Canada

Knowledge distillation has attracted a great deal of interest recently to compress large language models. However, existing knowledge distillation methods suffer from two limitations. First, the student model simply imitates the teacher&#39;s behavior while ignoring the reasoning behind it. Second, these methods usually focus on the transfer of sophisticated model-specific knowledge but overlook data-specific knowledge. In this paper, we present a novel attribution-driven knowledge distillation approach, which explores the token-level rationale behind the teacher model based on Integrated Gradients (IG) and transfers attribution knowledge to the student model. To enhance the knowledge transfer of model reasoning and generalization, we further explore multi-view attribution distillation on all potential decisions of the teacher. Comprehensive experiments are conducted with BERT on the GLUE benchmark. The experimental results demonstrate the superior performance of our approach to several state-of-the-art methods.

&lt;iframe src=&quot;https://app.sli.do/event/8PaBoicsuHK7beXFunvrtr/embed/polls/7fa1a85a-dbb9-4bf0-b362-4f511783ba03&quot; width=&quot;300&quot; height=&quot;400&quot;&gt;&lt;/iframe&gt;

ACL 2023

AD-KD: Attribution-Driven Knowledge Distillation for Language Model Compression

pre-trained language models

knowledge distillation

model compression

## Welcome to EMNLP 2022!
I am delighted to welcome you to EMNLP 2022!  I believe this conference has been complicated beyond any precedent.  Over the past year, it’s been thrilling to see the organization team approach each new puzzle with creativity and enthusiasm.  We hope that those participating in Abu Dhabi as well as those joining remotely will leave the conference feeling newly inspired by the program and newly connected to our ever-growing community.  Following EMNLP 2021 and major NLP conferences since, EMNLP 2022 is “hybrid,” serving both virtual and in-person participants.

Our key innovations for EMNLP 2022 include:

* EMNLP 2022 is “hybrid” in a second sense, as well:  we allowed both direct and rolling review paper submissions, building on the pilot experiment of EMNLP 2021, which considered a small number of ARR submissions. 
* Familiar from NAACL but new to EMNLP, we’ve added an industry track.
* During the conference, “portals” will link virtual poster sessions to in-person conference participants during poster sessions each day.
* The first ACL-family conference in the United Arab Emirates.

<br>*Message from Noah A. Smith, University of Washington and Allen Institute for AI, Seattle, Washington, USA*<br>
***EMNLP 2022 General Chair***
<br><br><br>
[![](https://assets.underline.io/uploads/markdown_image/1/image/9eec7d4a287ee18c278b08229290aa83.png)](https://drive.google.com/file/d/1OlPv6QBeo62VVTughj2jkiLeyHd1WnUt/view)
<br><br>
[![](https://assets.underline.io/uploads/markdown_image/1/image/a3db7a768409f05192210d98601edb25.png)](https://emnlp2022.rocket.chat/)



To access this site you need to register. Please register [here](https://2022.emnlp.org/registration/).

Register here

EMNLP 2022

Welcome!
EMNLP 2022 will take place in Abu Dhabi from December 7th to December 11th, 2022. And it will be held in hybrid mode, both online and offline.

Knowledge distillation has attracted a great deal of interest recently to compress large language models. However, existing knowledge distillation methods suffer from two limitations. First, the student model simply imitates the teacher's behavior while ignoring the reasoning behind it. Second, these methods usually focus on the transfer of sophisticated model-specific knowledge but overlook data-specific knowledge. In this paper, we present a novel attribution-driven knowledge distillation approach, which explores the token-level rationale behind the teacher model based on Integrated Gradients (IG) and transfers attribution knowledge to the student model. To enhance the knowledge transfer of model reasoning and generalization, we further explore multi-view attribution distillation on all potential decisions of the teacher. Comprehensive experiments are conducted with BERT on the GLUE benchmark. The experimental results demonstrate the superior performance of our approach to several state-of-the-art methods.

