST-Frame. Specifically, ATST-Clip uses segment-wise data augmentations, and ATST-Frame integrates frame-wise data augmentations and masking. Experimental results show that our ATST-Frame model obtains state-of-the-art (SOTA) performances on most of the clip-level and frame-level downstream tasks. Especially, it outperforms other models by a large margin on the frame-level sound event detection task. In addition, the performance can be further improved by combining the two models through knowledge distillation. Our code is available online.

arXiv:2306.03360공식 코드 · Papers with Code

Model-Based Reinforcement Learning with Multi-Task Offline Pretraining

Pretraining reinforcement learning (RL) models on offline datasets is a promising way to improve their training efficiency in online tasks, but challenging due to the inherent mismatch in dynamics and behaviors across various tasks. We present a model-based RL method that learns to transfer potentially useful dynamics and action demonstrations from offline data to a novel task. The main idea is to use the world models not only as simulators for behavior learning but also as tools to measure the task relevance for both dynamics representation transfer and policy transfer. We build a time-varying, domain-selective distillation loss to generate a set of offline-to-online similarity weights. These weights serve two purposes: (i) adaptively transferring the task-agnostic knowledge of physical dynamics to facilitate world model training, and (ii) learning to replay relevant source actions to guide the target policy. We demonstrate the advantages of our approach compared with the state-of-the-art methods in Meta-World and DeepMind Control Suite.

arXiv:2306.02371공식 코드 · Papers with Code

I^3 Retriever: Incorporating Implicit Interaction in Pre-trained Language Models for Passage Retrieval

Passage retrieval is a fundamental task in many information systems, such as web search and question answering, where both efficiency and effectiveness are critical concerns. In recent years, neural retrievers based on pre-trained language models (PLM), such as dual-encoders, have achieved huge success. Yet, studies have found that the performance of dual-encoders are often limited due to the neglecting of the interaction information between queries and candidate passages. Therefore, various interaction paradigms have been proposed to improve the performance of vanilla dual-encoders. Particularly, recent state-of-the-art methods often introduce late-interaction during the model inference process. However, such late-interaction based methods usually bring extensive computation and storage cost on large corpus. Despite their effectiveness, the concern of efficiency and space footprint is still an important factor that limits the application of interaction-based neural retrieval models. To tackle this issue, we incorporate implicit interaction into dual-encoders, and propose I^3 retriever. In particular, our implicit interaction paradigm leverages generated pseudo-queries to simulate query-passage interaction, which jointly optimizes with query and passage encoders in an end-to-end manner. It can be fully pre-computed and cached, and its inference process only involves simple dot product operation of the query vector and passage vector, which makes it as efficient as the vanilla dual encoders. We conduct comprehensive experiments on MSMARCO and TREC2019 Deep Learning Datasets, demonstrating the I^3 retriever's superiority in terms of both effectiveness and efficiency. Moreover, the proposed implicit interaction is compatible with special pre-training and knowledge distillation for passage retrieval, which brings a new state-of-the-art performance.

arXiv:2306.02368공식 코드 · Papers with Code

Revisiting Data-Free Knowledge Distillation with Poisoned Teachers

Data-free knowledge distillation (KD) helps transfer knowledge from a pre-trained model (known as the teacher model) to a smaller model (known as the student model) without access to the original training data used for training the teacher model. However, the security of the synthetic or out-of-distribution (OOD) data required in data-free KD is largely unknown and under-explored. In this work, we make the first effort to uncover the security risk of data-free KD w.r.t. untrusted pre-trained models. We then propose Anti-Backdoor Data-Free KD (ABD), the first plug-in defensive method for data-free KD methods to mitigate the chance of potential backdoors being transferred. We empirically evaluate the effectiveness of our proposed ABD in diminishing transferred backdoor knowledge while maintaining compatible downstream performances as the vanilla KD. We envision this work as a milestone for alarming and mitigating the potential backdoors in data-free KD. Codes are released at https://github.com/illidanlab/ABD.

arXiv:2306.02679공식 코드 · Papers with Code

Joint Pre-training and Local Re-training: Transferable Representation Learning on Multi-source Knowledge Graphs

