Over the past year, the emergence of transfer learning with large-scale language models (LM) has led to dramatic performance improvements across a broad range of natural language understanding tasks. However, the size and memory footprint of these large LMs makes them difficult to deploy in many scenarios (e.g. on mobile phones). Recent research points to knowledge distillation as a potential solution, showing that when training data for a given task is abundant, it is possible to distill a large (teacher) LM into a small task-specific (student) network with minimal loss of performance. However, when such data is scarce, there remains a significant performance gap between large pretrained LMs and smaller task-specific models, even when training via distillation. In this paper, we bridge this gap with a novel training approach, called generation-distillation, that leverages large finetuned LMs in two ways: (1) to generate new (unlabeled) training examples, and (2) to distill their knowledge into a small network using these examples. Across three low-resource text classification datsets, we achieve comparable performance to BERT while using 300x fewer parameters, and we outperform prior approaches to distillation for text classification while using 3x fewer parameters.
Knowledge Distillation (KD) is a model-agnostic technique to improve model quality while having a fixed capacity budget. It is a commonly used technique for model compression, where a larger capacity teacher model with better quality is used to train a more compact student model with better inference efficiency. Through distillation, one hopes to benefit from student's compactness, without sacrificing too much on model quality. Despite the large success of knowledge distillation, better understanding of how it benefits student model's training dynamics remains under-explored. In this paper, we…
Since deep learning became a key player in natural language processing (NLP), many deep learning models have been showing remarkable performances in a variety of NLP tasks, and in some cases, they are even outperforming humans. Such high performance can be explained by efficient knowledge representation of deep learning models. While many methods have been proposed to learn more efficient representation, knowledge distillation from pretrained deep networks suggest that we can use more information from the soft target probability to train other neural networks. In this paper, we propose a new k…
IST Austria (Institute of Science and Technology Austria)（奥地利科学技术研究所）
Knowledge distillation, i.e., one classifier being trained on the outputs of another classifier, is an empirically very successful technique for knowledge transfer between classifiers. It has even been observed that classifiers learn much faster and more reliably if trained with the outputs of another classifier as soft labels, instead of from ground truth data. So far, however, there is no satisfactory theoretical explanation of this phenomenon. In this work, we provide the first insights into the working mechanisms of distillation by studying the special case of linear and deep linear classi…
Knowledge distillation is widely used as a means of improving the performance of a relatively simple student model using the predictions from a complex teacher model. Several works have shown that distillation significantly boosts the student's overall performance; however, are these gains uniform across all data subgroups? In this paper, we show that distillation can harm performance on certain subgroups, e.g., classes with few associated samples. We trace this behaviour to errors made by the teacher distribution being transferred to and amplified by the student model. To mitigate this proble…
iFLYTEK Research（科大讯飞研究院）；Harbin Institute of Technology（哈尔滨工业大学）；iFLYTEK AI Research (Hebei)（科大讯飞人工智能研究院（河北））
In this paper, we introduce TextBrewer, an open-source knowledge distillation toolkit designed for natural language processing. It works with different neural network models and supports various kinds of supervised learning tasks, such as text classification, reading comprehension, sequence labeling. TextBrewer provides a simple and uniform workflow that enables quick setting up of distillation experiments with highly flexible configurations. It offers a set of predefined distillation methods and can be extended with custom code. As a case study, we use TextBrewer to distill BERT on several ty…
Meta FAIR Labs（Meta FAIR 实验室）；Mohamed bin Zayed University of AI（穆罕默德·本·扎耶德人工智能大学）
We formally study how ensemble of deep learning models can improve test accuracy, and how the superior performance of ensemble can be distilled into a single model using knowledge distillation. We consider the challenging case where the ensemble is simply an average of the outputs of a few independently trained neural networks with the SAME architecture, trained using the SAME algorithm on the SAME data set, and they only differ by the random seeds used in the initialization. We show that ensemble/knowledge distillation in Deep Learning works very differently from traditional learning theory (…
Huawei Noah’s Ark Lab（华为诺亚方舟实验室）；University of British Columbia（不列颠哥伦比亚大学）
Knowledge Distillation (KD) is a common knowledge transfer algorithm used for model compression across a variety of deep learning based natural language processing (NLP) solutions. In its regular manifestations, KD requires access to the teacher's training data for knowledge transfer to the student network. However, privacy concerns, data regulations and proprietary reasons may prevent access to such data. We present, to the best of our knowledge, the first work on Zero-Shot Knowledge Distillation for NLP, where the student learns from the much larger teacher without any task specific data. Ou…