<iframe src="https://app.sli.do/event/8PaBoicsuHK7beXFunvrtr/embed/polls/7fa1a85a-dbb9-4bf0-b362-4f511783ba03" width="300" height="400"></iframe>

poster

### Welcome to ACL 2023, the 61st Annual Meeting of the Association for Computational Linguistics! 
<br>The conference will be held in Toronto, Canada, July 9-14, 2023.<br><br>
Following the succession of the recent conferences in our field, ACL 2023 will adopt a hybrid format.
While the impact of Covid has considerably diminished in terms of traveling, obtaining visas to Canada
entails a very long process. Moreover, the global economic conditions pose challenges for many individuals to travel to conferences. Recognizing these circumstances, we know many participants may not be
able to attend the conference in person. Therefore, we are committed to providing a great virtual platform
so everyone has the opportunity to interact with other participants and enjoy the conference. Based on the
current registered participants, approxiately 30% have chosen to attend the conference virtually. Whether
you join us in person or virtually, we sincerely hope everyone has a remarkable conference experience.<br>
This General Chair’s message is where I express my gratitude to the many individuals who have made
enormous contributions to the conference over the past year.

Read [**ACL 2023 General Chair's message**](https://docs.google.com/document/d/1WobYM7norbG4dI48s75HfJoD89qgX5a_F-6U8AteLSA/edit?usp=sharing/) in full.

##### **[Conference Handbook](https://2023.aclweb.org/downloads/acl2023-handbook.pdf)**
<br><br><br><br>

The Association for Computational Linguistics (ACL) is the premier international scientific and professional society for people working on computational problems involving human language, a field often referred to as either computational linguistics or natural language processing.

Given a language model (LM), maximum probability is a poor decoding objective for open-ended generation, because it produces short and repetitive text. On the other hand, sampling can often produce incoherent text that drifts from the original topics. We propose contrastive decoding (CD), a reliable decoding approach that optimizes a contrastive objective subject to a plausibility constraint. The contrastive objective returns the difference between the likelihood under a large LM (called the expert, e.g. OPT-13B) and a small LM (called the amateur, e.g. OPT-125M), and the constraint ensures that the outputs are plausible. CD is inspired by the fact that the failures of larger LMs (e.g., repetition, inco- herence) are even more prevalent in smaller LMs, and that this difference signals which texts should be preferred. CD requires zero additional training, and produces higher quality text than decoding from the larger LM alone. It also works across model scales (OPT-13B and GPT2-1.5B) and significantly outperforms four strong decoding algorithms (e.g., nucleus, top-k) in automatic and human evaluations across wikipedia, news and story domains.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/d9219db4-af70-4cf4-a60f-453c1f306be5" width="300" height="400"></iframe>

Contrastive Decoding: Open-ended Text Generation as Optimization

Large-scale vision language (VL) models use Transformers to perform cross-modal interactions between the input text and image. These cross-modal interactions are computationally expensive and memory-intensive due to the quadratic complexity of processing the input image and text. We present PuMer: a token reduction framework that uses text-informed Pruning and modality-aware Merging strategies to progressively reduce the tokens of input image and text, improving model inference speed and reducing memory footprint. PuMer learns to keep salient image tokens related to the input text and merges similar textual and visual tokens by adding lightweight token reducer modules at several cross-modal layers in the VL model. Training PuMer is mostly the same as finetuning the original VL model but faster. Our evaluation for two vision language models on four downstream VL tasks shows PuMer increases inference throughput by up to 2x and reduces memory footprint by over 50% while incurring less than a 1% accuracy drop.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/c050e87a-7add-47c9-8d76-3a75f964c5e0" width="300" height="400"></iframe>

PuMer: Pruning and Merging Tokens for Efficient Vision Language Models

Current state-of-the-art models for natural language understanding require a preprocessing step to convert raw text into discrete tokens. This process known as tokenization relies on a pre-built vocabulary of words or sub-word morphemes. This fixed vocabulary limits the model's robustness to spelling errors and its capacity to adapt to new domains. In this work, we introduce a novel open-vocabulary language model that adopts a hierarchical two-level approach: one at the word level and another at the sequence level. Concretely, we design an intra-word module that uses a shallow Transformer architecture to learn word representations from their characters, and a deep inter-word Transformer module that contextualizes each word representation by attending to the entire word sequence. Our model thus directly operates on character sequences with explicit awareness of word boundaries, but without biased sub-word or word-level vocabulary. Experiments on various downstream tasks show that our method outperforms strong baselines. We also demonstrate that our hierarchical model is robust to textual corruption and domain shift.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/b5f49b1a-bd76-42ba-a3d4-9846e232c58a" width="300" height="400"></iframe>

From Characters to Words: Hierarchical Pre-trained Language Model for Open-vocabulary Language Understanding