In this paper, we present the ``joint pre-training and local re-training'' framework for learning and applying multi-source knowledge graph (KG) embeddings. We are motivated by the fact that different KGs contain complementary information to improve KG embeddings and downstream tasks. We pre-train a large teacher KG embedding model over linked multi-source KGs and distill knowledge to train a student model for a task-specific KG. To enable knowledge transfer across different KGs, we use entity alignment to build a linked subgraph for connecting the pre-trained KGs and the target KG. The linked subgraph is re-trained for three-level knowledge distillation from the teacher to the student, i.e., feature knowledge distillation, network knowledge distillation, and prediction knowledge distillation, to generate more expressive embeddings. The teacher model can be reused for different target KGs and tasks without having to train from scratch. We conduct extensive experiments to demonstrate the effectiveness and efficiency of our framework.

arXiv:2306.02027공식 코드 · Papers with Code

Evolving Knowledge Mining for Class Incremental Segmentation

Class Incremental Semantic Segmentation (CISS) has been a trend recently due to its great significance in real-world applications. Although the existing CISS methods demonstrate remarkable performance, they either leverage the high-level knowledge (feature) only while neglecting the rich and diverse knowledge in the low-level features, leading to poor old knowledge preservation and weak new knowledge exploration; or use multi-level features for knowledge distillation by retraining a heavy backbone, which is computationally intensive. In this paper, we for the first time investigate the efficient multi-grained knowledge reuse for CISS, and propose a novel method, Evolving kNowleDge minING (ENDING), employing a frozen backbone. ENDING incorporates two key modules: evolving fusion and semantic enhancement, for dynamic and comprehensive exploration of multi-grained knowledge. Evolving fusion is tailored to extract knowledge from individual low-level feature using a personalized lightweight network, which is generated from a meta-net, taking the high-level feature as input. This design enables the evolution of knowledge mining and fusing when applied to incremental new classes. In contrast, semantic enhancement is specifically crafted to aggregate prototype-based semantics from multi-level features, contributing to an enhanced representation. We evaluate our method on two widely used benchmarks and consistently demonstrate new state-of-the-art performance. The code is available at https://github.com/zhiheLu/ENDING_ISS.

arXiv:2306.02090공식 코드 · Papers with Code

Deep Classifier Mimicry without Data Access

Access to pre-trained models has recently emerged as a standard across numerous machine learning domains. Unfortunately, access to the original data the models were trained on may not equally be granted. This makes it tremendously challenging to fine-tune, compress models, adapt continually, or to do any other type of data-driven update. We posit that original data access may however not be required. Specifically, we propose Contrastive Abductive Knowledge Extraction (CAKE), a model-agnostic knowledge distillation procedure that mimics deep classifiers without access to the original data. To this end, CAKE generates pairs of noisy synthetic samples and diffuses them contrastively toward a model's decision boundary. We empirically corroborate CAKE's effectiveness using several benchmark datasets and various architectural choices, paving the way for broad application.

arXiv:2306.01526공식 코드 · Papers with Code

Group channel pruning and spatial attention distilling for object detection

Due to the over-parameterization of neural networks, many model compression methods based on pruning and quantization have emerged. They are remarkable in reducing the size, parameter number, and computational complexity of the model. However, most of the models compressed by such methods need the support of special hardware and software, which increases the deployment cost. Moreover, these methods are mainly used in classification tasks, and rarely directly used in detection tasks. To address these issues, for the object detection network we introduce a three-stage model compression method: dynamic sparse training, group channel pruning, and spatial attention distilling. Firstly, to select out the unimportant channels in the network and maintain a good balance between sparsity and accuracy, we put forward a dynamic sparse training method, which introduces a variable sparse rate, and the sparse rate will change with the training process of the network. Secondly, to reduce the effect of pruning on network accuracy, we propose a novel pruning method called group channel pruning. In particular, we divide the network into multiple groups according to the scales of the feature layer and the similarity of module structure in the network, and then we use different pruning thresholds to prune the channels in each group. Finally, to recover the accuracy of the pruned network, we use an improved knowledge distillation method for the pruned network. Especially, we extract spatial attention information from the feature maps of specific scales in each group as knowledge for distillation. In the experiments, we use YOLOv4 as the object detection network and PASCAL VOC as the training dataset. Our method reduces the parameters of the model by 64.7 % and the calculation by 34.9%.

