bibtex_url
stringlengths 41
53
| proceedings
stringlengths 38
50
| bibtext
stringlengths 528
3.02k
| abstract
stringlengths 17
2.35k
| authors
sequencelengths 1
44
| title
stringlengths 18
190
| id
stringlengths 7
19
| arxiv_id
stringlengths 0
10
| GitHub
sequencelengths 1
1
| paper_page
stringclasses 528
values | n_linked_authors
int64 -1
15
| upvotes
int64 -1
77
| num_comments
int64 -1
10
| n_authors
int64 -1
52
| Models
sequencelengths 0
100
| Datasets
sequencelengths 0
15
| Spaces
sequencelengths 0
46
| paper_page_exists_pre_conf
int64 0
1
| type
stringclasses 2
values |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
https://aclanthology.org/2023.emnlp-main.1.bib | https://aclanthology.org/2023.emnlp-main.1/ | @inproceedings{zhang-etal-2023-iag,
title = "{IAG}: Induction-Augmented Generation Framework for Answering Reasoning Questions",
author = "Zhang, Zhebin and
Zhang, Xinyu and
Ren, Yuanhang and
Shi, Saijiang and
Han, Meng and
Wu, Yongkang and
Lai, Ruofei and
Cao, Zhao",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.1",
doi = "10.18653/v1/2023.emnlp-main.1",
pages = "1--14",
abstract = "Retrieval-Augmented Generation (RAG), by incorporating external knowledge with parametric memory of language models, has become the state-of-the-art architecture for open-domain QA tasks. However, common knowledge bases are inherently constrained by limited coverage and noisy information, making retrieval-based approaches inadequate to answer implicit reasoning questions. In this paper, we propose an Induction-Augmented Generation (IAG) framework that utilizes inductive knowledge along with the retrieved documents for implicit reasoning. We leverage large language models (LLMs) for deriving such knowledge via a novel prompting method based on inductive reasoning patterns. On top of this, we implement two versions of IAG named IAG-GPT and IAG-Student, respectively. IAG-GPT directly utilizes the knowledge generated by GPT-3 for answer prediction, while IAG-Student gets rid of dependencies on GPT service at inference time by incorporating a student inductor model. The inductor is firstly trained via knowledge distillation and further optimized by back-propagating the generator feedback via differentiable beam scores. Experimental results show that IAG outperforms RAG baselines as well as ChatGPT on two Open-Domain QA tasks. Notably, our best models have won the first place in the official leaderboards of CSQA2.0 (since Nov 1, 2022) and StrategyQA (since Jan 8, 2023).",
}
| Retrieval-Augmented Generation (RAG), by incorporating external knowledge with parametric memory of language models, has become the state-of-the-art architecture for open-domain QA tasks. However, common knowledge bases are inherently constrained by limited coverage and noisy information, making retrieval-based approaches inadequate to answer implicit reasoning questions. In this paper, we propose an Induction-Augmented Generation (IAG) framework that utilizes inductive knowledge along with the retrieved documents for implicit reasoning. We leverage large language models (LLMs) for deriving such knowledge via a novel prompting method based on inductive reasoning patterns. On top of this, we implement two versions of IAG named IAG-GPT and IAG-Student, respectively. IAG-GPT directly utilizes the knowledge generated by GPT-3 for answer prediction, while IAG-Student gets rid of dependencies on GPT service at inference time by incorporating a student inductor model. The inductor is firstly trained via knowledge distillation and further optimized by back-propagating the generator feedback via differentiable beam scores. Experimental results show that IAG outperforms RAG baselines as well as ChatGPT on two Open-Domain QA tasks. Notably, our best models have won the first place in the official leaderboards of CSQA2.0 (since Nov 1, 2022) and StrategyQA (since Jan 8, 2023). | [
"Zhang, Zhebin",
"Zhang, Xinyu",
"Ren, Yuanhang",
"Shi, Saijiang",
"Han, Meng",
"Wu, Yongkang",
"Lai, Ruofei",
"Cao, Zhao"
] | IAG: Induction-Augmented Generation Framework for Answering Reasoning Questions | emnlp-main.1 | 2311.18397 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.2.bib | https://aclanthology.org/2023.emnlp-main.2/ | @inproceedings{yamamoto-matsuzaki-2023-absolute,
title = "Absolute Position Embedding Learns Sinusoid-like Waves for Attention Based on Relative Position",
author = "Yamamoto, Yuji and
Matsuzaki, Takuya",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.2",
doi = "10.18653/v1/2023.emnlp-main.2",
pages = "15--28",
abstract = "Attention weight is a clue to interpret how a Transformer-based model makes an inference. In some attention heads, the attention focuses on the neighbors of each token. This allows the output vector of each token to depend on the surrounding tokens and contributes to make the inference context-dependent. We analyze the mechanism behind the concentration of attention on nearby tokens. We show that the phenomenon emerges as follows: (1) learned position embedding has sinusoid-like components, (2) such components are transmitted to the query and the key in the self-attention, (3) the attention head shifts the phases of the sinusoid-like components so that the attention concentrates on nearby tokens at specific relative positions. In other words, a certain type of Transformer-based model acquires the sinusoidal positional encoding to some extent on its own through Masked Language Modeling.",
}
| Attention weight is a clue to interpret how a Transformer-based model makes an inference. In some attention heads, the attention focuses on the neighbors of each token. This allows the output vector of each token to depend on the surrounding tokens and contributes to make the inference context-dependent. We analyze the mechanism behind the concentration of attention on nearby tokens. We show that the phenomenon emerges as follows: (1) learned position embedding has sinusoid-like components, (2) such components are transmitted to the query and the key in the self-attention, (3) the attention head shifts the phases of the sinusoid-like components so that the attention concentrates on nearby tokens at specific relative positions. In other words, a certain type of Transformer-based model acquires the sinusoidal positional encoding to some extent on its own through Masked Language Modeling. | [
"Yamamoto, Yuji",
"Matsuzaki, Takuya"
] | Absolute Position Embedding Learns Sinusoid-like Waves for Attention Based on Relative Position | emnlp-main.2 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
||
https://aclanthology.org/2023.emnlp-main.3.bib | https://aclanthology.org/2023.emnlp-main.3/ | @inproceedings{qiang-etal-2023-chinese,
title = "{C}hinese Lexical Substitution: Dataset and Method",
author = "Qiang, Jipeng and
Liu, Kang and
Li, Ying and
Li, Yun and
Zhu, Yi and
Yuan, Yun-Hao and
Hu, Xiaocheng and
Ouyang, Xiaoye",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.3",
doi = "10.18653/v1/2023.emnlp-main.3",
pages = "29--42",
abstract = "Existing lexical substitution (LS) benchmarks were collected by asking human annotators to think of substitutes from memory, resulting in benchmarks with limited coverage and relatively small scales. To overcome this problem, we propose a novel annotation method to construct an LS dataset based on human and machine collaboration. Based on our annotation method, we construct the first Chinese LS dataset CHNLS which consists of 33,695 instances and 144,708 substitutes, covering three text genres (News, Novel, and Wikipedia). Specifically, we first combine four unsupervised LS methods as an ensemble method to generate the candidate substitutes, and then let human annotators judge these candidates or add new ones. This collaborative process combines the diversity of machine-generated substitutes with the expertise of human annotators. Experimental results that the ensemble method outperforms other LS methods. To our best knowledge, this is the first study for the Chinese LS task.",
}
| Existing lexical substitution (LS) benchmarks were collected by asking human annotators to think of substitutes from memory, resulting in benchmarks with limited coverage and relatively small scales. To overcome this problem, we propose a novel annotation method to construct an LS dataset based on human and machine collaboration. Based on our annotation method, we construct the first Chinese LS dataset CHNLS which consists of 33,695 instances and 144,708 substitutes, covering three text genres (News, Novel, and Wikipedia). Specifically, we first combine four unsupervised LS methods as an ensemble method to generate the candidate substitutes, and then let human annotators judge these candidates or add new ones. This collaborative process combines the diversity of machine-generated substitutes with the expertise of human annotators. Experimental results that the ensemble method outperforms other LS methods. To our best knowledge, this is the first study for the Chinese LS task. | [
"Qiang, Jipeng",
"Liu, Kang",
"Li, Ying",
"Li, Yun",
"Zhu, Yi",
"Yuan, Yun-Hao",
"Hu, Xiaocheng",
"Ouyang, Xiaoye"
] | Chinese Lexical Substitution: Dataset and Method | emnlp-main.3 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.4.bib | https://aclanthology.org/2023.emnlp-main.4/ | @inproceedings{sun-etal-2023-decoding,
title = "Decoding the Silent Majority: Inducing Belief Augmented Social Graph with Large Language Model for Response Forecasting",
author = "Sun, Chenkai and
Li, Jinning and
Fung, Yi and
Chan, Hou and
Abdelzaher, Tarek and
Zhai, ChengXiang and
Ji, Heng",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.4",
doi = "10.18653/v1/2023.emnlp-main.4",
pages = "43--57",
abstract = "Automatic response forecasting for news media plays a crucial role in enabling content producers to efficiently predict the impact of news releases and prevent unexpected negative outcomes such as social conflict and moral injury. To effectively forecast responses, it is essential to develop measures that leverage the social dynamics and contextual information surrounding individuals, especially in cases where explicit profiles or historical actions of the users are limited (referred to as lurkers). As shown in a previous study, 97{\%} of all tweets are produced by only the most active 25{\%} of users. However, existing approaches have limited exploration of how to best process and utilize these important features. To address this gap, we propose a novel framework, named SocialSense, that leverages a large language model to induce a belief-centered graph on top of an existent social network, along with graph-based propagation to capture social dynamics. We hypothesize that the induced graph that bridges the gap between distant users who share similar beliefs allows the model to effectively capture the response patterns. Our method surpasses existing state-of-the-art in experimental evaluations for both zero-shot and supervised settings, demonstrating its effectiveness in response forecasting. Moreover, the analysis reveals the framework{'}s capability to effectively handle unseen user and lurker scenarios, further highlighting its robustness and practical applicability.",
}
| Automatic response forecasting for news media plays a crucial role in enabling content producers to efficiently predict the impact of news releases and prevent unexpected negative outcomes such as social conflict and moral injury. To effectively forecast responses, it is essential to develop measures that leverage the social dynamics and contextual information surrounding individuals, especially in cases where explicit profiles or historical actions of the users are limited (referred to as lurkers). As shown in a previous study, 97{\%} of all tweets are produced by only the most active 25{\%} of users. However, existing approaches have limited exploration of how to best process and utilize these important features. To address this gap, we propose a novel framework, named SocialSense, that leverages a large language model to induce a belief-centered graph on top of an existent social network, along with graph-based propagation to capture social dynamics. We hypothesize that the induced graph that bridges the gap between distant users who share similar beliefs allows the model to effectively capture the response patterns. Our method surpasses existing state-of-the-art in experimental evaluations for both zero-shot and supervised settings, demonstrating its effectiveness in response forecasting. Moreover, the analysis reveals the framework{'}s capability to effectively handle unseen user and lurker scenarios, further highlighting its robustness and practical applicability. | [
"Sun, Chenkai",
"Li, Jinning",
"Fung, Yi",
"Chan, Hou",
"Abdelzaher, Tarek",
"Zhai, ChengXiang",
"Ji, Heng"
] | Decoding the Silent Majority: Inducing Belief Augmented Social Graph with Large Language Model for Response Forecasting | emnlp-main.4 | 2310.13297 | [
