Pre-trained language models (e.g., BERT (Devlin et al., 2018) and its variants) have achieved remarkable success in varieties of NLP tasks. However, these models usually consist of hundreds of millions of parameters which brings challenges for fine-tuning and online serving in real-life applications due to latency and capacity constraints. In this work, we present a simple and effective approach to compress large Transformer (Vaswani et al., 2017) based pre-trained models, termed as deep self-attention distillation. The small model (student) is trained by deeply mimicking the self-attention module, which plays a vital role in Transformer networks, of the large model (teacher). Specifically, we propose distilling the self-attention module of the last Transformer layer of the teacher, which is effective and flexible for the student. Furthermore, we introduce the scaled dot-product between values in the self-attention module as the new deep self-attention knowledge, in addition to the attention distributions (i.e., the scaled dot-product of queries and keys) that have been used in existing works. Moreover, we show that introducing a teacher assistant (Mirzadeh et al., 2019) also helps the distillation of large pre-trained Transformer models. Experimental results demonstrate that our monolingual model outperforms state-of-the-art baselines in different parameter size of student models. In particular, it retains more than 99% accuracy on SQuAD 2.0 and several GLUE benchmark tasks using 50% of the Transformer parameters and computations of the teacher model. We also obtain competitive results in applying deep self-attention distillation to multilingual pre-trained models.
We generalize deep self-attention distillation in MiniLM (Wang et al., 2020) by only using self-attention relation distillation for task-agnostic compression of pretrained Transformers. In particular, we define multi-head self-attention relations as scaled dot-product between the pairs of query, key, and value vectors within each self-attention module. Then we employ the above relational knowledge to train the student model. Besides its simplicity and unified principle, more favorably, there is no restriction in terms of the number of student's attention heads, while most previous work has to …
Task-agnostic knowledge distillation attempts to address the problem of deploying large pretrained language model in resource-constrained scenarios by compressing a large pretrained model called teacher into a smaller one called student such that the student can be directly finetuned on downstream tasks and retains comparable performance. However, we empirically find that there is a generalization gap between the student and the teacher in existing methods. In this work, we show that we can leverage multi-task learning in task-agnostic distillation to advance the generalization of the resulted…
Georgia Institute of Technology（佐治亚理工学院）；Amazon（亚马逊）
Knowledge distillation has been shown to be a powerful model compression approach to facilitate the deployment of pre-trained language models in practice. This paper focuses on task-agnostic distillation. It produces a compact pre-trained model that can be easily fine-tuned on various tasks with small computational costs and memory footprints. Despite the practical benefits, task-agnostic distillation is challenging. Since the teacher model has a significantly larger capacity and stronger representation power than the student model, it is very difficult for the student to produce predictions t…
The Pennsylvania State University（宾夕法尼亚州立大学）；Microsoft Research（微软研究院）
Knowledge distillation (KD) methods compress large models into smaller students with manually-designed student architectures given pre-specified computational cost. This requires several trials to find a viable student, and further repeating the process for each student or computational budget change. We use Neural Architecture Search (NAS) to automatically distill several compressed students with variable cost from a large model. Current works train a single SuperLM consisting of millions of subnetworks with weight-sharing, resulting in interference between subnetworks of different sizes. Our…
Pre-trained multilingual language models (LMs) have achieved state-of-the-art results in cross-lingual transfer, but they often lead to an inequitable representation of languages due to limited capacity, skewed pre-training data, and sub-optimal vocabularies. This has prompted the creation of an ever-growing pre-trained model universe, where each model is trained on large amounts of language or domain specific data with a carefully curated, linguistically informed vocabulary. However, doing so brings us back full circle and prevents one from leveraging the benefits of multilinguality. To addre…
University of Waterloo（滑铁卢大学）；Salesforce Research（赛富时研究院）
Pretrained language models have led to significant performance gains in many NLP tasks. However, the intensive computing resources to train such models remain an issue. Knowledge distillation alleviates this problem by learning a light-weight student model. So far the distillation approaches are all task-specific. In this paper, we explore knowledge distillation under the multi-task learning setting. The student is jointly distilled across different tasks. It acquires more general representation capacity through multi-tasking distillation and can be further fine-tuned to improve the model in t…
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 …
Deep and large pre-trained language models are the state-of-the-art for various natural language processing tasks. However, the huge size of these models could be a deterrent to use them in practice. Some recent and concurrent works use knowledge distillation to compress these huge models into shallow ones. In this work we study knowledge distillation with a focus on multi-lingual Named Entity Recognition (NER). In particular, we study several distillation strategies and propose a stage-wise optimization scheme leveraging teacher internal representations that is agnostic of teacher architectur…
Pre-trained language models (PLMs) have achieved great success in natural language processing. Most of PLMs follow the default setting of architecture hyper-parameters (e.g., the hidden dimension is a quarter of the intermediate dimension in feed-forward sub-networks) in BERT (Devlin et al., 2019). Few studies have been conducted to explore the design of architecture hyper-parameters in BERT, especially for the more efficient PLMs with tiny sizes, which are essential for practical deployment on resource-constrained devices. In this paper, we adopt the one-shot Neural Architecture Search (NAS) …
Max Planck Institute for Psycholinguistics（马克斯·普朗克心理语言学研究所）；University of Ulm（乌尔姆大学）
Large pretrained language models (PreLMs) are revolutionizing natural language processing across all benchmarks. However, their sheer size is prohibitive for small laboratories or for deployment on mobile devices. Approaches like pruning and distillation reduce the model size but typically retain the same model architecture. In contrast, we explore distilling PreLMs into a different, more efficient architecture, Continual Multiplication of Words (CMOW), which embeds each word as a matrix and uses matrix multiplication to encode sequences. We extend the CMOW architecture and its CMOW/CBOW-Hybri…
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…
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…
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 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…
Self-supervised learning (SSL) is seen as a very promising approach with high performance for several speech downstream tasks. Since the parameters of SSL models are generally so large that training and inference require a lot of memory and computational cost, it is desirable to produce compact SSL models without a significant performance degradation by applying compression methods such as knowledge distillation (KD). Although the KD approach is able to shrink the depth and/or width of SSL model structures, there has been little research on how varying the depth and width impacts the internal …
Recent model pruning methods have demonstrated the ability to remove redundant parameters without sacrificing model performance. Common methods remove redundant parameters according to the parameter sensitivity, a gradient-based measure reflecting the contribution of the parameters. In this paper, however, we argue that redundant parameters can be trained to make beneficial contributions. We first highlight the large sensitivity (contribution) gap among high-sensitivity and low-sensitivity parameters and show that the model generalization performance can be significantly improved after balanci…
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…
Beihang University（北京航空航天大学）；ETH Zurich（苏黎世联邦理工学院）；Bytedance AI Lab（字节跳动人工智能实验室）
Pre-trained vision-language models (VLMs) have achieved impressive results in a range of vision-language tasks. However, popular VLMs usually consist of hundreds of millions of parameters which brings challenges for fine-tuning and deployment in real-world applications due to space, memory, and latency constraints. In this work, we introduce a distilling then pruning framework to compress large vision-language models into smaller, faster, and more accurate ones. We first shrink the size of a pre-trained large VLM and apply knowledge distillation in the vision-language pre-training stage to obt…
