We perform knowledge distillation (KD) benchmark from task-specific BERT-base teacher models to various student models: BiLSTM, CNN, BERT-Tiny, BERT-Mini, and BERT-Small. Our experiment involves 12 datasets grouped in two tasks: text classification and sequence labeling in the Indonesian language. We also compare various aspects of distillations including the usage of word embeddings and unlabeled data augmentation. Our experiments show that, despite the rising popularity of Transformer-based models, using BiLSTM and CNN student models provide the best trade-off between performance and computational resource (CPU, RAM, and storage) compared to pruned BERT models. We further propose some quick wins on performing KD to produce small NLP models via efficient KD training mechanisms involving simple choices of loss functions, word embeddings, and unlabeled data preparation.
Recently, BERT has become an essential ingredient of various NLP deep models due to its effectiveness and universal-usability. However, the online deployment of BERT is often blocked by its large-scale parameters and high computational cost. There are plenty of studies showing that the knowledge distillation is efficient in transferring the knowledge from BERT into the model with a smaller size of parameters. Nevertheless, current BERT distillation approaches mainly focus on task-specified distillation, such methodologies lead to the loss of the general semantic knowledge of BERT for universal…
Hanyang University（汉阳大学）；Vision AI Labs（视觉人工智能实验室）；SK Telecom（SK电讯）
A pre-trained language model, BERT, has brought significant performance improvements across a range of natural language processing tasks. Since the model is trained on a large corpus of diverse topics, it shows robust performance for domain shift problems in which data distributions at training (source data) and testing (target data) differ while sharing similarities. Despite its great improvements compared to previous models, it still suffers from performance degradation due to domain shifts. To mitigate such problems, we propose a simple but effective unsupervised domain adaptation method, a…
Huazhong University of Science and Technology（华中科技大学）；Zhejiang University（浙江大学）；Huawei Noah’s Ark Lab（华为诺亚方舟实验室）；Huawei Technologies Co., Ltd.（华为技术有限公司）
Knowledge distillation (KD) which transfers the knowledge from a large teacher model to a small student model, has been widely used to compress the BERT model recently. Besides the supervision in the output in the original KD, recent works show that layer-level supervision is crucial to the performance of the student BERT model. However, previous works designed the layer mapping strategy heuristically (e.g., uniform or last-layer), which can lead to inferior performance. In this paper, we propose to use the genetic algorithm (GA) to search for the optimal layer mapping automatically. To accele…
In this paper, we propose Stochastic Knowledge Distillation (SKD) to obtain compact BERT-style language model dubbed SKDBERT. In each iteration, SKD samples a teacher model from a pre-defined teacher ensemble, which consists of multiple teacher models with multi-level capacities, to transfer knowledge into student model in an one-to-one manner. Sampling distribution plays an important role in SKD. We heuristically present three types of sampling distributions to assign appropriate probabilities for multi-level teacher models. SKD has two advantages: 1) it can preserve the diversities of multi-…
Tsinghua University（清华大学）；Shenzhen International Graduate School, Tsinghua University（清华大学深圳国际研究生院）
Task-agnostic knowledge distillation, a teacher-student framework, has been proved effective for BERT compression. Although achieving promising results on NLP tasks, it requires enormous computational resources. In this paper, we propose Extract Then Distill (ETD), a generic and flexible strategy to reuse the teacher's parameters for efficient and effective task-agnostic distillation, which can be applied to students of any size. Specifically, we introduce two variants of ETD, ETD-Rand and ETD-Impt, which extract the teacher's parameters in a random manner and by following an importance metric…
Tsinghua University（清华大学）；Alibaba Group（阿里巴巴集团）；Sun-Yat Sen University（中山大学）
Despite pre-trained language models such as BERT have achieved appealing performance in a wide range of natural language processing tasks, they are computationally expensive to be deployed in real-time applications. A typical method is to adopt knowledge distillation to compress these large pre-trained models (teacher models) to small student models. However, for a target domain with scarce training data, the teacher can hardly pass useful knowledge to the student, which yields performance degradation for the student models. To tackle this problem, we propose a method to learn to augment for d…
Institute of Information Engineering, Chinese Academy of Sciences（中国科学院信息工程研究所）；University of Chinese Academy of Sciences（中国科学院大学）；Tencent Inc（腾讯公司）
Recently, knowledge distillation (KD) has shown great success in BERT compression. Instead of only learning from the teacher's soft label as in conventional KD, researchers find that the rich information contained in the hidden layers of BERT is conducive to the student's performance. To better exploit the hidden knowledge, a common practice is to force the student to deeply mimic the teacher's hidden states of all the tokens in a layer-wise manner. In this paper, however, we observe that although distilling the teacher's hidden state knowledge (HSK) is helpful, the performance gain (marginal …
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 …
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 …
Recently, transformer-based language models such as BERT have shown tremendous performance improvement for a range of natural language processing tasks. However, these language models usually are computation expensive and memory intensive during inference. As a result, it is difficult to deploy them on resource-restricted devices. To improve the inference performance, as well as reduce the model size while maintaining the model accuracy, we propose a novel quantization method named KDLSQ-BERT that combines knowledge distillation (KD) with learned step size quantization (LSQ) for language model…