New Jersey Institute of Technology（新泽西理工学院）；Microsoft Research（微软研究院）；Rice University（莱斯大学）
Recent work has focused on compressing pre-trained language models (PLMs) like BERT where the major focus has been to improve the in-distribution performance for downstream tasks. However, very few of these studies have analyzed the impact of compression on the generalizability and robustness of compressed models for out-of-distribution (OOD) data. Towards this end, we study two popular model compression techniques including knowledge distillation and pruning and show that the compressed models are significantly less robust than their PLM counterparts on OOD test sets although they obtain simi…
End-to-end approaches open a new way for more accurate and efficient spoken language understanding (SLU) systems by alleviating the drawbacks of traditional pipeline systems. Previous works exploit textual information for an SLU model via pre-training with automatic speech recognition or fine-tuning with knowledge distillation. To utilize textual information more effectively, this work proposes a two-stage textual knowledge distillation method that matches utterance-level representations and predicted logits of two modalities during pre-training and fine-tuning, sequentially. We use vq-wav2vec…
Arm Research（Arm研究院）；Arizona State University（亚利桑那州立大学）
The success of deep learning has brought forth a wave of interest in computer hardware design to better meet the high demands of neural network inference. In particular, analog computing hardware has been heavily motivated specifically for accelerating neural networks, based on either electronic, optical or photonic devices, which may well achieve lower power consumption than conventional digital electronics. However, these proposed analog accelerators suffer from the intrinsic noise generated by their physical components, which makes it challenging to achieve high accuracy on deep neural netw…
The University of Texas at Austin（德克萨斯大学奥斯汀分校）；Massachusetts Institute of Technology（麻省理工学院）；South China University of Technology（华南理工大学）；Tsinghua University（清华大学）；Shanghai Qi Zhi Institute（上海期智研究院）
Crossmodal knowledge distillation (KD) extends traditional knowledge distillation to the area of multimodal learning and demonstrates great success in various applications. To achieve knowledge transfer across modalities, a pretrained network from one modality is adopted as the teacher to provide supervision signals to a student network learning from another modality. In contrast to the empirical success reported in prior works, the working mechanism of crossmodal KD remains a mystery. In this paper, we present a thorough understanding of crossmodal KD. We begin with two case studies and demon…
University of Southern California（南加州大学）；Facebook AI
To reduce a model size but retain performance, we often rely on knowledge distillation (KD) which transfers knowledge from a large "teacher" model to a smaller "student" model. However, KD on multimodal datasets such as vision-language tasks is relatively unexplored, and digesting multimodal information is challenging since different modalities present different types of information. In this paper, we perform a large-scale empirical study to investigate the importance and effects of each modality in knowledge distillation. Furthermore, we introduce a multimodal knowledge distillation framework…
Language model pre-training, such as BERT, has significantly improved the performances of many natural language processing tasks. However, pre-trained language models are usually computationally expensive, so it is difficult to efficiently execute them on resource-restricted devices. To accelerate inference and reduce model size while maintaining accuracy, we first propose a novel Transformer distillation method that is specially designed for knowledge distillation (KD) of the Transformer-based models. By leveraging this new KD method, the plenty of knowledge encoded in a large teacher BERT ca…
While deep and large pre-trained models are the state-of-the-art for various natural language processing tasks, their huge size poses significant challenges for practical uses in resource constrained settings. Recent works in knowledge distillation propose task-agnostic as well as task-specific methods to compress these models, with task-specific ones often yielding higher compression rate. In this work, we develop a new task-agnostic distillation framework XtremeDistilTransformers that leverages the advantage of task-specific methods for learning a small universal model that can be applied to…
The performance of autoregressive models on natural language generation tasks has dramatically improved due to the adoption of deep, self-attentive architectures. However, these gains have come at the cost of hindering inference speed, making state-of-the-art models cumbersome to deploy in real-world, time-sensitive settings. We develop a compression technique for autoregressive models that is driven by an imitation learning perspective on knowledge distillation. The algorithm is designed to address the exposure bias problem. On prototypical language generation tasks such as translation and su…
Significant advances have been made in Natural Language Processing (NLP) modelling since the beginning of 2018. The new approaches allow for accurate results, even when there is little labelled data, because these NLP models can benefit from training on both task-agnostic and task-specific unlabelled data. However, these advantages come with significant size and computational costs. This workshop paper outlines how our proposed convolutional student architecture, having been trained by a distillation process from a large-scale model, can achieve 300x inference speedup and 39x reduction in para…