"https://github.com/chenkaisun/socialsense"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.5.bib | https://aclanthology.org/2023.emnlp-main.5/ | @inproceedings{yao-etal-2023-fine,
title = "Fine-grained Conversational Decoding via Isotropic and Proximal Search",
author = "Yao, Yuxuan and
Wu, Han and
Xu, Qiling and
Song, Linqi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.5",
doi = "10.18653/v1/2023.emnlp-main.5",
pages = "58--70",
abstract = "General-purpose text decoding approaches are usually adopted for dialogue response generation. Although the quality of the generated responses can be improved with dialogue-specific encoding methods, conversational decoding methods are still under-explored. Inspired by SimDRC that a good dialogue feature space should follow the rules of locality and isotropy, we present a fine-grained conversational decoding method, termed isotropic and proximal search (IPS). Our method is designed to generate the semantic-concentrated response, while still maintaining informativeness and discrimination against the context. Experiments show that our approach significantly outperforms existing decoding strategies in the dialogue field across both automatic and human evaluation metrics. More in-depth analyses further confirm the effectiveness of our approach.",
}
| General-purpose text decoding approaches are usually adopted for dialogue response generation. Although the quality of the generated responses can be improved with dialogue-specific encoding methods, conversational decoding methods are still under-explored. Inspired by SimDRC that a good dialogue feature space should follow the rules of locality and isotropy, we present a fine-grained conversational decoding method, termed isotropic and proximal search (IPS). Our method is designed to generate the semantic-concentrated response, while still maintaining informativeness and discrimination against the context. Experiments show that our approach significantly outperforms existing decoding strategies in the dialogue field across both automatic and human evaluation metrics. More in-depth analyses further confirm the effectiveness of our approach. | [
"Yao, Yuxuan",
"Wu, Han",
"Xu, Qiling",
"Song, Linqi"
] | Fine-grained Conversational Decoding via Isotropic and Proximal Search | emnlp-main.5 | 2310.08130 | [
"https://github.com/starrYYxuan/IPS"
] | https://huggingface.co/papers/2310.08130 | 0 | 0 | 0 | 4 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.6.bib | https://aclanthology.org/2023.emnlp-main.6/ | @inproceedings{stefanovitch-piskorski-2023-holistic,
title = "Holistic Inter-Annotator Agreement and Corpus Coherence Estimation in a Large-scale Multilingual Annotation Campaign",
author = "Stefanovitch, Nicolas and
Piskorski, Jakub",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.6",
doi = "10.18653/v1/2023.emnlp-main.6",
pages = "71--86",
abstract = "In this paper we report on the complexity of persuasion technique annotation in the context of a large multilingual annotation campaign involving 6 languages and approximately 40 annotators. We highlight the techniques that appear to be difficult for humans to annotate and elaborate on our findings on the causes of this phenomenon. We introduce Holistic IAA, a new word embedding-based annotator agreement metric and we report on various experiments using this metric and its correlation with the traditional Inter Annotator Agreement (IAA) metrics. However, given somewhat limited and loose interaction between annotators, i.e., only a few annotators annotate the same document subsets, we try to devise a way to assess the coherence of the entire dataset and strive to find a good proxy for IAA between annotators tasked to annotate different documents and in different languages, for which classical IAA metrics can not be applied.",
}
| In this paper we report on the complexity of persuasion technique annotation in the context of a large multilingual annotation campaign involving 6 languages and approximately 40 annotators. We highlight the techniques that appear to be difficult for humans to annotate and elaborate on our findings on the causes of this phenomenon. We introduce Holistic IAA, a new word embedding-based annotator agreement metric and we report on various experiments using this metric and its correlation with the traditional Inter Annotator Agreement (IAA) metrics. However, given somewhat limited and loose interaction between annotators, i.e., only a few annotators annotate the same document subsets, we try to devise a way to assess the coherence of the entire dataset and strive to find a good proxy for IAA between annotators tasked to annotate different documents and in different languages, for which classical IAA metrics can not be applied. | [
"Stefanovitch, Nicolas",
"Piskorski, Jakub"
] | Holistic Inter-Annotator Agreement and Corpus Coherence Estimation in a Large-scale Multilingual Annotation Campaign | emnlp-main.6 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.7.bib | https://aclanthology.org/2023.emnlp-main.7/ | @inproceedings{borenstein-etal-2023-phd,
title = "{PHD}: Pixel-Based Language Modeling of Historical Documents",
author = "Borenstein, Nadav and
Rust, Phillip and
Elliott, Desmond and
Augenstein, Isabelle",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.7",
doi = "10.18653/v1/2023.emnlp-main.7",
pages = "87--107",
abstract = "The digitisation of historical documents has provided historians with unprecedented research opportunities. Yet, the conventional approach to analysing historical documents involves converting them from images to text using OCR, a process that overlooks the potential benefits of treating them as images and introduces high levels of noise. To bridge this gap, we take advantage of recent advancements in pixel-based language models trained to reconstruct masked patches of pixels instead of predicting token distributions. Due to the scarcity of real historical scans, we propose a novel method for generating synthetic scans to resemble real historical documents. We then pre-train our model, PHD, on a combination of synthetic scans and real historical newspapers from the 1700-1900 period. Through our experiments, we demonstrate that PHD exhibits high proficiency in reconstructing masked image patches and provide evidence of our model{'}s noteworthy language understanding capabilities. Notably, we successfully apply our model to a historical QA task, highlighting its usefulness in this domain.",
}
| The digitisation of historical documents has provided historians with unprecedented research opportunities. Yet, the conventional approach to analysing historical documents involves converting them from images to text using OCR, a process that overlooks the potential benefits of treating them as images and introduces high levels of noise. To bridge this gap, we take advantage of recent advancements in pixel-based language models trained to reconstruct masked patches of pixels instead of predicting token distributions. Due to the scarcity of real historical scans, we propose a novel method for generating synthetic scans to resemble real historical documents. We then pre-train our model, PHD, on a combination of synthetic scans and real historical newspapers from the 1700-1900 period. Through our experiments, we demonstrate that PHD exhibits high proficiency in reconstructing masked image patches and provide evidence of our model{'}s noteworthy language understanding capabilities. Notably, we successfully apply our model to a historical QA task, highlighting its usefulness in this domain. | [
"Borenstein, Nadav",
"Rust, Phillip",
"Elliott, Desmond",
"Augenstein, Isabelle"
] | PHD: Pixel-Based Language Modeling of Historical Documents | emnlp-main.7 | 2310.18343 | [
"https://github.com/nadavborenstein/pixel-bw"
] | https://huggingface.co/papers/2310.18343 | 1 | 1 | 0 | 4 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.8.bib | https://aclanthology.org/2023.emnlp-main.8/ | @inproceedings{wang-etal-2023-primacy,
title = "Primacy Effect of {C}hat{GPT}",
author = "Wang, Yiwei and
Cai, Yujun and
Chen, Muhao and
Liang, Yuxuan and
Hooi, Bryan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.8",
doi = "10.18653/v1/2023.emnlp-main.8",
pages = "108--115",
abstract = "Instruction-tuned large language models (LLMs), such as ChatGPT, have led to promising zero-shot performance in discriminative natural language understanding (NLU) tasks. This involves querying the LLM using a prompt containing the question, and the candidate labels to choose from. The question-answering capabilities of ChatGPT arise from its pre-training on large amounts of human-written text, as well as its subsequent fine-tuning on human preferences, which motivates us to ask: Does ChatGPT also inherit humans{'} cognitive biases? In this paper, we study the primacy effect of ChatGPT: the tendency of selecting the labels at earlier positions as the answer. We have two main findings: i) ChatGPT{'}s decision is sensitive to the order of labels in the prompt; ii) ChatGPT has a clearly higher chance to select the labels at earlier positions as the answer. We hope that our experiments and analyses provide additional insights into building more reliable ChatGPT-based solutions. We release the source code at https://github.com/wangywUST/PrimacyEffectGPT.",
}
| Instruction-tuned large language models (LLMs), such as ChatGPT, have led to promising zero-shot performance in discriminative natural language understanding (NLU) tasks. This involves querying the LLM using a prompt containing the question, and the candidate labels to choose from. The question-answering capabilities of ChatGPT arise from its pre-training on large amounts of human-written text, as well as its subsequent fine-tuning on human preferences, which motivates us to ask: Does ChatGPT also inherit humans{'} cognitive biases? In this paper, we study the primacy effect of ChatGPT: the tendency of selecting the labels at earlier positions as the answer. We have two main findings: i) ChatGPT{'}s decision is sensitive to the order of labels in the prompt; ii) ChatGPT has a clearly higher chance to select the labels at earlier positions as the answer. We hope that our experiments and analyses provide additional insights into building more reliable ChatGPT-based solutions. We release the source code at https://github.com/wangywUST/PrimacyEffectGPT. | [
"Wang, Yiwei",
"Cai, Yujun",
"Chen, Muhao",
"Liang, Yuxuan",
"Hooi, Bryan"
] | Primacy Effect of ChatGPT | emnlp-main.8 | 2310.13206 | [
"https://github.com/wangywust/primacyeffectgpt"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.9.bib | https://aclanthology.org/2023.emnlp-main.9/ | @inproceedings{kawabata-sugawara-2023-evaluating,
title = "Evaluating the Rationale Understanding of Critical Reasoning in Logical Reading Comprehension",
author = "Kawabata, Akira and
Sugawara, Saku",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.9",
doi = "10.18653/v1/2023.emnlp-main.9",
pages = "116--143",
abstract = "To precisely evaluate a language model{'}s capability for logical reading comprehension, we present a dataset for testing the understanding of the rationale behind critical reasoning. For questions taken from an existing multiple-choice logical reading comprehension dataset, we crowdsource rationale texts that explain why we should select or eliminate answer options, resulting in 3,003 multiple-choice subquestions that are associated with 943 main questions. Experiments on our dataset show that recent large language models (e.g., InstructGPT) struggle to answer the subquestions even if they are able to answer the main questions correctly. We find that the models perform particularly poorly in answering subquestions written for the incorrect options of the main questions, implying that the models have a limited capability for explaining why incorrect alternatives should be eliminated. These results suggest that our dataset encourages further investigation into the critical reasoning ability of language models while focusing on the elimination process of relevant alternatives.",
}
| To precisely evaluate a language model{'}s capability for logical reading comprehension, we present a dataset for testing the understanding of the rationale behind critical reasoning. For questions taken from an existing multiple-choice logical reading comprehension dataset, we crowdsource rationale texts that explain why we should select or eliminate answer options, resulting in 3,003 multiple-choice subquestions that are associated with 943 main questions. Experiments on our dataset show that recent large language models (e.g., InstructGPT) struggle to answer the subquestions even if they are able to answer the main questions correctly. We find that the models perform particularly poorly in answering subquestions written for the incorrect options of the main questions, implying that the models have a limited capability for explaining why incorrect alternatives should be eliminated. These results suggest that our dataset encourages further investigation into the critical reasoning ability of language models while focusing on the elimination process of relevant alternatives. | [