Transformer-based pre-training models like BERT have achieved remarkable performance in many natural language processing tasks.However, these models are both computation and memory expensive, hindering their deployment to resource-constrained devices. In this work, we propose TernaryBERT, which ternarizes the weights in a fine-tuned BERT model. Specifically, we use both approximation-based and loss-aware ternarization methods and empirically investigate the ternarization granularity of different parts of BERT. Moreover, to reduce the accuracy degradation caused by the lower capacity of low bit…
University of California, Berkeley（加州大学伯克利分校）；DeepMind（深度思维）；University of Oxford（牛津大学）
Textual representation learners trained on large amounts of data have achieved notable success on downstream tasks; intriguingly, they have also performed well on challenging tests of syntactic competence. Given this success, it remains an open question whether scalable learners like BERT can become fully proficient in the syntax of natural language by virtue of data scale alone, or whether they still benefit from more explicit syntactic biases. To answer this question, we introduce a knowledge distillation strategy for injecting syntactic biases into BERT pretraining, by distilling the syntac…
As Transfer Learning from large-scale pre-trained models becomes more prevalent in Natural Language Processing (NLP), operating these large models in on-the-edge and/or under constrained computational training or inference budgets remains challenging. In this work, we propose a method to pre-train a smaller general-purpose language representation model, called DistilBERT, which can then be fine-tuned with good performances on a wide range of tasks like its larger counterparts. While most prior work investigated the use of distillation for building task-specific models, we leverage knowledge di…
Tsinghua University（清华大学）；Microsoft Research Asia（微软亚洲研究院）；Peking University（北京大学）
BERT is a cutting-edge language representation model pre-trained by a large corpus, which achieves superior performances on various natural language understanding tasks. However, a major blocking issue of applying BERT to online services is that it is memory-intensive and leads to unsatisfactory latency of user requests, raising the necessity of model compression. Existing solutions leverage the knowledge distillation framework to learn a smaller model that imitates the behaviors of BERT. However, the training procedure of knowledge distillation is expensive itself as it requires sufficient tr…
University of California, San Diego（加州大学圣地亚哥分校）；Beihang University（北京航空航天大学）；Microsoft Research Asia（微软亚洲研究院）
In this paper, we propose a novel model compression approach to effectively compress BERT by progressive module replacing. Our approach first divides the original BERT into several modules and builds their compact substitutes. Then, we randomly replace the original modules with their substitutes to train the compact modules to mimic the behavior of the original modules. We progressively increase the probability of replacement through the training. In this way, our approach brings a deeper level of interaction between the original and compact models. Compared to the previous knowledge distillat…
Stanford University（斯坦福大学）；University of California, San Diego（加州大学圣地亚哥分校）
We present Knowledge Distillation with Meta Learning (MetaDistil), a simple yet effective alternative to traditional knowledge distillation (KD) methods where the teacher model is fixed during training. We show the teacher network can learn to better transfer knowledge to the student network (i.e., learning to teach) with the feedback from the performance of the distilled student network in a meta learning framework. Moreover, we introduce a pilot update mechanism to improve the alignment between the inner-learner and meta-learner in meta learning algorithms that focus on an improved inner-lea…
Shenzhen Campus of Sun Yat-sen University（中山大学深圳校区）；University of Oxford（牛津大学）；Huawei Noah’s Ark Lab（华为诺亚方舟实验室）；DarkMatter AI Research（暗物质人工智能研究院）
Pre-trained language models have shown remarkable results on various NLP tasks. Nevertheless, due to their bulky size and slow inference speed, it is hard to deploy them on edge devices. In this paper, we have a critical insight that improving the feed-forward network (FFN) in BERT has a higher gain than improving the multi-head attention (MHA) since the computational cost of FFN is 2$\sim$3 times larger than MHA. Hence, to compact BERT, we are devoted to designing efficient FFN as opposed to previous works that pay attention to MHA. Since FFN comprises a multilayer perceptron (MLP) that is es…
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…
Recent advances in pre-training huge models on large amounts of text through self supervision have obtained state-of-the-art results in various natural language processing tasks. However, these huge and expensive models are difficult to use in practise for downstream tasks. Some recent efforts use knowledge distillation to compress these models. However, we see a gap between the performance of the smaller student models as compared to that of the large teacher. In this work, we leverage large amounts of in-domain unlabeled transfer data in addition to a limited amount of labeled training insta…
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…
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…
Knowledge distillation is typically conducted by training a small model (the student) to mimic a large and cumbersome model (the teacher). The idea is to compress the knowledge from the teacher by using its output probabilities as soft-labels to optimize the student. However, when the teacher is considerably large, there is no guarantee that the internal knowledge of the teacher will be transferred into the student; even if the student closely matches the soft-labels, its internal representations may be considerably different. This internal mismatch can undermine the generalization capabilitie…
In this work we introduce a new transformer architecture called SparseDistilBERT (SDBERT), which is a combination of sparse attention and knowledge distillantion (KD). We implemented sparse attention mechanism to reduce quadratic dependency on input length to linear. In addition to reducing computational complexity of the model, we used knowledge distillation (KD). We were able to reduce the size of BERT model by 60% while retaining 97% performance and it only took 40% of time to train.
