Paul G. Allen School of Computer Science & Engineering, University of Washington（华盛顿大学保罗·G·艾伦计算机科学与工程学院）；Microsoft（微软）；DeepMind；Allen Institute for AI（艾伦人工智能研究所）
Jungo Kasai, Hao Peng, Yizhe Zhang, Dani Yogatama, Gabriel Ilharco, Nikolaos Pappas, Yi Mao, Weizhu Chen, Noah A. Smith
Transformers have outperformed recurrent neural networks (RNNs) in natural language generation. But this comes with a significant computational cost, as the attention mechanism's complexity scales quadratically with sequence length. Efficient transformer variants have received increasing interest in recent works. Among them, a linear-complexity recurrent variant has proven well suited for autoregressive generation. It approximates the softmax attention with randomized or heuristic feature maps, but can be difficult to train and may yield suboptimal accuracy. This work aims to convert a pretrained transformer into its efficient recurrent counterpart, improving efficiency while maintaining accuracy. Specifically, we propose a swap-then-finetune procedure: in an off-the-shelf pretrained transformer, we replace the softmax attention with its linear-complexity recurrent alternative and then finetune. With a learned feature map, our approach provides an improved tradeoff between efficiency and accuracy over the standard transformer and other recurrent variants. We also show that the finetuning process has lower training cost relative to training these recurrent variants from scratch. As many models for natural language tasks are increasingly dependent on large-scale pretrained transformers, this work presents a viable approach to improving inference efficiency without repeating the expensive pretraining process.
We present a simple yet effective approach to build multilingual speech-to-text (ST) translation by efficient transfer learning from pretrained speech encoder and text decoder. Our key finding is that a minimalistic LNA (LayerNorm and Attention) finetuning can achieve zero-shot crosslingual and cross-modality transfer ability by only finetuning less than 10% of the pretrained parameters. This enables effectively leveraging large pretrained models with low training cost. Using wav2vec 2.0 for acoustic modeling, and mBART for multilingual text generation, our approach advanced the new state-of-t…
We present an efficient method of utilizing pretrained language models, where we learn selective binary masks for pretrained weights in lieu of modifying them through finetuning. Extensive evaluations of masking BERT and RoBERTa on a series of NLP tasks show that our masking scheme yields performance comparable to finetuning, yet has a much smaller memory footprint when several tasks need to be inferred simultaneously. Through intrinsic evaluations, we show that representations computed by masked language models encode information necessary for solving downstream tasks. Analyzing the loss land…
We introduce equi-tuning, a novel fine-tuning method that transforms (potentially non-equivariant) pretrained models into group equivariant models while incurring minimum $L_2$ loss between the feature representations of the pretrained and the equivariant models. Large pretrained models can be equi-tuned for different groups to satisfy the needs of various downstream tasks. Equi-tuned models benefit from both group equivariance as an inductive bias and semantic priors from pretrained models. We provide applications of equi-tuning on three different tasks: image classification, compositional ge…
Recent work demonstrates the potential of multilingual pretraining of creating one model that can be used for various tasks in different languages. Previous work in multilingual pretraining has demonstrated that machine translation systems can be created by finetuning on bitext. In this work, we show that multilingual translation models can be created through multilingual finetuning. Instead of finetuning on one direction, a pretrained model is finetuned on many directions at the same time. Compared to multilingual models trained from scratch, starting from pretrained models incorporates the b…
Tsinghua University（清华大学）；Microsoft Research Asia（微软亚洲研究院）
Effectively finetuning pretrained language models (PLMs) is critical for their success in downstream tasks. However, PLMs may have risks in overfitting the pretraining tasks and data, which usually have gap with the target downstream tasks. Such gap may be difficult for existing PLM finetuning methods to overcome and lead to suboptimal performance. In this paper, we propose a very simple yet effective method named NoisyTune to help better finetune PLMs on downstream tasks by adding some noise to the parameters of PLMs before fine-tuning. More specifically, we propose a matrix-wise perturbing m…
Text variational autoencoders (VAEs) are notorious for posterior collapse, a phenomenon where the model's decoder learns to ignore signals from the encoder. Because posterior collapse is known to be exacerbated by expressive decoders, Transformers have seen limited adoption as components of text VAEs. Existing studies that incorporate Transformers into text VAEs (Li et al., 2020; Fang et al., 2021) mitigate posterior collapse using massive pretraining, a technique unavailable to most of the research community without extensive computing resources. We present a simple two-phase training scheme …