"Kawabata, Akira",
"Sugawara, Saku"
] | Evaluating the Rationale Understanding of Critical Reasoning in Logical Reading Comprehension | emnlp-main.9 | 2311.18353 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.10.bib | https://aclanthology.org/2023.emnlp-main.10/ | @inproceedings{muller-etal-2023-evaluating,
title = "Evaluating and Modeling Attribution for Cross-Lingual Question Answering",
author = "Muller, Benjamin and
Wieting, John and
Clark, Jonathan and
Kwiatkowski, Tom and
Ruder, Sebastian and
Soares, Livio and
Aharoni, Roee and
Herzig, Jonathan and
Wang, Xinyi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.10",
doi = "10.18653/v1/2023.emnlp-main.10",
pages = "144--157",
abstract = "Trustworthy answer content is abundant in many high-resource languages and is instantly accessible through question answering systems {---} yet this content can be hard to access for those that do not speak these languages. The leap forward in cross-lingual modeling quality offered by generative language models offers much promise, yet their raw generations often fall short in factuality. To improve trustworthiness in these systems, a promising direction is to attribute the answer to a retrieved source, possibly in a content-rich language different from the query. Our work is the first to study attribution for cross-lingual question answering. First, we collect data in 5 languages to assess the attribution level of a state-of-the-art cross-lingual QA system. To our surprise, we find that a substantial portion of the answers is not attributable to any retrieved passages (up to 50{\%} of answers exactly matching a gold reference) despite the system being able to attend directly to the retrieved text. Second, to address this poor attribution level, we experiment with a wide range of attribution detection techniques. We find that Natural Language Inference models and PaLM 2 fine-tuned on a very small amount of attribution data can accurately detect attribution. With these models, we improve the attribution level of a cross-lingual QA system. Overall, we show that current academic generative cross-lingual QA systems have substantial shortcomings in attribution and we build tooling to mitigate these issues.",
}
| Trustworthy answer content is abundant in many high-resource languages and is instantly accessible through question answering systems {---} yet this content can be hard to access for those that do not speak these languages. The leap forward in cross-lingual modeling quality offered by generative language models offers much promise, yet their raw generations often fall short in factuality. To improve trustworthiness in these systems, a promising direction is to attribute the answer to a retrieved source, possibly in a content-rich language different from the query. Our work is the first to study attribution for cross-lingual question answering. First, we collect data in 5 languages to assess the attribution level of a state-of-the-art cross-lingual QA system. To our surprise, we find that a substantial portion of the answers is not attributable to any retrieved passages (up to 50{\%} of answers exactly matching a gold reference) despite the system being able to attend directly to the retrieved text. Second, to address this poor attribution level, we experiment with a wide range of attribution detection techniques. We find that Natural Language Inference models and PaLM 2 fine-tuned on a very small amount of attribution data can accurately detect attribution. With these models, we improve the attribution level of a cross-lingual QA system. Overall, we show that current academic generative cross-lingual QA systems have substantial shortcomings in attribution and we build tooling to mitigate these issues. | [
"Muller, Benjamin",
"Wieting, John",
"Clark, Jonathan",
"Kwiatkowski, Tom",
"Ruder, Sebastian",
"Soares, Livio",
"Aharoni, Roee",
"Herzig, Jonathan",
"Wang, Xinyi"
] | Evaluating and Modeling Attribution for Cross-Lingual Question Answering | emnlp-main.10 | 2305.14332 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.11.bib | https://aclanthology.org/2023.emnlp-main.11/ | @inproceedings{oladipo-etal-2023-better,
title = "Better Quality Pre-training Data and T5 Models for {A}frican Languages",
author = "Oladipo, Akintunde and
Adeyemi, Mofetoluwa and
Ahia, Orevaoghene and
Owodunni, Abraham and
Ogundepo, Odunayo and
Adelani, David and
Lin, Jimmy",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.11",
doi = "10.18653/v1/2023.emnlp-main.11",
pages = "158--168",
abstract = "In this study, we highlight the importance of enhancing the quality of pretraining data in multilingual language models. Existing web crawls have demonstrated quality issues, particularly in the context of low-resource languages. Consequently, we introduce a new multilingual pretraining corpus for 16 African languages, designed by carefully auditing existing pretraining corpora to understand and rectify prevalent quality issues. To compile this dataset, we undertake a rigorous examination of current data sources for thirteen languages within one of the most extensive multilingual web crawls, mC4, and extract cleaner data through meticulous auditing and improved web crawling strategies. Subsequently, we pretrain a new T5-based model on this dataset and evaluate its performance on multiple downstream tasks. Our model demonstrates better downstream effectiveness over existing pretrained models across four NLP tasks, underscoring the critical role data quality plays in pretraining language models in low-resource scenarios. Specifically, on cross-lingual QA evaluation, our new model is more than twice as effective as multilingual T5. All code, data and models are publicly available at https://github.com/castorini/AfriTeVa-keji.",
}
| In this study, we highlight the importance of enhancing the quality of pretraining data in multilingual language models. Existing web crawls have demonstrated quality issues, particularly in the context of low-resource languages. Consequently, we introduce a new multilingual pretraining corpus for 16 African languages, designed by carefully auditing existing pretraining corpora to understand and rectify prevalent quality issues. To compile this dataset, we undertake a rigorous examination of current data sources for thirteen languages within one of the most extensive multilingual web crawls, mC4, and extract cleaner data through meticulous auditing and improved web crawling strategies. Subsequently, we pretrain a new T5-based model on this dataset and evaluate its performance on multiple downstream tasks. Our model demonstrates better downstream effectiveness over existing pretrained models across four NLP tasks, underscoring the critical role data quality plays in pretraining language models in low-resource scenarios. Specifically, on cross-lingual QA evaluation, our new model is more than twice as effective as multilingual T5. All code, data and models are publicly available at https://github.com/castorini/AfriTeVa-keji. | [
"Oladipo, Akintunde",
"Adeyemi, Mofetoluwa",
"Ahia, Orevaoghene",
"Owodunni, Abraham",
"Ogundepo, Odunayo",
"Adelani, David",
"Lin, Jimmy"
] | Better Quality Pre-training Data and T5 Models for African Languages | emnlp-main.11 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.12.bib | https://aclanthology.org/2023.emnlp-main.12/ | @inproceedings{tan-etal-2023-sparse,
title = "Sparse Universal Transformer",
author = "Tan, Shawn and
Shen, Yikang and
Chen, Zhenfang and
Courville, Aaron and
Gan, Chuang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.12",
doi = "10.18653/v1/2023.emnlp-main.12",
pages = "169--179",
abstract = "The Universal Transformer (UT) is a variant of the Transformer that shares parameters across its layers and is Turing-complete under certain assumptions. Empirical evidence also shows that UTs have better compositional generalization than Vanilla Transformers (VTs) in formal language tasks. The parameter-sharing also affords it better parameter efficiency than VTs. Despite its many advantages, most state-of-the-art NLP systems use VTs as their backbone model instead of UTs. This is mainly because scaling UT parameters is more compute and memory intensive than scaling up a VT. This paper proposes the Sparse Universal Transformer (SUT), which leverages Sparse Mixture of Experts (SMoE) to reduce UT{'}s computation complexity while retaining its parameter efficiency and generalization ability. Experiments show that SUT combines the best of both worlds, achieving strong generalization results on formal language tasks (Logical inference and CFQ) and impressive parameter and computation efficiency on standard natural language benchmarks like WMT{'}14.",
}
| The Universal Transformer (UT) is a variant of the Transformer that shares parameters across its layers and is Turing-complete under certain assumptions. Empirical evidence also shows that UTs have better compositional generalization than Vanilla Transformers (VTs) in formal language tasks. The parameter-sharing also affords it better parameter efficiency than VTs. Despite its many advantages, most state-of-the-art NLP systems use VTs as their backbone model instead of UTs. This is mainly because scaling UT parameters is more compute and memory intensive than scaling up a VT. This paper proposes the Sparse Universal Transformer (SUT), which leverages Sparse Mixture of Experts (SMoE) to reduce UT{'}s computation complexity while retaining its parameter efficiency and generalization ability. Experiments show that SUT combines the best of both worlds, achieving strong generalization results on formal language tasks (Logical inference and CFQ) and impressive parameter and computation efficiency on standard natural language benchmarks like WMT{'}14. | [
"Tan, Shawn",
"Shen, Yikang",
"Chen, Zhenfang",
"Courville, Aaron",
"Gan, Chuang"
] | Sparse Universal Transformer | emnlp-main.12 | 2310.07096 | [
""
] | https://huggingface.co/papers/2310.07096 | 1 | 0 | 0 | 5 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.13.bib | https://aclanthology.org/2023.emnlp-main.13/ | @inproceedings{li-etal-2023-theory,
title = "Theory of Mind for Multi-Agent Collaboration via Large Language Models",
author = "Li, Huao and
Chong, Yu and
Stepputtis, Simon and
Campbell, Joseph and
Hughes, Dana and
Lewis, Charles and
Sycara, Katia",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.13",
doi = "10.18653/v1/2023.emnlp-main.13",
pages = "180--192",
abstract = "While Large Language Models (LLMs) have demonstrated impressive accomplishments in both reasoning and planning, their abilities in multi-agent collaborations remains largely unexplored. This study evaluates LLM-based agents in a multi-agent cooperative text game with Theory of Mind (ToM) inference tasks, comparing their performance with Multi-Agent Reinforcement Learning (MARL) and planning-based baselines. We observed evidence of emergent collaborative behaviors and high-order Theory of Mind capabilities among LLM-based agents. Our results reveal limitations in LLM-based agents{'} planning optimization due to systematic failures in managing long-horizon contexts and hallucination about the task state. We explore the use of explicit belief state representations to mitigate these issues, finding that it enhances task performance and the accuracy of ToM inferences for LLM-based agents.",
}
| While Large Language Models (LLMs) have demonstrated impressive accomplishments in both reasoning and planning, their abilities in multi-agent collaborations remains largely unexplored. This study evaluates LLM-based agents in a multi-agent cooperative text game with Theory of Mind (ToM) inference tasks, comparing their performance with Multi-Agent Reinforcement Learning (MARL) and planning-based baselines. We observed evidence of emergent collaborative behaviors and high-order Theory of Mind capabilities among LLM-based agents. Our results reveal limitations in LLM-based agents{'} planning optimization due to systematic failures in managing long-horizon contexts and hallucination about the task state. We explore the use of explicit belief state representations to mitigate these issues, finding that it enhances task performance and the accuracy of ToM inferences for LLM-based agents. | [
"Li, Huao",
"Chong, Yu",
"Stepputtis, Simon",
"Campbell, Joseph",
"Hughes, Dana",
"Lewis, Charles",
"Sycara, Katia"
] | Theory of Mind for Multi-Agent Collaboration via Large Language Models | emnlp-main.13 | 2310.10701 | [
"https://github.com/romanlee6/multi_LLM_comm"
] | https://huggingface.co/papers/2310.10701 | 0 | 0 | 0 | 7 | [] | [] | [
"agentharbor/agenta"
] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.14.bib | https://aclanthology.org/2023.emnlp-main.14/ | @inproceedings{litschko-etal-2023-establishing,
title = "Establishing Trustworthiness: Rethinking Tasks and Model Evaluation",
author = {Litschko, Robert and
M{\"u}ller-Eberstein, Max and
van der Goot, Rob and
Weber-Genzel, Leon and
Plank, Barbara},
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.14",
doi = "10.18653/v1/2023.emnlp-main.14",
pages = "193--203",
abstract = "Language understanding is a multi-faceted cognitive capability, which the Natural Language Processing (NLP) community has striven to model computationally for decades. Traditionally, facets of linguistic intelligence have been compartmentalized into tasks with specialized model architectures and corresponding evaluation protocols. With the advent of large language models (LLMs) the community has witnessed a dramatic shift towards general purpose, task-agnostic approaches powered by generative models. As a consequence, the traditional compartmentalized notion of language tasks is breaking down, followed by an increasing challenge for evaluation and analysis. At the same time, LLMs are being deployed in more real-world scenarios, including previously unforeseen zero-shot setups, increasing the need for trustworthy and reliable systems. Therefore, we argue that it is time to rethink what constitutes tasks and model evaluation in NLP, and pursue a more holistic view on language, placing trustworthiness at the center. Towards this goal, we review existing compartmentalized approaches for understanding the origins of a model{'}s functional capacity, and provide recommendations for more multi-faceted evaluation protocols.",
}
| Language understanding is a multi-faceted cognitive capability, which the Natural Language Processing (NLP) community has striven to model computationally for decades. Traditionally, facets of linguistic intelligence have been compartmentalized into tasks with specialized model architectures and corresponding evaluation protocols. With the advent of large language models (LLMs) the community has witnessed a dramatic shift towards general purpose, task-agnostic approaches powered by generative models. As a consequence, the traditional compartmentalized notion of language tasks is breaking down, followed by an increasing challenge for evaluation and analysis. At the same time, LLMs are being deployed in more real-world scenarios, including previously unforeseen zero-shot setups, increasing the need for trustworthy and reliable systems. Therefore, we argue that it is time to rethink what constitutes tasks and model evaluation in NLP, and pursue a more holistic view on language, placing trustworthiness at the center. Towards this goal, we review existing compartmentalized approaches for understanding the origins of a model{'}s functional capacity, and provide recommendations for more multi-faceted evaluation protocols. | [
"Litschko, Robert",
"M{\\\"u}ller-Eberstein, Max",
"van der Goot, Rob",
"Weber-Genzel, Leon",
"Plank, Barbara"
] | Establishing Trustworthiness: Rethinking Tasks and Model Evaluation | emnlp-main.14 | 2310.05442 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.15.bib | https://aclanthology.org/2023.emnlp-main.15/ | @inproceedings{himakunthala-etal-2023-lets,
title = "Let{'}s Think Frame by Frame with {VIP}: A Video Infilling and Prediction Dataset for Evaluating Video Chain-of-Thought",
author = "Himakunthala, Vaishnavi and
Ouyang, Andy and
Rose, Daniel and
He, Ryan and
Mei, Alex and
Lu, Yujie and
Sonar, Chinmay and
Saxon, Michael and
Wang, William",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.15",
doi = "10.18653/v1/2023.emnlp-main.15",
pages = "204--219",
abstract = "Despite exciting recent results showing vision-language systems{'} capacity to reason about images using natural language, their capacity for video reasoning remains underexplored. We motivate framing video reasoning as the sequential understanding of a small number of keyframes, thereby leveraging the power and robustness of vision-language while alleviating the computational complexities of processing videos. To evaluate this novel application, we introduce VIP, an inference-time challenge dataset designed to explore models{'} reasoning capabilities through video chain-of-thought. Inspired by visually descriptive scene plays, we propose two formats for keyframe description: unstructured dense captions and structured scene descriptions that identify the focus, action, mood, objects, and setting (FAMOuS) of the keyframe. To evaluate video reasoning, we propose two tasks: Video Infilling and Video Prediction, which test abilities to generate multiple intermediate keyframes and predict future keyframes, respectively. We benchmark GPT-4, GPT-3, and VICUNA on VIP, demonstrate the performance gap in these complex video reasoning tasks, and encourage future work to prioritize language models for efficient and generalized video reasoning.",
}
| Despite exciting recent results showing vision-language systems{'} capacity to reason about images using natural language, their capacity for video reasoning remains underexplored. We motivate framing video reasoning as the sequential understanding of a small number of keyframes, thereby leveraging the power and robustness of vision-language while alleviating the computational complexities of processing videos. To evaluate this novel application, we introduce VIP, an inference-time challenge dataset designed to explore models{'} reasoning capabilities through video chain-of-thought. Inspired by visually descriptive scene plays, we propose two formats for keyframe description: unstructured dense captions and structured scene descriptions that identify the focus, action, mood, objects, and setting (FAMOuS) of the keyframe. To evaluate video reasoning, we propose two tasks: Video Infilling and Video Prediction, which test abilities to generate multiple intermediate keyframes and predict future keyframes, respectively. We benchmark GPT-4, GPT-3, and VICUNA on VIP, demonstrate the performance gap in these complex video reasoning tasks, and encourage future work to prioritize language models for efficient and generalized video reasoning. | [
"Himakunthala, Vaishnavi",
"Ouyang, Andy",
"Rose, Daniel",
"He, Ryan",
"Mei, Alex",
"Lu, Yujie",
"Sonar, Chinmay",
"Saxon, Michael",
"Wang, William"
] | Let's Think Frame by Frame with VIP: A Video Infilling and Prediction Dataset for Evaluating Video Chain-of-Thought | emnlp-main.15 | 2305.13903 | [
"https://github.com/vaishnavihimakunthala/vip"
] | https://huggingface.co/papers/2305.13903 | 2 | 0 | 0 | 9 | [] | [
"ryanhe/VIP"
] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.16.bib | https://aclanthology.org/2023.emnlp-main.16/ | @inproceedings{khondaker-etal-2023-gptaraeval,
title = "{GPTA}ra{E}val: A Comprehensive Evaluation of {C}hat{GPT} on {A}rabic {NLP}",
author = "Khondaker, Md Tawkat Islam and
Waheed, Abdul and
Nagoudi, El Moatez Billah and
Abdul-Mageed, Muhammad",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.16",
doi = "10.18653/v1/2023.emnlp-main.16",
pages = "220--247",
abstract = "ChatGPT{'}s emergence heralds a transformative phase in NLP, particularly demonstrated through its excellent performance on many English benchmarks. However, the model{'}s efficacy across diverse linguistic contexts remains largely uncharted territory. This work aims to bridge this knowledge gap, with a primary focus on assessing ChatGPT{'}s capabilities on Arabic languages and dialectal varieties. Our comprehensive study conducts a large-scale automated and human evaluation of ChatGPT, encompassing 44 distinct language understanding and generation tasks on over 60 different datasets. To our knowledge, this marks the first extensive performance analysis of ChatGPT{'}s deployment in Arabic NLP. Our findings indicate that, despite its remarkable performance in English, ChatGPT is consistently surpassed by smaller models that have undergone finetuning on Arabic. We further undertake a meticulous comparison of ChatGPT and GPT-4{'}s Modern Standard Arabic (MSA) and Dialectal Arabic (DA), unveiling the relative shortcomings of both models in handling Arabic dialects compared to MSA. Although we further explore and confirm the utility of employing GPT-4 as a potential alternative for human evaluation, our work adds to a growing body of research underscoring the limitations of ChatGPT.",
}
| ChatGPT{'}s emergence heralds a transformative phase in NLP, particularly demonstrated through its excellent performance on many English benchmarks. However, the model{'}s efficacy across diverse linguistic contexts remains largely uncharted territory. This work aims to bridge this knowledge gap, with a primary focus on assessing ChatGPT{'}s capabilities on Arabic languages and dialectal varieties. Our comprehensive study conducts a large-scale automated and human evaluation of ChatGPT, encompassing 44 distinct language understanding and generation tasks on over 60 different datasets. To our knowledge, this marks the first extensive performance analysis of ChatGPT{'}s deployment in Arabic NLP. Our findings indicate that, despite its remarkable performance in English, ChatGPT is consistently surpassed by smaller models that have undergone finetuning on Arabic. We further undertake a meticulous comparison of ChatGPT and GPT-4{'}s Modern Standard Arabic (MSA) and Dialectal Arabic (DA), unveiling the relative shortcomings of both models in handling Arabic dialects compared to MSA. Although we further explore and confirm the utility of employing GPT-4 as a potential alternative for human evaluation, our work adds to a growing body of research underscoring the limitations of ChatGPT. | [
"Khondaker, Md Tawkat Islam",
"Waheed, Abdul",
"Nagoudi, El Moatez Billah",
"Abdul-Mageed, Muhammad"
] | GPTAraEval: A Comprehensive Evaluation of ChatGPT on Arabic NLP | emnlp-main.16 | 2305.14976 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.17.bib | https://aclanthology.org/2023.emnlp-main.17/ | @inproceedings{li-etal-2023-dual-channel,
title = "Dual-Channel Span for Aspect Sentiment Triplet Extraction",
author = "Li, Pan and
Li, Ping and
Zhang, Kai",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.17",
doi = "10.18653/v1/2023.emnlp-main.17",
pages = "248--261",
abstract = "Aspect Sentiment Triplet Extraction (ASTE) is one of the compound tasks of fine-grained aspect-based sentiment analysis (ABSA), aiming at extracting the triplets of aspect terms, corresponding opinion terms and the associated sentiment orientation. Recent efforts in exploiting span-level semantic interaction shown superior performance on ASTE task. However, most of the existing span-based approaches suffer from enumerating all possible spans, since it can introduce too much noise in sentiment triplet extraction. To ease this burden, we propose a dual-channel span generation method to coherently constrain the search space of span candidates. Specifically, we leverage the syntactic relations among aspect/opinion terms and the associated part-of-speech characteristics in those terms to generate span candidates, which reduces span enumeration by nearly half. Besides, feature representations are learned from syntactic and part-of-speech correlation among terms, which renders span representation fruitful linguistic information. Extensive experiments on two versions of public datasets demonstrate both the effectiveness of our design and the superiority on ASTE/ATE/OTE tasks.",
}
| Aspect Sentiment Triplet Extraction (ASTE) is one of the compound tasks of fine-grained aspect-based sentiment analysis (ABSA), aiming at extracting the triplets of aspect terms, corresponding opinion terms and the associated sentiment orientation. Recent efforts in exploiting span-level semantic interaction shown superior performance on ASTE task. However, most of the existing span-based approaches suffer from enumerating all possible spans, since it can introduce too much noise in sentiment triplet extraction. To ease this burden, we propose a dual-channel span generation method to coherently constrain the search space of span candidates. Specifically, we leverage the syntactic relations among aspect/opinion terms and the associated part-of-speech characteristics in those terms to generate span candidates, which reduces span enumeration by nearly half. Besides, feature representations are learned from syntactic and part-of-speech correlation among terms, which renders span representation fruitful linguistic information. Extensive experiments on two versions of public datasets demonstrate both the effectiveness of our design and the superiority on ASTE/ATE/OTE tasks. | [
"Li, Pan",
"Li, Ping",
"Zhang, Kai"
] | Dual-Channel Span for Aspect Sentiment Triplet Extraction | emnlp-main.17 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.18.bib | https://aclanthology.org/2023.emnlp-main.18/ | @inproceedings{li-zhang-2023-cultural,