arXiv:2306.00393공식 코드 · Papers with Code

Teacher Agent: A Knowledge Distillation-Free Framework for Rehearsal-based Video Incremental Learning

Rehearsal-based video incremental learning often employs knowledge distillation to mitigate catastrophic forgetting of previously learned data. However, this method faces two major challenges for video task: substantial computing resources from loading teacher model and limited replay capability from performance-limited teacher model. To address these problems, we first propose a knowledge distillation-free framework for rehearsal-based video incremental learning called \textit{Teacher Agent}. Instead of loading parameter-heavy teacher networks, we introduce an agent generator that is either parameter-free or uses only a few parameters to obtain accurate and reliable soft labels. This method not only greatly reduces the computing requirement but also circumvents the problem of knowledge misleading caused by inaccurate predictions of the teacher model. Moreover, we put forward a self-correction loss which provides an effective regularization signal for the review of old knowledge, which in turn alleviates the problem of catastrophic forgetting. Further, to ensure that the samples in the memory buffer are memory-efficient and representative, we introduce a unified sampler for rehearsal-based video incremental learning to mine fixed-length key video frames. Interestingly, based on the proposed strategies, the network exhibits a high level of robustness against spatial resolution reduction when compared to the baseline. Extensive experiments demonstrate the advantages of our method, yielding significant performance improvements while utilizing only half the spatial resolution of video clips as network inputs in the incremental phases.

arXiv:2305.19502공식 코드 · Papers with Code

Graph Entropy Minimization for Semi-supervised Node Classification

Node classifiers are required to comprehensively reduce prediction errors, training resources, and inference latency in the industry. However, most graph neural networks (GNN) concentrate only on one or two of them. The compromised aspects thus are the shortest boards on the bucket, hindering their practical deployments for industrial-level tasks. This work proposes a novel semi-supervised learning method termed Graph Entropy Minimization (GEM) to resolve the three issues simultaneously. GEM benefits its one-hop aggregation from massive uncategorized nodes, making its prediction accuracy comparable to GNNs with two or more hops message passing. It can be decomposed to support stochastic training with mini-batches of independent edge samples, achieving extremely fast sampling and space-saving training. While its one-hop aggregation is faster in inference than deep GNNs, GEM can be further accelerated to an extreme by deriving a non-hop classifier via online knowledge distillation. Thus, GEM can be a handy choice for latency-restricted and error-sensitive services running on resource-constraint hardware. Code is available at https://github.com/cf020031308/GEM.

arXiv:2305.19472공식 코드 · Papers with Code

PlaSma: Making Small Language Models Better Procedural Knowledge Models for (Counterfactual) Planning

Procedural planning, which entails decomposing a high-level goal into a sequence of temporally ordered steps, is an important yet intricate task for machines. It involves integrating common-sense knowledge to reason about complex and often contextualized situations, e.g. ``scheduling a doctor's appointment without a phone''. While current approaches show encouraging results using large language models (LLMs), they are hindered by drawbacks such as costly API calls and reproducibility issues. In this paper, we advocate planning using smaller language models. We present PlaSma, a novel two-pronged approach to endow small language models with procedural knowledge and (constrained) language planning capabilities. More concretely, we develop symbolic procedural knowledge distillation to enhance the commonsense knowledge in small language models and an inference-time algorithm to facilitate more structured and accurate reasoning. In addition, we introduce a new related task, Replanning, that requires a revision of a plan to cope with a constrained situation. In both the planning and replanning settings, we show that orders-of-magnitude smaller models (770M-11B parameters) can compete and often surpass their larger teacher models' capabilities. Finally, we showcase successful application of PlaSma in an embodied environment, VirtualHome.

arXiv:2305.19412공식 코드 · Papers with Code

Are Large Kernels Better Teachers than Transformers for ConvNets?