Nanjing University of Science and Technology（南京理工大学）；Peking University（北京大学）；Microsoft Research Asia（微软亚洲研究院）
While pre-training and fine-tuning, e.g., BERT~\citep{devlin2018bert}, GPT-2~\citep{radford2019language}, have achieved great success in language understanding and generation tasks, the pre-trained models are usually too big for online deployment in terms of both memory cost and inference speed, which hinders them from practical online usage. In this paper, we propose LightPAFF, a Lightweight Pre-training And Fine-tuning Framework that leverages two-stage knowledge distillation to transfer knowledge from a big teacher model to a lightweight student model in both pre-training and fine-tuning st…
University of Liverpool（利物浦大学）；Huawei Ireland Research Center（华为爱尔兰研究中心）
Pruning aims to reduce the number of parameters while maintaining performance close to the original network. This work proposes a novel \emph{self-distillation} based pruning strategy, whereby the representational similarity between the pruned and unpruned versions of the same network is maximized. Unlike previous approaches that treat distillation and pruning separately, we use distillation to inform the pruning criteria, without requiring a separate student network as in knowledge distillation. We show that the proposed {\em cross-correlation objective for self-distilled pruning} implicitly …
Huawei Noah’s Ark Lab（华为诺亚方舟实验室）；University of British Columbia（英属哥伦比亚大学）
The advent of large pre-trained language models has given rise to rapid progress in the field of Natural Language Processing (NLP). While the performance of these models on standard benchmarks has scaled with size, compression techniques such as knowledge distillation have been key in making them practical. We present, MATE-KD, a novel text-based adversarial training algorithm which improves the performance of knowledge distillation. MATE-KD first trains a masked language model based generator to perturb text by maximizing the divergence between teacher and student logits. Then using knowledge…
Learning on a massive amount of speech corpus leads to the recent success of many self-supervised speech models. With knowledge distillation, these models may also benefit from the knowledge encoded by language models that are pre-trained on rich sources of texts. The distillation process, however, is challenging due to the modal disparity between textual and speech embedding spaces. This paper studies metric-based distillation to align the embedding space of text and speech with only a small amount of data without modifying the model structure. Since the semantic and granularity gap between t…
Fine-tuning pre-trained language models like BERT has become an effective way in NLP and yields state-of-the-art results on many downstream tasks. Recent studies on adapting BERT to new tasks mainly focus on modifying the model structure, re-designing the pre-train tasks, and leveraging external data and knowledge. The fine-tuning strategy itself has yet to be fully explored. In this paper, we improve the fine-tuning of BERT with two effective mechanisms: self-ensemble and self-distillation. The experiments on text classification and natural language inference tasks show our proposed methods c…
DeepMind；Google Research（谷歌研究）；EECS Dept., University of California at Berkeley（加州大学伯克利分校电气工程与计算机科学系）；University of California at Berkeley（加州大学伯克利分校）；Google（谷歌）
Knowledge distillation introduced in the deep learning context is a method to transfer knowledge from one architecture to another. In particular, when the architectures are identical, this is called self-distillation. The idea is to feed in predictions of the trained model as new target values for retraining (and iterate this loop possibly a few times). It has been empirically observed that the self-distilled model often achieves higher accuracy on held out data. Why this happens, however, has been a mystery: the self-distillation dynamics does not receive any new information about the task an…
Microsoft Dynamics 365 AI Research（微软Dynamics 365人工智能研究院）
Pre-trained language models such as BERT have proven to be highly effective for natural language processing (NLP) tasks. However, the high demand for computing resources in training such models hinders their application in practice. In order to alleviate this resource hunger in large-scale model training, we propose a Patient Knowledge Distillation approach to compress an original large model (teacher) into an equally-effective lightweight shallow network (student). Different from previous knowledge distillation methods, which only use the output from the last layer of the teacher network for …
The remarkable performance of the pre-trained language model (LM) using self-supervised learning has led to a major paradigm shift in the study of natural language processing. In line with these changes, leveraging the performance of speech recognition systems with massive deep learning-based LMs is a major topic of speech recognition research. Among the various methods of applying LMs to speech recognition systems, in this paper, we focus on a cross-modal knowledge distillation method that transfers knowledge between two types of deep neural networks with different modalities. We propose an a…
USC Information Sciences Institute（南加州大学信息科学研究所）；Google Research（谷歌研究院）