title = "Cultural Concept Adaptation on Multimodal Reasoning",
author = "Li, Zhi and
Zhang, Yin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.18",
doi = "10.18653/v1/2023.emnlp-main.18",
pages = "262--276",
abstract = "Developing cultural adaptation methods is important, which can improve the model performance on the low-resource ones and provide more equitable opportunities for everyone to benefit from advanced technology. Past methods primarily focused on multilingual and multimodal capabilities, and the improvement of multicultural competence is still an unexplored problem. This is largely due to the difficulty of data scarcity and expensive annotation. In this paper, we navigate this uncharted territory by leveraging high-resource cultures to facilitate comprehension of low-resource ones. We first introduce an annotation-free method for cultural-concept adaptation and construct a concept mapping set. To facilitate the model{'}s comprehension of cultural-concept mappings, we propose a new multimodal data augmentation called CultureMixup. This approach employs a three-tier code-switching strategy on textual sentences. Additionally, it uses a cultural concept-based mixup method for the images. This combination effectively generates new data instances across culture, phrase, word, and image levels. For visually grounded reasoning across languages and cultures, experimental results on five languages show that our method consistently improves performance for four existing multilingual and multimodal models on both zero-shot and few-shot settings.",
}
| Developing cultural adaptation methods is important, which can improve the model performance on the low-resource ones and provide more equitable opportunities for everyone to benefit from advanced technology. Past methods primarily focused on multilingual and multimodal capabilities, and the improvement of multicultural competence is still an unexplored problem. This is largely due to the difficulty of data scarcity and expensive annotation. In this paper, we navigate this uncharted territory by leveraging high-resource cultures to facilitate comprehension of low-resource ones. We first introduce an annotation-free method for cultural-concept adaptation and construct a concept mapping set. To facilitate the model{'}s comprehension of cultural-concept mappings, we propose a new multimodal data augmentation called CultureMixup. This approach employs a three-tier code-switching strategy on textual sentences. Additionally, it uses a cultural concept-based mixup method for the images. This combination effectively generates new data instances across culture, phrase, word, and image levels. For visually grounded reasoning across languages and cultures, experimental results on five languages show that our method consistently improves performance for four existing multilingual and multimodal models on both zero-shot and few-shot settings. | [
"Li, Zhi",
"Zhang, Yin"
] | Cultural Concept Adaptation on Multimodal Reasoning | emnlp-main.18 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
||
https://aclanthology.org/2023.emnlp-main.19.bib | https://aclanthology.org/2023.emnlp-main.19/ | @inproceedings{samir-silfverberg-2023-understanding,
title = "Understanding Compositional Data Augmentation in Typologically Diverse Morphological Inflection",
author = "Samir, Farhan and
Silfverberg, Miikka",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.19",
doi = "10.18653/v1/2023.emnlp-main.19",
pages = "277--291",
abstract = "Data augmentation techniques are widely used in low-resource automatic morphological inflection to address the issue of data sparsity. However, the full implications of these techniques remain poorly understood. In this study, we aim to shed light on the theoretical aspects of the data augmentation strategy StemCorrupt, a method that generates synthetic examples by randomly substituting stem characters in existing gold standard training examples. Our analysis uncovers that StemCorrupt brings about fundamental changes in the underlying data distribution, revealing inherent compositional concatenative structure. To complement our theoretical analysis, we investigate the data-efficiency of StemCorrupt. Through evaluation across a diverse set of seven typologically distinct languages, we demonstrate that selecting a subset of datapoints with both high diversity \textit{and} high predictive uncertainty significantly enhances the data-efficiency of compared to competitive baselines. Furthermore, we explore the impact of typological features on the choice of augmentation strategy and find that languages incorporating non-concatenativity, such as morphonological alternations, derive less benefit from synthetic examples with high predictive uncertainty. We attribute this effect to phonotactic violations induced by StemCorrupt, emphasizing the need for further research to ensure optimal performance across the entire spectrum of natural language morphology.",
}
| Data augmentation techniques are widely used in low-resource automatic morphological inflection to address the issue of data sparsity. However, the full implications of these techniques remain poorly understood. In this study, we aim to shed light on the theoretical aspects of the data augmentation strategy StemCorrupt, a method that generates synthetic examples by randomly substituting stem characters in existing gold standard training examples. Our analysis uncovers that StemCorrupt brings about fundamental changes in the underlying data distribution, revealing inherent compositional concatenative structure. To complement our theoretical analysis, we investigate the data-efficiency of StemCorrupt. Through evaluation across a diverse set of seven typologically distinct languages, we demonstrate that selecting a subset of datapoints with both high diversity \textit{and} high predictive uncertainty significantly enhances the data-efficiency of compared to competitive baselines. Furthermore, we explore the impact of typological features on the choice of augmentation strategy and find that languages incorporating non-concatenativity, such as morphonological alternations, derive less benefit from synthetic examples with high predictive uncertainty. We attribute this effect to phonotactic violations induced by StemCorrupt, emphasizing the need for further research to ensure optimal performance across the entire spectrum of natural language morphology. | [
"Samir, Farhan",
"Silfverberg, Miikka"
] | Understanding Compositional Data Augmentation in Typologically Diverse Morphological Inflection | emnlp-main.19 | 2305.13658 | [
"https://github.com/smfsamir/understanding-augmentation-morphology"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.20.bib | https://aclanthology.org/2023.emnlp-main.20/ | @inproceedings{li-etal-2023-evaluating,
title = "Evaluating Object Hallucination in Large Vision-Language Models",
author = "Li, Yifan and
Du, Yifan and
Zhou, Kun and
Wang, Jinpeng and
Zhao, Xin and
Wen, Ji-Rong",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.20",
doi = "10.18653/v1/2023.emnlp-main.20",
pages = "292--305",
abstract = "Inspired by the superior language abilities of large language models (LLM), large vision-language models (LVLM) have been recently proposed by integrating powerful LLMs for improving the performance on complex multimodal tasks. Despite the promising progress on LVLMs, we find that they suffer from object hallucinations, i.e., they tend to generate objects inconsistent with the target images in the descriptions. To investigate it, this work presents the first systematic study on object hallucination of LVLMs. We conduct the evaluation experiments on several representative LVLMs, and show that they mostly suffer from severe object hallucination issues. We further discuss that the visual instructions may influence the hallucination, and find that: objects that frequently appear in the visual instructions or co-occur with the image objects are obviously prone to be hallucinated by LVLMs. Besides, we further design a polling-based query method called POPE for better evaluation of object hallucination. Experiment results show that our POPE can evaluate object hallucination in a more stable and flexible way.",
}
| Inspired by the superior language abilities of large language models (LLM), large vision-language models (LVLM) have been recently proposed by integrating powerful LLMs for improving the performance on complex multimodal tasks. Despite the promising progress on LVLMs, we find that they suffer from object hallucinations, i.e., they tend to generate objects inconsistent with the target images in the descriptions. To investigate it, this work presents the first systematic study on object hallucination of LVLMs. We conduct the evaluation experiments on several representative LVLMs, and show that they mostly suffer from severe object hallucination issues. We further discuss that the visual instructions may influence the hallucination, and find that: objects that frequently appear in the visual instructions or co-occur with the image objects are obviously prone to be hallucinated by LVLMs. Besides, we further design a polling-based query method called POPE for better evaluation of object hallucination. Experiment results show that our POPE can evaluate object hallucination in a more stable and flexible way. | [
"Li, Yifan",
"Du, Yifan",
"Zhou, Kun",
"Wang, Jinpeng",
"Zhao, Xin",
"Wen, Ji-Rong"
] | Evaluating Object Hallucination in Large Vision-Language Models | emnlp-main.20 | 2305.10355 | [
"https://github.com/rucaibox/pope"
] | https://huggingface.co/papers/2305.10355 | 0 | 0 | 0 | 6 | [
"google/paligemma-3b-pt-224",
"google/paligemma-3b-pt-896",
"google/paligemma-3b-mix-448",
"google/paligemma-3b-mix-224",
"google/paligemma-3b-pt-448",
"google/paligemma-3b-ft-ocrvqa-896",
"google/paligemma-3b-ft-vqav2-448",
"google/paligemma-3b-ft-refcoco-seg-896",
"google/paligemma-3b-ft-ocrvqa-448",
"google/paligemma-3b-ft-cococap-448",
"google/paligemma-3b-ft-ai2d-224-jax",
"google/paligemma-3b-ft-docvqa-896",
"google/paligemma-3b-ft-vizwizvqa-448-jax",
"google/paligemma-3b-ft-vqav2-224",
"google/paligemma-3b-ft-widgetcap-448",
"google/paligemma-3b-ft-rsvqa-hr-224",
"google/paligemma-3b-pt-896-jax",
"google/paligemma-3b-ft-vqav2-224-jax",
"google/paligemma-3b-ft-docvqa-896-jax",
"google/paligemma-3b-ft-widgetcap-448-jax",
"google/paligemma-3b-ft-vizwizvqa-448",
"google/paligemma-3b-ft-nlvr2-224",
"google/paligemma-3b-ft-refcoco-seg-448-jax",
"google/paligemma-3b-ft-vqav2-448-jax",
"google/paligemma-3b-ft-okvqa-224",
"google/paligemma-3b-ft-ocrvqa-224",
"google/paligemma-3b-ft-cococap-224",
"google/paligemma-3b-ft-ocrvqa-896-jax",
"google/paligemma-3b-ft-science-qa-224",
"google/paligemma-3b-ft-textvqa-896-jax",
"google/paligemma-3b-ft-coco35l-224",
"google/paligemma-3b-ft-docvqa-448",
"google/paligemma-3b-ft-science-qa-448",
"google/paligemma-3b-ft-widgetcap-224",
"google/paligemma-3b-ft-textcaps-448",
"google/paligemma-3b-ft-textvqa-896",
"leo009/paligemma-3b-mix-224",
"google/paligemma-3b-ft-infovqa-896-jax",
"google/paligemma-3b-ft-screen2words-224",
"google/paligemma-3b-ft-screen2words-448-jax",
"google/paligemma-3b-ft-textvqa-448",
"google/paligemma-3b-ft-rsvqa-lr-224-jax",
"google/paligemma-3b-ft-tallyqa-224-jax",
"google/paligemma-3b-ft-refcoco-seg-896-jax",
"google/paligemma-3b-ft-scicap-224",
"google/paligemma-3b-ft-okvqa-224-jax",
"google/paligemma-3b-ft-nlvr2-448-jax",
"google/paligemma-3b-ft-science-qa-224-jax",
"google/paligemma-3b-ft-infovqa-896",
"google/paligemma-3b-ft-docvqa-448-jax",
"google/paligemma-3b-ft-gqa-448",
"google/paligemma-3b-ft-okvqa-448-jax",
"google/paligemma-3b-ft-textcaps-224",
"google/paligemma-3b-ft-rsvqa-hr-448-jax",
"google/paligemma-3b-ft-tallyqa-448-jax",
"google/paligemma-3b-ft-tallyqa-224",
"google/paligemma-3b-ft-stvqa-448-jax",
"google/paligemma-3b-ft-stvqa-224-jax",
"google/paligemma-3b-ft-ai2d-224",
"google/paligemma-3b-ft-widgetcap-224-jax",
"google/paligemma-3b-ft-aokvqa-da-224-jax",
"google/paligemma-3b-ft-refcoco-seg-224-jax",
"google/paligemma-3b-ft-nlvr2-448",
"google/paligemma-3b-ft-infovqa-448",
"google/paligemma-3b-ft-coco35l-224-jax",
"google/paligemma-3b-ft-scicap-224-jax",
"google/paligemma-3b-ft-aokvqa-da-448",
"google/paligemma-3b-ft-tallyqa-448",
"google/paligemma-3b-ft-cococap-448-jax",
"google/paligemma-3b-ft-stvqa-896",
"google/paligemma-3b-ft-vizwizvqa-224",
"google/paligemma-3b-ft-aokvqa-mc-224-jax",
"google/paligemma-3b-ft-gqa-448-jax",
"google/paligemma-3b-ft-docvqa-224",
"google/paligemma-3b-ft-cococap-224-jax",
"google/paligemma-3b-ft-gqa-224",
"google/paligemma-3b-ft-textvqa-448-jax",
"google/paligemma-3b-ft-aokvqa-mc-448",
"google/paligemma-3b-ft-ai2d-448-jax",
"google/paligemma-3b-ft-coco35l-448-jax",
"google/paligemma-3b-ft-rsvqa-hr-448",
"google/paligemma-3b-ft-refcoco-seg-224",
"google/paligemma-3b-ft-scicap-448",
"google/paligemma-3b-ft-aokvqa-da-224",