Paul G. Allen School of Computer Science & Engineering, University of Washington（华盛顿大学保罗·G·艾伦计算机科学与工程学院）；Allen Institute for Artificial Intelligence（艾伦人工智能研究所）
For natural language processing systems, two kinds of evidence support the use of text representations from neural language models "pretrained" on large unannotated corpora: performance on application-inspired benchmarks (Peters et al., 2018, inter alia), and the emergence of syntactic abstractions in those representations (Tenney et al., 2019, inter alia). On the other hand, the lack of grounded supervision calls into question how well these representations can ever capture meaning (Bender and Koller, 2020). We apply novel probes to recent language models -- specifically focusing on predicate…
This paper propose to combine pretrained language models with the modular dialogue paradigm for open-domain dialogue modeling. Our method, semantic-enhanced finetuning, instantiates conversation understanding, planning, and response generation as a language model finetuning task. At inference, we disentangle semantic and token variations by specifying sampling methods and constraints for each module separately. For training and evaluation, we present X-Weibo, a Chinese multi-turn open-domain dialogue dataset with automatic annotation for emotions, DAs, and topical words. Experiments show that …
Institute of Automation, CAS（中国科学院自动化研究所）；University of Chinese Academy of Sciences（中国科学院大学）；Beijing Fanyu Technology Co., Ltd（北京凡语科技有限公司）
Spoken Language Understanding (SLU) is one essential step in building a dialogue system. Due to the expensive cost of obtaining the labeled data, SLU suffers from the data scarcity problem. Therefore, in this paper, we focus on data augmentation for slot filling task in SLU. To achieve that, we aim at generating more diverse data based on existing data. Specifically, we try to exploit the latent language knowledge from pretrained language models by finetuning them. We propose two strategies for finetuning process: value-based and context-based augmentation. Experimental results on two public S…
Pretrained Transformers achieve state-of-the-art performance in various code-processing tasks but may be too large to be deployed. As software development tools often incorporate modules for various purposes which may potentially use a single instance of the pretrained model, it appears relevant to utilize parameter-efficient fine-tuning for the pretrained models of code. In this work, we test two widely used approaches, adapters and LoRA, which were initially tested on NLP tasks, on four code-processing tasks. We find that though the efficient fine-tuning approaches may achieve comparable or …
National Technical University of Athens（雅典国家技术大学）；Athens University of Economics and Business（雅典经济与商业大学）；Behavioral Signal Technologies（行为信号技术公司）；University of Athens（雅典大学）
In this paper we present a deep-learning model that competed at SemEval-2018 Task 2 "Multilingual Emoji Prediction". We participated in subtask A, in which we are called to predict the most likely associated emoji in English tweets. The proposed architecture relies on a Long Short-Term Memory network, augmented with an attention mechanism, that conditions the weight of each word, on a "context vector" which is taken as the aggregation of a tweet's meaning. Moreover, we initialize the embedding layer of our model, with word2vec word embeddings, pretrained on a dataset of 550 million English twe…
How does language model pretraining help transfer learning? We consider a simple ablation technique for determining the impact of each pretrained layer on transfer task performance. This method, partial reinitialization, involves replacing different layers of a pretrained model with random weights, then finetuning the entire model on the transfer task and observing the change in performance. This technique reveals that in BERT, layers with high probing performance on downstream GLUE tasks are neither necessary nor sufficient for high accuracy on those tasks. Furthermore, the benefit of using p…
Tsinghua University（清华大学）；Sogou Inc.（搜狗公司）；Beijing Academy of Artificial Intelligence（北京智源人工智能研究院）；Institute for AI Industry Research, Tsinghua University（清华大学智能产业研究院）
Multi-source sequence generation (MSG) is an important kind of sequence generation tasks that takes multiple sources, including automatic post-editing, multi-source translation, multi-document summarization, etc. As MSG tasks suffer from the data scarcity problem and recent pretrained models have been proven to be effective for low-resource downstream tasks, transferring pretrained sequence-to-sequence models to MSG tasks is essential. Although directly finetuning pretrained models on MSG tasks and concatenating multiple sources into a single long sequence is regarded as a simple method to tra…
West Greenlandic, known by native speakers as Kalaallisut, is an extremely low-resource polysynthetic language spoken by around 56,000 people in Greenland. Here, we attempt to finetune a pretrained Kalaallisut-to-English neural machine translation (NMT) system using web-crawled pseudoparallel sentences from around 30 multilingual websites. We compile a corpus of over 93,000 Kalaallisut sentences and over 140,000 Danish sentences, then use cross-lingual sentence embeddings and approximate nearest-neighbors search in an attempt to mine near-translations from these corpora. Finally, we translate …