Online resources such as WikiHow compile a wide range of scripts for performing everyday tasks, which can assist models in learning to reason about procedures. However, the scripts are always presented in a linear manner, which does not reflect the flexibility displayed by people executing tasks in real life. For example, in the CrossTask Dataset, 64.5% of consecutive step pairs are also observed in the reverse order, suggesting their ordering is not fixed. In addition, each step has an average of 2.56 frequent next steps, demonstrating "branching". In this paper, we propose the new challenging task of non-sequential graph script induction, aiming to capture optional and interchangeable steps in procedural planning. To automate the induction of such graph scripts for given tasks, we propose to take advantage of loosely aligned videos of people performing the tasks. In particular, we design a multimodal framework to ground procedural videos to WikiHow textual steps and thus transform each video into an observed step path on the latent ground truth graph script. This key transformation enables us to train a script knowledge model capable of both generating explicit graph scripts for learnt tasks and predicting future steps given a partial step sequence. Our best model outperforms the strongest pure text/vision baselines by 17.52% absolute gains on F1@3 for next step prediction and 13.8% absolute gains on Acc@1 for partial sequence completion. Human evaluation shows our model outperforming the WikiHow linear baseline by 48.76% absolute gains in capturing sequential and non-sequential step relationships.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/4ff53d03-d703-4cec-8889-82e90a948444" width="300" height="400"></iframe>

Non-Sequential Graph Script Induction via Multimedia Grounding

Direct speech-to-speech translation (S2ST) aims to convert speech from one language into another, and has demonstrated significant progress to date. Despite the recent success, current S2ST models still suffer from distinct degradation in noisy environments and fail to translate visual speech (i.e., the movement of lips and teeth). In this work, we present AV-TranSpeech, the first audio-visual speech-to-speech (AV-S2ST) translation model without relying on intermediate text. AV-TranSpeech complements the audio stream with visual information to promote system robustness and opens up a host of practical applications: dictation or dubbing archival films. To mitigate the data scarcity with limited parallel AV-S2ST data, we 1) explore self-supervised pre-training with unlabeled audio-visual data to learn contextual representation, and 2) introduce cross-modal distillation with S2ST models trained on the audio-only corpus to further reduce the requirements of visual data. Experimental results on two language pairs demonstrate that AV-TranSpeech outperforms audio-only models under all settings regardless of the type of noise. With low-resource audio-visual data (10h, 30h), cross-modal distillation yields an improvement of 7.6 BLEU on average compared with baselines.\footnote{Audio samples are available at https://AV-TranSpeech.github.io/.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/0206acd0-0522-4516-bce1-c4b675b3e2f9" width="300" height="400"></iframe>

AV-TranSpeech: Audio-Visual Robust Speech-to-Speech Translation

Similes occur in the creative context of describing a concept (i.e., tenor) by making a literally false yet figuratively meaningful comparison to another (i.e., vehicle). Previous efforts form simile generation as a context-free generation task, focusing on simile-style transfer or writing a simile from a given prefix. However, generated texts under such settings might be undesirable, such as hardly meeting the simile definition (e.g., missing vehicle) or difficult to address certain preferences of content as humans wish (e.g., describe the color of apples through the simile). We believe that a simile could be more qualified and user-oriented if incorporated with pre-specified constraints. To this end, we introduce controllable simile generation (CSG), a new task that requires the model to generate a simile with multiple simile elements, e.g., context and vehicle. To facilitate this task, we present GraCe, including 61.3k simile-element annotated Chinese similes. Based on it, we propose a CSG model Similor to benchmark this task, including a vehicle retrieval module Scorer to obtain the explicable comparison for a given tenor in the vehicle-unknown situation. Both statistical and experimental analyses show that GraCe is of high quality beyond all other Chinese simile datasets, in terms of the number (8 vs. 3) of annotation elements, Is-Simile accuracy (98.9% vs. 78.7%), and increasing model-performance gains for both uncontrollable and controllable simile generation. Meanwhile, Similor can serve as a strong baseline for CSG, especially with Scorer, which beats model-based retrieval methods without any re-training.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/aa72ea9e-c0df-41b3-8849-af8371122079" width="300" height="400"></iframe>

Fantastic Expressions and Where to Find Them: Chinese Simile Generation with Multiple Constraints