"google/paligemma-3b-ft-science-qa-448-jax",
"google/paligemma-3b-ft-gqa-224-jax",
"google/paligemma-3b-ft-infovqa-224",
"google/paligemma-3b-ft-scicap-448-jax",
"google/paligemma-3b-ft-aokvqa-mc-224",
"google/paligemma-3b-ft-stvqa-224",
"google/paligemma-3b-ft-stvqa-448",
"google/paligemma-3b-ft-infovqa-448-jax",
"google/paligemma-3b-ft-textvqa-224-jax",
"google/paligemma-3b-ft-coco35l-448",
"google/paligemma-3b-ft-refcoco-seg-448",
"google/paligemma-3b-ft-aokvqa-mc-448-jax",
"hermanhelf/paligemma",
"google/paligemma-3b-ft-screen2words-448",
"google/paligemma-3b-ft-okvqa-448",
"google/paligemma-3b-ft-rsvqa-lr-224"
] | [
"HuggingFaceM4/POPE_modif"
] | [
"big-vision/paligemma-hf",
"manu/ColPali-demo",
"merve/paligemma-doc",
"merve/paligemma-tracking",
"agentsea/paligemma-waveui",
"Justinrune/LLaMA-Factory",
"Saee/vQA-exploration",
"dwb2023/model_explorer2",
"dwb2023/model_explorer4",
"rynmurdock/Blue_Tigers",
"beingcognitive/Image_to_Music",
"dwb2023/hf_extractor",
"Scharbhen/paligemma-vqa",
"NSTiwari/PaliGemma-ZeroShotDetection-Video",
"kenken999/fastapi_django_main_live",
"LEAHWA/Artificial_Intel_project",
"dwb2023/omniscience",
"triphuong57/paligemma_finetune",
"triphuong57/paligemma_finetune_v2",
"triphuong57/paligemma_ft_v1",
"taufiqdp/paligemma",
"hermanhelf/paligemma-hf",
"gabrielaltay/vlmqa",
"HUANG-Stephanie/cvquest-colpali",
"anthony-chen/Chem-210-Autograder",
"mattraj/curacel-demo-1",
"mattraj/curacel-demo-2"
] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.21.bib | https://aclanthology.org/2023.emnlp-main.21/ | @inproceedings{cao-etal-2023-event,
title = "Event Ontology Completion with Hierarchical Structure Evolution Networks",
author = "Cao, Pengfei and
Hao, Yupu and
Chen, Yubo and
Liu, Kang and
Xu, Jiexin and
Li, Huaijun and
Jiang, Xiaojian and
Zhao, Jun",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.21",
doi = "10.18653/v1/2023.emnlp-main.21",
pages = "306--320",
abstract = "Traditional event detection methods require predefined event schemas. However, manually defining event schemas is expensive and the coverage of schemas is limited. To this end, some works study the event type induction (ETI) task, which discovers new event types via clustering. However, the setting of ETI suffers from two limitations: event types are not linked into the existing hierarchy and have no semantic names. In this paper, we propose a new research task named Event Ontology Completion (EOC), which aims to simultaneously achieve event clustering, hierarchy expansion and type naming. Furthermore, we develop a Hierarchical Structure Evolution Network (HalTon) for this new task. Specifically, we first devise a Neighborhood Contrastive Clustering module to cluster unlabeled event instances. Then, we propose a Hierarchy-Aware Linking module to incorporate the hierarchical information for event expansion. Finally, we generate meaningful names for new types via an In-Context Learning-based Naming module. Extensive experiments indicate that our method achieves the best performance, outperforming the baselines by 8.23{\%}, 8.79{\%} and 8.10{\%} of ARI score on three datasets.",
}
| Traditional event detection methods require predefined event schemas. However, manually defining event schemas is expensive and the coverage of schemas is limited. To this end, some works study the event type induction (ETI) task, which discovers new event types via clustering. However, the setting of ETI suffers from two limitations: event types are not linked into the existing hierarchy and have no semantic names. In this paper, we propose a new research task named Event Ontology Completion (EOC), which aims to simultaneously achieve event clustering, hierarchy expansion and type naming. Furthermore, we develop a Hierarchical Structure Evolution Network (HalTon) for this new task. Specifically, we first devise a Neighborhood Contrastive Clustering module to cluster unlabeled event instances. Then, we propose a Hierarchy-Aware Linking module to incorporate the hierarchical information for event expansion. Finally, we generate meaningful names for new types via an In-Context Learning-based Naming module. Extensive experiments indicate that our method achieves the best performance, outperforming the baselines by 8.23{\%}, 8.79{\%} and 8.10{\%} of ARI score on three datasets. | [
"Cao, Pengfei",
"Hao, Yupu",
"Chen, Yubo",
"Liu, Kang",
"Xu, Jiexin",
"Li, Huaijun",
"Jiang, Xiaojian",
"Zhao, Jun"
] | Event Ontology Completion with Hierarchical Structure Evolution Networks | emnlp-main.21 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.22.bib | https://aclanthology.org/2023.emnlp-main.22/ | @inproceedings{jin-etal-2023-parameter,
title = "Parameter-efficient Tuning for Large Language Model without Calculating Its Gradients",
author = "Jin, Feihu and
Zhang, Jiajun and
Zong, Chengqing",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.22",
doi = "10.18653/v1/2023.emnlp-main.22",
pages = "321--330",
abstract = "Fine-tuning all parameters of large language models (LLMs) requires significant computational resources and is time-consuming. Recent parameter-efficient tuning methods such as Adapter tuning, Prefix tuning, and LoRA allow for updating a small subset of parameters in large language models. However, they can only save approximately 30{\%} of the training memory requirements, due to the problem that gradient computation and backpropagation are still necessary for these methods. This paper proposes a novel parameter-efficient tuning method for LLMs without calculating their gradients. Leveraging the discernible similarities between the parameter-efficient modules of the same task learned by both large and small language models, we put forward a strategy for transferring the parameter-efficient modules, originally derived from small language models to much larger ones. To ensure a smooth and effective adaptation process, we further introduce a Bridge model to guarantee dimensional consistency while also stimulating a dynamic interaction between the models. We demonstrate the effectiveness of our method using the T5 and GPT-2 series of language models on the SuperGLUE benchmark. Our method achieves comparable performance to both fine-tuning and parameter-efficient tuning on large language models without needing gradient-based optimization. Additionally, our method achieves up to 5.7x memory reduction compared to parameter-efficient tuning.",
}
| Fine-tuning all parameters of large language models (LLMs) requires significant computational resources and is time-consuming. Recent parameter-efficient tuning methods such as Adapter tuning, Prefix tuning, and LoRA allow for updating a small subset of parameters in large language models. However, they can only save approximately 30{\%} of the training memory requirements, due to the problem that gradient computation and backpropagation are still necessary for these methods. This paper proposes a novel parameter-efficient tuning method for LLMs without calculating their gradients. Leveraging the discernible similarities between the parameter-efficient modules of the same task learned by both large and small language models, we put forward a strategy for transferring the parameter-efficient modules, originally derived from small language models to much larger ones. To ensure a smooth and effective adaptation process, we further introduce a Bridge model to guarantee dimensional consistency while also stimulating a dynamic interaction between the models. We demonstrate the effectiveness of our method using the T5 and GPT-2 series of language models on the SuperGLUE benchmark. Our method achieves comparable performance to both fine-tuning and parameter-efficient tuning on large language models without needing gradient-based optimization. Additionally, our method achieves up to 5.7x memory reduction compared to parameter-efficient tuning. | [
"Jin, Feihu",
"Zhang, Jiajun",
"Zong, Chengqing"
] | Parameter-efficient Tuning for Large Language Model without Calculating Its Gradients | emnlp-main.22 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.23.bib | https://aclanthology.org/2023.emnlp-main.23/ | @inproceedings{lei-huang-2023-discourse,
title = "Discourse Structures Guided Fine-grained Propaganda Identification",
author = "Lei, Yuanyuan and
Huang, Ruihong",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.23",
doi = "10.18653/v1/2023.emnlp-main.23",
pages = "331--342",
abstract = "Propaganda is a form of deceptive narratives that instigate or mislead the public, usually with a political purpose. In this paper, we aim to identify propaganda in political news at two fine-grained levels: sentence-level and token-level. We observe that propaganda content is more likely to be embedded in sentences that attribute causality or assert contrast to nearby sentences, as well as seen in opinionated evaluation, speculation and discussions of future expectation. Hence, we propose to incorporate both local and global discourse structures for propaganda discovery and construct two teacher models for identifying PDTB-style discourse relations between nearby sentences and common discourse roles of sentences in a news article respectively. We further devise two methods to incorporate the two types of discourse structures for propaganda identification by either using teacher predicted probabilities as additional features or soliciting guidance in a knowledge distillation framework. Experiments on the benchmark dataset demonstrate that leveraging guidance from discourse structures can significantly improve both precision and recall of propaganda content identification.",
}
| Propaganda is a form of deceptive narratives that instigate or mislead the public, usually with a political purpose. In this paper, we aim to identify propaganda in political news at two fine-grained levels: sentence-level and token-level. We observe that propaganda content is more likely to be embedded in sentences that attribute causality or assert contrast to nearby sentences, as well as seen in opinionated evaluation, speculation and discussions of future expectation. Hence, we propose to incorporate both local and global discourse structures for propaganda discovery and construct two teacher models for identifying PDTB-style discourse relations between nearby sentences and common discourse roles of sentences in a news article respectively. We further devise two methods to incorporate the two types of discourse structures for propaganda identification by either using teacher predicted probabilities as additional features or soliciting guidance in a knowledge distillation framework. Experiments on the benchmark dataset demonstrate that leveraging guidance from discourse structures can significantly improve both precision and recall of propaganda content identification. | [
"Lei, Yuanyuan",
"Huang, Ruihong"
] | Discourse Structures Guided Fine-grained Propaganda Identification | emnlp-main.23 | 2310.18544 | [
"https://github.com/yuanyuanlei-nlp/propaganda_emnlp_2023"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.24.bib | https://aclanthology.org/2023.emnlp-main.24/ | @inproceedings{minixhofer-etal-2023-compoundpiece,
title = "{C}ompound{P}iece: Evaluating and Improving Decompounding Performance of Language Models",
author = "Minixhofer, Benjamin and
Pfeiffer, Jonas and
Vuli{\'c}, Ivan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.24",
doi = "10.18653/v1/2023.emnlp-main.24",
pages = "343--359",
abstract = "While many languages possess processes of joining two or more words to create compound words, previous studies have been typically limited only to languages with excessively productive compound formation (e.g., German, Dutch) and there is no public dataset containing compound and non-compound words across a large number of languages. In this work, we systematically study decompounding, the task of splitting compound words into their constituents, at a wide scale. We first address the data gap by introducing a dataset of 255k compound and non-compound words across 56 diverse languages obtained from Wiktionary. We then use this dataset to evaluate an array of Large Language Models (LLMs) on the decompounding task. We find that LLMs perform poorly, especially on words which are tokenized unfavorably by subword tokenization. We thus introduce a novel methodology to train dedicated models for decompounding. The proposed two-stage procedure relies on a fully self-supervised objective in the first stage, while the second, supervised learning stage optionally fine-tunes the model on the annotated Wiktionary data. Our self-supervised models outperform the prior best unsupervised decompounding models by 13.9{\%} accuracy on average. Our fine-tuned models outperform all prior (language-specific) decompounding tools. Furthermore, we use our models to leverage decompounding during the creation of a subword tokenizer, which we refer to as CompoundPiece. CompoundPiece tokenizes compound words more favorably on average, leading to improved performance on decompounding over an otherwise equivalent model using SentencePiece tokenization.",