This paper reveals a new appeal of the recently emerged large-kernel Convolutional Neural Networks (ConvNets): as the teacher in Knowledge Distillation (KD) for small-kernel ConvNets. While Transformers have led state-of-the-art (SOTA) performance in various fields with ever-larger models and labeled data, small-kernel ConvNets are considered more suitable for resource-limited applications due to the efficient convolution operation and compact weight sharing. KD is widely used to boost the performance of small-kernel ConvNets. However, previous research shows that it is not quite effective to distill knowledge (e.g., global information) from Transformers to small-kernel ConvNets, presumably due to their disparate architectures. We hereby carry out a first-of-its-kind study unveiling that modern large-kernel ConvNets, a compelling competitor to Vision Transformers, are remarkably more effective teachers for small-kernel ConvNets, due to more similar architectures. Our findings are backed up by extensive experiments on both logit-level and feature-level KD ``out of the box", with no dedicated architectural nor training recipe modifications. Notably, we obtain the \textbf{best-ever pure ConvNet} under 30M parameters with \textbf{83.1\%} top-1 accuracy on ImageNet, outperforming current SOTA methods including ConvNeXt V2 and Swin V2. We also find that beneficial characteristics of large-kernel ConvNets, e.g., larger effective receptive fields, can be seamlessly transferred to students through this large-to-small kernel distillation. Code is available at: \url{https://github.com/VITA-Group/SLaK}.

arXiv:2305.18464공식 코드 · Papers with Code

Bridging the Sim-to-Real Gap from the Information Bottleneck Perspective

Reinforcement Learning (RL) has recently achieved remarkable success in robotic control. However, most works in RL operate in simulated environments where privileged knowledge (e.g., dynamics, surroundings, terrains) is readily available. Conversely, in real-world scenarios, robot agents usually rely solely on local states (e.g., proprioceptive feedback of robot joints) to select actions, leading to a significant sim-to-real gap. Existing methods address this gap by either gradually reducing the reliance on privileged knowledge or performing a two-stage policy imitation. However, we argue that these methods are limited in their ability to fully leverage the available privileged knowledge, resulting in suboptimal performance. In this paper, we formulate the sim-to-real gap as an information bottleneck problem and therefore propose a novel privileged knowledge distillation method called the Historical Information Bottleneck (HIB). In particular, HIB learns a privileged knowledge representation from historical trajectories by capturing the underlying changeable dynamic information. Theoretical analysis shows that the learned privileged knowledge representation helps reduce the value discrepancy between the oracle and learned policies. Empirical experiments on both simulated and real-world tasks demonstrate that HIB yields improved generalizability compared to previous methods. Videos of real-world experiments are available at https://sites.google.com/view/history-ib .

arXiv:2305.18830공식 코드 · Papers with Code

Semi-supervised Pathological Image Segmentation via Cross Distillation of Multiple Attentions

Segmentation of pathological images is a crucial step for accurate cancer diagnosis. However, acquiring dense annotations of such images for training is labor-intensive and time-consuming. To address this issue, Semi-Supervised Learning (SSL) has the potential for reducing the annotation cost, but it is challenged by a large number of unlabeled training images. In this paper, we propose a novel SSL method based on Cross Distillation of Multiple Attentions (CDMA) to effectively leverage unlabeled images. Firstly, we propose a Multi-attention Tri-branch Network (MTNet) that consists of an encoder and a three-branch decoder, with each branch using a different attention mechanism that calibrates features in different aspects to generate diverse outputs. Secondly, we introduce Cross Decoder Knowledge Distillation (CDKD) between the three decoder branches, allowing them to learn from each other's soft labels to mitigate the negative impact of incorrect pseudo labels in training. Additionally, uncertainty minimization is applied to the average prediction of the three branches, which further regularizes predictions on unlabeled images and encourages inter-branch consistency. Our proposed CDMA was compared with eight state-of-the-art SSL methods on the public DigestPath dataset, and the experimental results showed that our method outperforms the other approaches under different annotation ratios. The code is available at \href{https://github.com/HiLab-git/CDMA}{https://github.com/HiLab-git/CDMA.}