Knowledge distillation is considered as a training and compression strategy in which two neural networks, namely a teacher and a student, are coupled together during training. The teacher network is supposed to be a trustworthy predictor and the student tries to mimic its predictions. Usually, a student with a lighter architecture is selected so we can achieve compression and yet deliver high-quality results. In such a setting, distillation only happens for final predictions whereas the student could also benefit from teacher's supervision for internal components. Motivated by this, we studied…
It can be challenging to train multi-task neural networks that outperform or even match their single-task counterparts. To help address this, we propose using knowledge distillation where single-task models teach a multi-task model. We enhance this training with teacher annealing, a novel method that gradually transitions the model from distillation to supervised learning, helping the multi-task model surpass its single-task teachers. We evaluate our approach by multi-task fine-tuning BERT on the GLUE benchmark. Our method consistently improves over standard single-task and multi-task training…
In this demonstration, we present an efficient BERT-based multi-task (MT) framework that is particularly suitable for iterative and incremental development of the tasks. The proposed framework is based on the idea of partial fine-tuning, i.e. only fine-tune some top layers of BERT while keep the other layers frozen. For each task, we train independently a single-task (ST) model using partial fine-tuning. Then we compress the task-specific layers in each ST model using knowledge distillation. Those compressed ST models are finally merged into one MT model so that the frozen layers of the former…
Apple（苹果公司）；Cornell Tech（康奈尔大学科技校区）；Massachusetts Institute of Technology（麻省理工学院）
Quantization, knowledge distillation, and magnitude pruning are among the most popular methods for neural network compression in NLP. Independently, these methods reduce model size and can accelerate inference, but their relative benefit and combinatorial interactions have not been rigorously studied. For each of the eight possible subsets of these techniques, we compare accuracy vs. model size tradeoffs across six BERT architecture sizes and eight GLUE tasks. We find that quantization and distillation consistently provide greater benefit than pruning. Surprisingly, except for the pair of prun…
Huazhong University of Science and Technology（华中科技大学）；Huawei Technologies Co., Ltd.（华为技术有限公司）
The multilingual pre-trained language models (e.g, mBERT, XLM and XLM-R) have shown impressive performance on cross-lingual natural language understanding tasks. However, these models are computationally intensive and difficult to be deployed on resource-restricted devices. In this paper, we propose a simple yet effective distillation method (LightMBERT) for transferring the cross-lingual generalization ability of the multilingual BERT to a small student model. The experiment results empirically demonstrate the efficiency and effectiveness of LightMBERT, which is significantly better than the …
Microsoft Dynamics 365 AI Research（微软Dynamics 365人工智能研究院）
Existing language model compression methods mostly use a simple L2 loss to distill knowledge in the intermediate representations of a large BERT model to a smaller one. Although widely used, this objective by design assumes that all the dimensions of hidden representations are independent, failing to capture important structural knowledge in the intermediate layers of the teacher network. To achieve better distillation efficacy, we propose Contrastive Distillation on Intermediate Representations (CoDIR), a principled knowledge distillation framework where the student is trained to distill know…
Deep Learning for Medical Text Processing: BERT Model Fine-Tuning and Comparative Study
Tulane University（杜兰大学）；Northeastern University（东北大学）；San Francisco State University（旧金山州立大学）；Brandeis University（布兰迪斯大学）；University of California, Los Angeles（加州大学洛杉矶分校）；Johns Hopkins University（约翰斯·霍普金斯大学）
The huge size of the widely used BERT family models has led to recent efforts about model distillation. The main goal of distillation is to create a task-agnostic pre-trained model that can be fine-tuned on downstream tasks without fine-tuning its full-sized version. Despite the progress of distillation, to what degree and for what reason a task-agnostic model can be created from distillation has not been well studied. Also, the mechanisms behind transfer learning of those BERT models are not well investigated either. Therefore, this work focuses on analyzing the acceptable deduction when dist…
As a pre-trained Transformer model, BERT (Bidirectional Encoder Representations from Transformers) has achieved ground-breaking performance on multiple NLP tasks. On the other hand, Boosting is a popular ensemble learning technique which combines many base classifiers and has been demonstrated to yield better generalization performance in many machine learning tasks. Some works have indicated that ensemble of BERT can further improve the application performance. However, current ensemble approaches focus on bagging or stacking and there has not been much effort on exploring the boosting. In th…