Despite the popularity and efficacy of knowledge distillation, there is limited understanding of why it helps. In order to study the generalization behavior of a distilled student, we propose a new theoretical framework that leverages supervision complexity: a measure of alignment between teacher-provided supervision and the student's neural tangent kernel. The framework highlights a delicate interplay among the teacher's accuracy, the student's margin with respect to the teacher predictions, and the complexity of the teacher predictions. Specifically, it provides a rigorous justification for …
Deep neural networks (DNNs) have been extremely successful in solving many challenging AI tasks in natural language processing, speech recognition, and computer vision nowadays. However, DNNs are typically computation intensive, memory demanding, and power hungry, which significantly limits their usage on platforms with constrained resources. Therefore, a variety of compression techniques (e.g. quantization, pruning, and knowledge distillation) have been proposed to reduce the size and power consumption of DNNs. Blockwise knowledge distillation is one of the compression techniques that can eff…
In recent years the empirical success of transfer learning with neural networks has stimulated an increasing interest in obtaining a theoretical understanding of its core properties. Knowledge distillation where a smaller neural network is trained using the outputs of a larger neural network is a particularly interesting case of transfer learning. In the present work, we introduce a statistical physics framework that allows an analytic characterization of the properties of knowledge distillation (KD) in shallow neural networks. Focusing the analysis on a solvable model that exhibits a non-triv…
The pre-training models such as BERT have achieved great results in various natural language processing problems. However, a large number of parameters need significant amounts of memory and the consumption of inference time, which makes it difficult to deploy them on edge devices. In this work, we propose a knowledge distillation method LRC-BERT based on contrastive learning to fit the output of the intermediate layer from the angular distance aspect, which is not considered by the existing distillation methods. Furthermore, we introduce a gradient perturbation-based training architecture in …
David R. Cheriton School of Computer Science, University of Waterloo（滑铁卢大学大卫·R·切里顿计算机科学学院）；RALI/DIRO, Université de Montréal（蒙特利尔大学RALI/DIRO）
Knowledge Distillation (KD) is extensively used to compress and deploy large pre-trained language models on edge devices for real-world applications. However, one neglected area of research is the impact of noisy (corrupted) labels on KD. We present, to the best of our knowledge, the first study on KD with noisy labels in Natural Language Understanding (NLU). We document the scope of the problem and present two methods to mitigate the impact of label noise. Experiments on the GLUE benchmark show that our methods are effective even under high noise levels. Nevertheless, our results indicate tha…
Carnegie Mellon University（卡内基梅隆大学）；Facebook AI Research（脸书人工智能研究）
Non-autoregressive machine translation (NAT) systems predict a sequence of output tokens in parallel, achieving substantial improvements in generation speed compared to autoregressive models. Existing NAT models usually rely on the technique of knowledge distillation, which creates the training data from a pretrained autoregressive model for better performance. Knowledge distillation is empirically useful, leading to large gains in accuracy for NAT models, but the reason for this success has, as of yet, been unclear. In this paper, we first design systematic experiments to investigate why know…
Knowledge distillation (KD) has been a ubiquitous method for model compression to strengthen the capability of a lightweight model with the transferred knowledge from the teacher. In particular, KD has been employed in quantization-aware training (QAT) of Transformer encoders like BERT to improve the accuracy of the student model with the reduced-precision weight parameters. However, little is understood about which of the various KD approaches best fits the QAT of Transformers. In this work, we provide an in-depth analysis of the mechanism of KD on attention recovery of quantized large Transf…
Microsoft Research（微软研究院）；Microsoft Dynamics 365 AI（微软 Dynamics 365 AI）
This paper explores the use of knowledge distillation to improve a Multi-Task Deep Neural Network (MT-DNN) (Liu et al., 2019) for learning text representations across multiple natural language understanding tasks. Although ensemble learning can improve model performance, serving an ensemble of large DNNs such as MT-DNN can be prohibitively expensive. Here we apply the knowledge distillation method (Hinton et al., 2015) in the multi-task learning setting. For each task, we train an ensemble of different MT-DNNs (teacher) that outperforms any single model, and then train a single MT-DNN (student…
AFD-SLU: Adaptive Feature Distillation for Spoken Language Understanding
School of Advanced Manufacturing and Robotics（先进制造与机器人学院）；Peking University（北京大学）；National Institute of Defense Technology Innovation（国防科技创新国家研究院）；Academy of Military Sciences（军事科学院）；Tianjin Artificial Intelligence Innovation Center（天津人工智能创新中心）
Large Language Models (LLMs) have shown impressive results on a variety of text understanding tasks. Search queries though pose a unique challenge, given their short-length and lack of nuance or context. Complicated feature engineering efforts do not always lead to downstream improvements as their performance benefits may be offset by increased complexity of knowledge distillation. Thus, in this paper we make the following contributions: (1) We demonstrate that Retrieval Augmentation of queries provides LLMs with valuable additional context enabling improved understanding. While Retrieval Augm…