Short-form video hashtag recommendation (SVHR) aims to recommend hashtags to content creators from videos and corresponding descriptions. Most prior studies regard SVHR as a classification or ranking problem and select hashtags from a set of limited candidates. However, in reality, users can create new hashtags, and trending hashtags change rapidly over time on social media. Both of these properties cannot be easily modeled with classification approaches. To bridge this gap, we formulate SVHR as a generation task that better represents how hashtags are created naturally. Additionally, we propose the CLIP-Guided Generative Model (CGGM) where we augment the input features by retrieving relevant hashtags from a large-scale hashtag pool as extra guidance signals. Experimental results on two datasets (SFVD and YFCC100M) show that our generative models outperform classification baselines, and the guidance signals further boost the performance by 8.11 and 2.17 ROUGE-1 on average, respectively. We also perform extensive analyses including human evaluation, demonstrating that our generative model can create meaningful and relevant novel hashtags while achieving state-of-the-art performance on known hashtags.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/985f9245-38e6-4251-b789-b81817e0c9e3" width="300" height="400"></iframe>

Generating Hashtags for Short-form Videos with Guided Signals

Information Extraction (IE) spans several tasks with different output structures, such as named entity recognition, relation extraction and event extraction. Previously, those tasks were solved with different models because of diverse task output structures. Through re-examining IE tasks, we find that all of them can be interpreted as extracting spans and span relations. They can further be decomposed into token-pair classification tasks by using the start and end token of a span to pinpoint the span, and using the start-to-start and end-to-end token pairs of two spans to determine the relation. Based on the reformulation, we propose a Unified Token-pair Classification architecture for Information Extraction (UTC-IE), where we introduce Plusformer on top of the token-pair feature matrix. Specifically, it models axis-aware interaction with plus-shaped self-attention and local interaction with Convolutional Neural Network over token pairs. Experiments show that our approach outperforms task-specific and unified models on all tasks in 10 datasets, and achieves better or comparable results on 2 joint IE datasets. Moreover, UTC-IE speeds up over state-of-the-art models on IE tasks significantly in most datasets, which verifies the effectiveness of our architecture.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/3691afe4-35e3-43d6-bc30-106150699043" width="300" height="400"></iframe>

UTC-IE: A Unified Token-pair Classification Architecture for Information Extraction

Chinese Spelling Correction (CSC) aims to detect and correct erroneous characters in Chinese texts. Although efforts have been made to introduce phonetic information (Hanyu Pinyin) in this task, they typically merge phonetic representations with character representations, which tends to weaken the representation effect of normal texts. In this work, we propose to disentangle the two types of features to allow for direct interaction between textual and phonetic information. To learn useful phonetic representations, we introduce a pinyin-to-character objective to ask the model to predict the correct characters based solely on phonetic information, where a separation mask is imposed to disable attention from phonetic input to text. To avoid overfitting the phonetics, we further design a self-distillation module to ensure that semantic information plays a major role in the prediction. Extensive experiments on three CSC benchmarks demonstrate the superiority of our method in using phonetic information.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/735b6dd6-5eba-420e-b70e-f51bc2d33ed3" width="300" height="400"></iframe>

Disentangled Phonetic Representation for Chinese Spelling Correction

Large language models (LLMs) have demonstrated powerful capabilities in both text understanding and generation. Companies have begun to offer Embedding as a Service (EaaS) based on these LLMs, which can benefit various natural language processing (NLP) tasks for customers. However, previous studies have shown that EaaS is vulnerable to model extraction attacks, which can cause significant losses for the owners of LLMs, as training these models is extremely expensive. To protect the copyright of LLMs for EaaS, we propose an Embedding Watermark method called \method that implants backdoors on embeddings. Our method selects a group of moderate-frequency words from a general text corpus to form a trigger set, then selects a target embedding as the watermark, and inserts it into the embeddings of texts containing trigger words as the backdoor. The weight of insertion is proportional to the number of trigger words included in the text. This allows the watermark backdoor to be effectively transferred to EaaS-stealer's model for copyright verification while minimizing the adverse impact on the original embeddings' utility. Our extensive experiments on various datasets show that our method can effectively protect the copyright of EaaS models without compromising service quality.Our code is available at \url{https://github.com/yjw1029/EmbMarker}.

<iframe src="https://app.sli.do/event/or9aypsi961h595NQtD7kk/embed/polls/15f63d67-c40b-4fe4-b50c-a81e44472a69" width="300" height="400"></iframe>

Downloads

Next from ACL 2023

Contrastive Decoding: Open-ended Text Generation as Optimization

Similar lecture

Improved Knowledge Distillation for Pre-trained Language Models via Knowledge Selection