While task-specific finetuning of pretrained networks has led to significant empirical advances in NLP, the large size of networks makes finetuning difficult to deploy in multi-task, memory-constrained settings. We propose diff pruning as a simple approach to enable parameter-efficient transfer learning within the pretrain-finetune framework. This approach views finetuning as learning a task-specific diff vector that is applied on top of the pretrained parameter vector, which remains fixed and is shared across different tasks. The diff vector is adaptively pruned during training with a differe…
University of Chicago（芝加哥大学）；University of Southern California（南加州大学）
Fine-tuning over large pretrained language models (PLMs) has established many state-of-the-art results. Despite its superior performance, such fine-tuning can be unstable, resulting in significant variance in performance and potential risks for practical applications. Previous works have attributed such instability to the catastrophic forgetting problem in the top layers of PLMs, which indicates iteratively that fine-tuning layers in a top-down manner is a promising solution. In this paper, we first point out that this method does not always work out due to the different convergence speeds of …
University of Bonn（波恩大学）；Ruhr University Bochum（波鸿鲁尔大学）；Fraunhofer IAIS（弗劳恩霍夫应用信息技术研究所）
Answering simple questions over knowledge graphs is a well-studied problem in question answering. Previous approaches for this task built on recurrent and convolutional neural network based architectures that use pretrained word embeddings. It was recently shown that finetuning pretrained transformer networks (e.g. BERT) can outperform previous approaches on various natural language processing tasks. In this work, we investigate how well BERT performs on SimpleQuestions and provide an evaluation of both BERT and BiLSTM-based models in datasparse scenarios.
Computer Science Division, University of California, Berkeley（加利福尼亚大学伯克利分校计算机科学系）
Larger language models have higher accuracy on average, but are they better on every single instance (datapoint)? Some work suggests larger models have higher out-of-distribution robustness, while other work suggests they have lower accuracy on rare subgroups. To understand these differences, we investigate these models at the level of individual instances. However, one major challenge is that individual predictions are highly sensitive to noise in the randomness in training. We develop statistically rigorous methods to address this, and after accounting for pretraining and finetuning noise, w…
SpiroLLM: Finetuning Pretrained LLMs to Understand Spirogram Time Series with Clinical Validation in COPD Reporting
Guangzhou Institute of Technology, Xidian University, Xi’an, China（广州科技研究院，西安电子科技大学，中国）；Department of Computer Science, Tianjin University of Technology, Tianjin, China（天津理工大学计算机学院，天津，中国）；Department of Respiratory, The Second Hospital of Tianjin Medical University, China（天津医科大学第二医院呼吸科，中国）；College of Pulmonary and Critical Care Medicine, Chinese PLA General Hospital, Beijing, China（中国人民解放军总医院呼吸与危重症医学科，北京，中国）；HeartVoice Medical Technology, Hefei, China（合肥心声医疗技术有限公司，中国）；School of Life Science and Technology, Xidian University, Xi’an, China（西安电子科技大学生命科学与技术学院，中国）；National Institute of Health Data Science, Peking University, Beijing, China（北京大学国家健康数据科学研究院，北京，中国）；Institute for Artificial Intelligence, Peking University, Beijing, China（北京大学人工智能研究院，北京，中国）
Hugging Face；Google（谷歌）；LightOn；Allen Institute for AI（艾伦人工智能研究所）；LPENS, École Normale Supérieure（巴黎高等师范学院物理实验室）
Large pretrained Transformer language models have been shown to exhibit zero-shot generalization, i.e. they can perform a wide variety of tasks that they were not explicitly trained on. However, the architectures and pretraining objectives used across state-of-the-art models differ significantly, and there has been limited systematic comparison of these factors. In this work, we present a large-scale evaluation of modeling choices and their impact on zero-shot generalization. In particular, we focus on text-to-text models and experiment with three model architectures (causal/non-causal decoder…
Finetuning Pretrained Language Models (PLM) for IR has been de facto the standard practice since their breakthrough effectiveness few years ago. But, is this approach well understood? In this paper, we study the impact of the pretraining collection on the final IR effectiveness. In particular, we challenge the current hypothesis that PLM shall be trained on a large enough generic collection and we show that pretraining from scratch on the collection of interest is surprisingly competitive with the current approach. We benchmark first-stage ranking rankers and cross-encoders for reranking on th…
Existing knowledge-grounded dialogue systems typically use finetuned versions of a pretrained language model (LM) and large-scale knowledge bases. These models typically fail to generalize on topics outside of the knowledge base, and require maintaining separate potentially large checkpoints each time finetuning is needed. In this paper, we aim to address these limitations by leveraging the inherent knowledge stored in the pretrained LM as well as its powerful generation ability. We propose a multi-stage prompting approach to generate knowledgeable responses from a single pretrained LM. We fir…