}
| While many languages possess processes of joining two or more words to create compound words, previous studies have been typically limited only to languages with excessively productive compound formation (e.g., German, Dutch) and there is no public dataset containing compound and non-compound words across a large number of languages. In this work, we systematically study decompounding, the task of splitting compound words into their constituents, at a wide scale. We first address the data gap by introducing a dataset of 255k compound and non-compound words across 56 diverse languages obtained from Wiktionary. We then use this dataset to evaluate an array of Large Language Models (LLMs) on the decompounding task. We find that LLMs perform poorly, especially on words which are tokenized unfavorably by subword tokenization. We thus introduce a novel methodology to train dedicated models for decompounding. The proposed two-stage procedure relies on a fully self-supervised objective in the first stage, while the second, supervised learning stage optionally fine-tunes the model on the annotated Wiktionary data. Our self-supervised models outperform the prior best unsupervised decompounding models by 13.9{\%} accuracy on average. Our fine-tuned models outperform all prior (language-specific) decompounding tools. Furthermore, we use our models to leverage decompounding during the creation of a subword tokenizer, which we refer to as CompoundPiece. CompoundPiece tokenizes compound words more favorably on average, leading to improved performance on decompounding over an otherwise equivalent model using SentencePiece tokenization. | [
"Minixhofer, Benjamin",
"Pfeiffer, Jonas",
"Vuli{\\'c}, Ivan"
] | CompoundPiece: Evaluating and Improving Decompounding Performance of Language Models | emnlp-main.24 | 2305.14214 | [
"https://github.com/bminixhofer/compoundpiece"
] | https://huggingface.co/papers/2305.14214 | 1 | 0 | 0 | 3 | [
"benjamin/compoundpiece",
"benjamin/compoundpiece-stage1"
] | [
"benjamin/compoundpiece"
] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.25.bib | https://aclanthology.org/2023.emnlp-main.25/ | @inproceedings{wang-etal-2023-improving,
title = "Improving Image Captioning via Predicting Structured Concepts",
author = "Wang, Ting and
Chen, Weidong and
Tian, Yuanhe and
Song, Yan and
Mao, Zhendong",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.25",
doi = "10.18653/v1/2023.emnlp-main.25",
pages = "360--370",
abstract = "Having the difficulty of solving the semantic gap between images and texts for the image captioning task, conventional studies in this area paid some attention to treating semantic concepts as a bridge between the two modalities and improved captioning performance accordingly. Although promising results on concept prediction were obtained, the aforementioned studies normally ignore the relationship among concepts, which relies on not only objects in the image, but also word dependencies in the text, so that offers a considerable potential for improving the process of generating good descriptions. In this paper, we propose a structured concept predictor (SCP) to predict concepts and their structures, then we integrate them into captioning, so that enhance the contribution of visual signals in this task via concepts and further use their relations to distinguish cross-modal semantics for better description generation. Particularly, we design weighted graph convolutional networks (W-GCN) to depict concept relations driven by word dependencies, and then learns differentiated contributions from these concepts for following decoding process. Therefore, our approach captures potential relations among concepts and discriminatively learns different concepts, so that effectively facilitates image captioning with inherited information across modalities. Extensive experiments and their results demonstrate the effectiveness of our approach as well as each proposed module in this work.",
}
| Having the difficulty of solving the semantic gap between images and texts for the image captioning task, conventional studies in this area paid some attention to treating semantic concepts as a bridge between the two modalities and improved captioning performance accordingly. Although promising results on concept prediction were obtained, the aforementioned studies normally ignore the relationship among concepts, which relies on not only objects in the image, but also word dependencies in the text, so that offers a considerable potential for improving the process of generating good descriptions. In this paper, we propose a structured concept predictor (SCP) to predict concepts and their structures, then we integrate them into captioning, so that enhance the contribution of visual signals in this task via concepts and further use their relations to distinguish cross-modal semantics for better description generation. Particularly, we design weighted graph convolutional networks (W-GCN) to depict concept relations driven by word dependencies, and then learns differentiated contributions from these concepts for following decoding process. Therefore, our approach captures potential relations among concepts and discriminatively learns different concepts, so that effectively facilitates image captioning with inherited information across modalities. Extensive experiments and their results demonstrate the effectiveness of our approach as well as each proposed module in this work. | [
"Wang, Ting",
"Chen, Weidong",
"Tian, Yuanhe",
"Song, Yan",
"Mao, Zhendong"
] | Improving Image Captioning via Predicting Structured Concepts | emnlp-main.25 | 2311.08223 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.26.bib | https://aclanthology.org/2023.emnlp-main.26/ | @inproceedings{jones-etal-2023-gatitos,
title = "{GATITOS}: Using a New Multilingual Lexicon for Low-resource Machine Translation",
author = "Jones, Alexander and
Caswell, Isaac and
Firat, Orhan and
Saxena, Ishank",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.26",
doi = "10.18653/v1/2023.emnlp-main.26",
pages = "371--405",
abstract = "Modern machine translation models and language models are able to translate without having been trained on parallel data, greatly expanding the set of languages that they can serve. However, these models still struggle in a variety of predictable ways, a problem that cannot be overcome without at least some trusted bilingual data. This work expands on a cheap and abundant resource to combat this problem: bilingual lexica. We test the efficacy of bilingual lexica in a real-world set-up, on 200-language translation models trained on web-crawled text. We present several findings: (1) using lexical data augmentation, we demonstrate sizable performance gains for unsupervised translation; (2) we compare several families of data augmentation, demonstrating that they yield similar improvements, and can be combined for even greater improvements; (3) we demonstrate the importance of carefully curated lexica over larger, noisier ones, especially with larger models; and (4) we compare the efficacy of multilingual lexicon data versus human-translated parallel data. Based on results from (3), we develop and open-source GATITOS, a high-quality, curated dataset in 168 tail languages, one of the first human-translated resources to cover many of these languages.",
}
| Modern machine translation models and language models are able to translate without having been trained on parallel data, greatly expanding the set of languages that they can serve. However, these models still struggle in a variety of predictable ways, a problem that cannot be overcome without at least some trusted bilingual data. This work expands on a cheap and abundant resource to combat this problem: bilingual lexica. We test the efficacy of bilingual lexica in a real-world set-up, on 200-language translation models trained on web-crawled text. We present several findings: (1) using lexical data augmentation, we demonstrate sizable performance gains for unsupervised translation; (2) we compare several families of data augmentation, demonstrating that they yield similar improvements, and can be combined for even greater improvements; (3) we demonstrate the importance of carefully curated lexica over larger, noisier ones, especially with larger models; and (4) we compare the efficacy of multilingual lexicon data versus human-translated parallel data. Based on results from (3), we develop and open-source GATITOS, a high-quality, curated dataset in 168 tail languages, one of the first human-translated resources to cover many of these languages. | [
"Jones, Alex",
"er",
"Caswell, Isaac",
"Firat, Orhan",
"Saxena, Ishank"
] | GATITOS: Using a New Multilingual Lexicon for Low-resource Machine Translation | emnlp-main.26 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.27.bib | https://aclanthology.org/2023.emnlp-main.27/ | @inproceedings{gao-etal-2023-continually,
title = "Continually Improving Extractive {QA} via Human Feedback",
author = "Gao, Ge and
Chen, Hung-Ting and
Artzi, Yoav and
Choi, Eunsol",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.27",
doi = "10.18653/v1/2023.emnlp-main.27",
pages = "406--423",
abstract = "We study continually improving an extractive question answering (QA) system via human user feedback. We design and deploy an iterative approach, where information-seeking users ask questions, receive model-predicted answers, and provide feedback. We conduct experiments involving thousands of user interactions under diverse setups to broaden the understanding of learning from feedback over time. Our experiments show effective improvement from user feedback of extractive QA models over time across different data regimes, including significant potential for domain adaptation.",
}
| We study continually improving an extractive question answering (QA) system via human user feedback. We design and deploy an iterative approach, where information-seeking users ask questions, receive model-predicted answers, and provide feedback. We conduct experiments involving thousands of user interactions under diverse setups to broaden the understanding of learning from feedback over time. Our experiments show effective improvement from user feedback of extractive QA models over time across different data regimes, including significant potential for domain adaptation. | [
"Gao, Ge",
"Chen, Hung-Ting",
"Artzi, Yoav",
"Choi, Eunsol"
] | Continually Improving Extractive QA via Human Feedback | emnlp-main.27 | 2305.12473 | [
"https://github.com/lil-lab/qa-from-hf"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.28.bib | https://aclanthology.org/2023.emnlp-main.28/ | @inproceedings{chen-etal-2023-using,
title = "Using Interpretation Methods for Model Enhancement",
author = "Chen, Zhuo and
Jiang, Chengyue and
Tu, Kewei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.28",
doi = "10.18653/v1/2023.emnlp-main.28",
pages = "424--438",
abstract = "In the age of neural natural language processing, there are plenty of works trying to derive interpretations of neural models. Intuitively, when gold rationales exist during training, one can additionally train the model to match its interpretation with the rationales. However, this intuitive idea has not been fully explored. In this paper, we propose a framework of utilizing interpretation methods and gold rationales to enhance models. Our framework is very general in the sense that it can incorporate various interpretation methods. Previously proposed gradient-based methods can be shown as an instance of our framework. We also propose two novel instances utilizing two other types of interpretation methods, erasure/replace-based and extractor-based methods, for model enhancement. We conduct comprehensive experiments on a variety of tasks. Experimental results show that our framework is effective especially in low-resource settings in enhancing models with various interpretation methods, and our two newly-proposed methods outperform gradient-based methods in most settings. Code is available at https://github.com/Chord-Chen-30/UIMER.",
}
| In the age of neural natural language processing, there are plenty of works trying to derive interpretations of neural models. Intuitively, when gold rationales exist during training, one can additionally train the model to match its interpretation with the rationales. However, this intuitive idea has not been fully explored. In this paper, we propose a framework of utilizing interpretation methods and gold rationales to enhance models. Our framework is very general in the sense that it can incorporate various interpretation methods. Previously proposed gradient-based methods can be shown as an instance of our framework. We also propose two novel instances utilizing two other types of interpretation methods, erasure/replace-based and extractor-based methods, for model enhancement. We conduct comprehensive experiments on a variety of tasks. Experimental results show that our framework is effective especially in low-resource settings in enhancing models with various interpretation methods, and our two newly-proposed methods outperform gradient-based methods in most settings. Code is available at https://github.com/Chord-Chen-30/UIMER. | [