Selecting radiology examination protocol is a repetitive, and time-consuming process. In this paper, we present a deep learning approach to automatically assign protocols to computer tomography examinations, by pre-training a domain-specific BERT model ($BERT_{rad}$). To handle the high data imbalance across exam protocols, we used a knowledge distillation approach that up-sampled the minority classes through data augmentation. We compared classification performance of the described approach with the statistical n-gram models using Support Vector Machine (SVM), Gradient Boosting Machine (GBM),…
We present an approach to ranking with dense representations that applies knowledge distillation to improve the recently proposed late-interaction ColBERT model. Specifically, we distill the knowledge from ColBERT's expressive MaxSim operator for computing relevance scores into a simple dot product, thus enabling single-step ANN search. Our key insight is that during distillation, tight coupling between the teacher model and the student model enables more flexible distillation strategies and yields better learned representations. We empirically show that our approach improves query latency and…
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…
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…
David R. Cheriton School of Computer Science, University of Waterloo（滑铁卢大学大卫·切里顿计算机科学学院）；Huawei Noah’s Ark Lab（华为诺亚方舟实验室）；Department of Statistics and Actuarial Science, University of Waterloo（滑铁卢大学统计与精算科学系）
Significant memory and computational requirements of large deep neural networks restrict their application on edge devices. Knowledge distillation (KD) is a prominent model compression technique for deep neural networks in which the knowledge of a trained large teacher model is transferred to a smaller student model. The success of knowledge distillation is mainly attributed to its training objective function, which exploits the soft-target information (also known as "dark knowledge") besides the given regular hard labels in a training set. However, it is shown in the literature that the large…
Tsinghua University（清华大学）；Microsoft Research Asia（微软亚洲研究院）
Pre-trained language models (PLMs) achieve great success in NLP. However, their huge model sizes hinder their applications in many practical systems. Knowledge distillation is a popular technique to compress PLMs, which learns a small student model from a large teacher PLM. However, the knowledge learned from a single teacher may be limited and even biased, resulting in low-quality student model. In this paper, we propose a multi-teacher knowledge distillation framework named MT-BERT for pre-trained language model compression, which can train high-quality student model from multiple teacher PL…
University of Pittsburgh（匹兹堡大学）；Google Research（谷歌研究院）；Kuaishou Technology（快手科技）；Google Brain（谷歌大脑）
Pretrained language models like BERT have achieved good results on NLP tasks, but are impractical on resource-limited devices due to memory footprint. A large fraction of this footprint comes from the input embeddings with large input vocabulary and embedding dimensions. Existing knowledge distillation methods used for model compression cannot be directly applied to train student models with reduced vocabulary sizes. To this end, we propose a distillation method to align the teacher and student embeddings via mixed-vocabulary training. Our method compresses BERT-LARGE to a task-agnostic model …
Pretrained language models (PLMs) such as BERT adopt a training paradigm which first pretrain the model in general data and then finetune the model on task-specific data, and have recently achieved great success. However, PLMs are notorious for their enormous parameters and hard to be deployed on real-life applications. Knowledge distillation has been prevailing to address this problem by transferring knowledge from a large teacher to a much smaller student over a set of data. We argue that the selection of thee three key components, namely teacher, training data, and learning objective, is cr…
Due to the ambiguity of homophones, Chinese Spell Checking (CSC) has widespread applications. Existing systems typically utilize BERT for text encoding. However, CSC requires the model to account for both phonetic and graphemic information. To adapt BERT to the CSC task, we propose a token-level self-distillation contrastive learning method. We employ BERT to encode both the corrupted and corresponding correct sentence. Then, we use contrastive learning loss to regularize corrupted tokens' hidden states to be closer to counterparts in the correct sentence. On three CSC datasets, we confirmed o…
National Center for High-performance Computing（国家高速网络与计算中心）；National Applied Research Laboratories（国家实验研究院）
Self-supervised speech representation learning methods like wav2vec 2.0 and Hidden-unit BERT (HuBERT) leverage unlabeled speech data for pre-training and offer good representations for numerous speech processing tasks. Despite the success of these methods, they require large memory and high pre-training costs, making them inaccessible for researchers in academia and small companies. Therefore, this paper introduces DistilHuBERT, a novel multi-task learning framework to distill hidden representations from a HuBERT model directly. This method reduces HuBERT's size by 75% and 73% faster while ret…