arXiv:2305.18413공식 코드 · Papers with Code

Learning to Learn from APIs: Black-Box Data-Free Meta-Learning

Data-free meta-learning (DFML) aims to enable efficient learning of new tasks by meta-learning from a collection of pre-trained models without access to the training data. Existing DFML work can only meta-learn from (i) white-box and (ii) small-scale pre-trained models (iii) with the same architecture, neglecting the more practical setting where the users only have inference access to the APIs with arbitrary model architectures and model scale inside. To solve this issue, we propose a Bi-level Data-free Meta Knowledge Distillation (BiDf-MKD) framework to transfer more general meta knowledge from a collection of black-box APIs to one single meta model. Specifically, by just querying APIs, we inverse each API to recover its training data via a zero-order gradient estimator and then perform meta-learning via a novel bi-level meta knowledge distillation structure, in which we design a boundary query set recovery technique to recover a more informative query set near the decision boundary. In addition, to encourage better generalization within the setting of limited API budgets, we propose task memory replay to diversify the underlying task distribution by covering more interpolated tasks. Extensive experiments in various real-world scenarios show the superior performance of our BiDf-MKD framework.

arXiv:2305.17394공식 코드 · Papers with Code

One-Step Knowledge Distillation and Fine-Tuning in Using Large Pre-Trained Self-Supervised Learning Models for Speaker Verification

The application of speech self-supervised learning (SSL) models has achieved remarkable performance in speaker verification (SV). However, there is a computational cost hurdle in employing them, which makes development and deployment difficult. Several studies have simply compressed SSL models through knowledge distillation (KD) without considering the target task. Consequently, these methods could not extract SV-tailored features. This paper suggests One-Step Knowledge Distillation and Fine-Tuning (OS-KDFT), which incorporates KD and fine-tuning (FT). We optimize a student model for SV during KD training to avert the distillation of inappropriate information for the SV. OS-KDFT could downsize Wav2Vec 2.0 based ECAPA-TDNN size by approximately 76.2%, and reduce the SSL model's inference time by 79% while presenting an EER of 0.98%. The proposed OS-KDFT is validated across VoxCeleb1 and VoxCeleb2 datasets and W2V2 and HuBERT SSL models. Experiments are available on our GitHub.

arXiv:2305.17651공식 코드 · Papers with Code

DPHuBERT: Joint Distillation and Pruning of Self-Supervised Speech Models

Self-supervised learning (SSL) has achieved notable success in many speech processing tasks, but the large model size and heavy computational cost hinder the deployment. Knowledge distillation trains a small student model to mimic the behavior of a large teacher model. However, the student architecture usually needs to be manually designed and will remain fixed during training, which requires prior knowledge and can lead to suboptimal performance. Inspired by recent success of task-specific structured pruning, we propose DPHuBERT, a novel task-agnostic compression method for speech SSL based on joint distillation and pruning. Experiments on SUPERB show that DPHuBERT outperforms pure distillation methods in almost all tasks. Moreover, DPHuBERT requires little training time and performs well with limited training data, making it suitable for resource-constrained applications. Our method can also be applied to various speech SSL models. Our code and models will be publicly available.

arXiv:2305.17371공식 코드 · Papers with Code

Towards Better Entity Linking with Multi-View Enhanced Distillation

Dense retrieval is widely used for entity linking to retrieve entities from large-scale knowledge bases. Mainstream techniques are based on a dual-encoder framework, which encodes mentions and entities independently and calculates their relevances via rough interaction metrics, resulting in difficulty in explicitly modeling multiple mention-relevant parts within entities to match divergent mentions. Aiming at learning entity representations that can match divergent mentions, this paper proposes a Multi-View Enhanced Distillation (MVD) framework, which can effectively transfer knowledge of multiple fine-grained and mention-relevant parts within entities from cross-encoders to dual-encoders. Each entity is split into multiple views to avoid irrelevant information being over-squashed into the mention-relevant view. We further design cross-alignment and self-alignment mechanisms for this framework to facilitate fine-grained knowledge distillation from the teacher model to the student model. Meanwhile, we reserve a global-view that embeds the entity as a whole to prevent dispersal of uniform information. Experiments show our method achieves state-of-the-art performance on several entity linking benchmarks.