"Chen, Zhuo",
"Jiang, Chengyue",
"Tu, Kewei"
] | Using Interpretation Methods for Model Enhancement | emnlp-main.28 | 2404.02068 | [
"https://github.com/chord-chen-30/uimer"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.29.bib | https://aclanthology.org/2023.emnlp-main.29/ | @inproceedings{zhang-etal-2023-expression,
title = "An Expression Tree Decoding Strategy for Mathematical Equation Generation",
author = "Zhang, Wenqi and
Shen, Yongliang and
Nong, Qingpeng and
Tan, Zeqi and
Ma, Yanna and
Lu, Weiming",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.29",
doi = "10.18653/v1/2023.emnlp-main.29",
pages = "439--456",
abstract = "Generating mathematical equations from natural language requires an accurate understanding of the relations among math expressions. Existing approaches can be broadly categorized into token-level and expression-level generation. The former treats equations as a mathematical language, sequentially generating math tokens. Expression-level methods generate each expression one by one. However, each expression represents a solving step, and there naturally exist parallel or dependent relations between these steps, which are ignored by current sequential methods. Therefore, we integrate tree structure into the expression-level generation and advocate an expression tree decoding strategy. To generate a tree with expression as its node, we employ a layer-wise parallel decoding strategy: we decode multiple independent expressions (leaf nodes) in parallel at each layer and repeat parallel decoding layer by layer to sequentially generate these parent node expressions that depend on others. Besides, a bipartite matching algorithm is adopted to align multiple predictions with annotations for each layer. Experiments show our method outperforms other baselines, especially for these equations with complex structures.",
}
| Generating mathematical equations from natural language requires an accurate understanding of the relations among math expressions. Existing approaches can be broadly categorized into token-level and expression-level generation. The former treats equations as a mathematical language, sequentially generating math tokens. Expression-level methods generate each expression one by one. However, each expression represents a solving step, and there naturally exist parallel or dependent relations between these steps, which are ignored by current sequential methods. Therefore, we integrate tree structure into the expression-level generation and advocate an expression tree decoding strategy. To generate a tree with expression as its node, we employ a layer-wise parallel decoding strategy: we decode multiple independent expressions (leaf nodes) in parallel at each layer and repeat parallel decoding layer by layer to sequentially generate these parent node expressions that depend on others. Besides, a bipartite matching algorithm is adopted to align multiple predictions with annotations for each layer. Experiments show our method outperforms other baselines, especially for these equations with complex structures. | [
"Zhang, Wenqi",
"Shen, Yongliang",
"Nong, Qingpeng",
"Tan, Zeqi",
"Ma, Yanna",
"Lu, Weiming"
] | An Expression Tree Decoding Strategy for Mathematical Equation Generation | emnlp-main.29 | 2310.09619 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.30.bib | https://aclanthology.org/2023.emnlp-main.30/ | @inproceedings{yang-etal-2023-bootstrapping,
title = "Bootstrapping Small {\&} High Performance Language Models with Unmasking-Removal Training Policy",
author = "Yang, Yahan and
Sulem, Elior and
Lee, Insup and
Roth, Dan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.30",
doi = "10.18653/v1/2023.emnlp-main.30",
pages = "457--464",
abstract = "BabyBERTa, a language model trained on small-scale child-directed speech while none of the words are unmasked during training, has been shown to achieve a level of grammaticality comparable to that of RoBERTa-base, which is trained on 6,000 times more words and 15 times more parameters. Relying on this promising result, we explore in this paper the performance of BabyBERTa-based models in downstream tasks, focusing on Semantic Role Labeling (SRL) and two Extractive Question Answering tasks, with the aim of building more efficient systems that rely on less data and smaller models. We investigate the influence of these models both alone and as a starting point to larger pre-trained models, separately examining the contribution of the pre-training data, the vocabulary, and the masking policy on the downstream task performance. Our results show that BabyBERTa trained with unmasking-removal policy is a much stronger starting point for downstream tasks compared to the use of RoBERTa masking policy when 10M words are used for training and that this tendency persists, although to a lesser extent, when adding more training data.",
}
| BabyBERTa, a language model trained on small-scale child-directed speech while none of the words are unmasked during training, has been shown to achieve a level of grammaticality comparable to that of RoBERTa-base, which is trained on 6,000 times more words and 15 times more parameters. Relying on this promising result, we explore in this paper the performance of BabyBERTa-based models in downstream tasks, focusing on Semantic Role Labeling (SRL) and two Extractive Question Answering tasks, with the aim of building more efficient systems that rely on less data and smaller models. We investigate the influence of these models both alone and as a starting point to larger pre-trained models, separately examining the contribution of the pre-training data, the vocabulary, and the masking policy on the downstream task performance. Our results show that BabyBERTa trained with unmasking-removal policy is a much stronger starting point for downstream tasks compared to the use of RoBERTa masking policy when 10M words are used for training and that this tendency persists, although to a lesser extent, when adding more training data. | [
"Yang, Yahan",
"Sulem, Elior",
"Lee, Insup",
"Roth, Dan"
] | Bootstrapping Small & High Performance Language Models with Unmasking-Removal Training Policy | emnlp-main.30 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.31.bib | https://aclanthology.org/2023.emnlp-main.31/ | @inproceedings{yoon-bak-2023-diversity,
title = "Diversity Enhanced Narrative Question Generation for Storybooks",
author = "Yoon, Hokeun and
Bak, JinYeong",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.31",
doi = "10.18653/v1/2023.emnlp-main.31",
pages = "465--482",
abstract = "Question generation (QG) from a given context can enhance comprehension, engagement, assessment, and overall efficacy in learning or conversational environments. Despite recent advancements in QG, the challenge of enhancing or measuring the diversity of generated questions often remains unaddressed. In this paper, we introduce a multi-question generation model (mQG), which is capable of generating multiple, diverse, and answerable questions by focusing on context and questions. To validate the answerability of the generated questions, we employ a SQuAD 2.0 fine-tuned question answering model, classifying the questions as answerable or not. We train and evaluate mQG on the FairytaleQA dataset, a well-structured QA dataset based on storybooks, with narrative questions. We further apply a zero-shot adaptation on the TellMeWhy and SQuAD1.1 datasets. mQG shows promising results across various evaluation metrics, among strong baselines.",
}
| Question generation (QG) from a given context can enhance comprehension, engagement, assessment, and overall efficacy in learning or conversational environments. Despite recent advancements in QG, the challenge of enhancing or measuring the diversity of generated questions often remains unaddressed. In this paper, we introduce a multi-question generation model (mQG), which is capable of generating multiple, diverse, and answerable questions by focusing on context and questions. To validate the answerability of the generated questions, we employ a SQuAD 2.0 fine-tuned question answering model, classifying the questions as answerable or not. We train and evaluate mQG on the FairytaleQA dataset, a well-structured QA dataset based on storybooks, with narrative questions. We further apply a zero-shot adaptation on the TellMeWhy and SQuAD1.1 datasets. mQG shows promising results across various evaluation metrics, among strong baselines. | [
"Yoon, Hokeun",
"Bak, JinYeong"
] | Diversity Enhanced Narrative Question Generation for Storybooks | emnlp-main.31 | 2310.16446 | [
"https://github.com/hkyoon95/mqg"
] | https://huggingface.co/papers/2310.16446 | 0 | 0 | 0 | 2 | [] | [] | [] | 1 | Oral |
https://aclanthology.org/2023.emnlp-main.32.bib | https://aclanthology.org/2023.emnlp-main.32/ | @inproceedings{dong-etal-2023-debiasing,
title = "Debiasing Made State-of-the-art: Revisiting the Simple Seed-based Weak Supervision for Text Classification",
author = "Dong, Chengyu and
Wang, Zihan and
Shang, Jingbo",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.32",
doi = "10.18653/v1/2023.emnlp-main.32",
pages = "483--493",
abstract = "Recent advances in weakly supervised text classification mostly focus on designing sophisticated methods to turn high-level human heuristics into quality pseudo-labels. In this paper, we revisit the seed matching-based method, which is arguably the simplest way to generate pseudo-labels, and show that its power was greatly underestimated. We show that the limited performance of seed matching is largely due to the label bias injected by the simple seed-match rule, which prevents the classifier from learning reliable confidence for selecting high-quality pseudo-labels. Interestingly, simply deleting the seed words present in the matched input texts can mitigate the label bias and help learn better confidence. Subsequently, the performance achieved by seed matching can be improved significantly, making it on par with or even better than the state-of-the-art. Furthermore, to handle the case when the seed words are not made known, we propose to simply delete the word tokens in the input text randomly with a high deletion ratio. Remarkably, seed matching equipped with this random deletion method can often achieve even better performance than that with seed deletion.",
}
| Recent advances in weakly supervised text classification mostly focus on designing sophisticated methods to turn high-level human heuristics into quality pseudo-labels. In this paper, we revisit the seed matching-based method, which is arguably the simplest way to generate pseudo-labels, and show that its power was greatly underestimated. We show that the limited performance of seed matching is largely due to the label bias injected by the simple seed-match rule, which prevents the classifier from learning reliable confidence for selecting high-quality pseudo-labels. Interestingly, simply deleting the seed words present in the matched input texts can mitigate the label bias and help learn better confidence. Subsequently, the performance achieved by seed matching can be improved significantly, making it on par with or even better than the state-of-the-art. Furthermore, to handle the case when the seed words are not made known, we propose to simply delete the word tokens in the input text randomly with a high deletion ratio. Remarkably, seed matching equipped with this random deletion method can often achieve even better performance than that with seed deletion. | [
"Dong, Chengyu",
"Wang, Zihan",
"Shang, Jingbo"
] | Debiasing Made State-of-the-art: Revisiting the Simple Seed-based Weak Supervision for Text Classification | emnlp-main.32 | 2305.14794 | [
"https://github.com/shwinshaker/simseed"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.33.bib | https://aclanthology.org/2023.emnlp-main.33/ | @inproceedings{chen-etal-2023-enhance,
title = "How to Enhance Causal Discrimination of Utterances: A Case on Affective Reasoning",
author = "Chen, Hang and
Yang, Xinyu and
Luo, Jing and
Zhu, Wenjing",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.33",
doi = "10.18653/v1/2023.emnlp-main.33",
pages = "494--512",
abstract = "Our investigation into the Affective Reasoning in Conversation (ARC) task highlights the challenge of causal discrimination. Almost all existing models, including large language models (LLMs), excel at capturing semantic correlations within utterance embeddings but fall short in determining the specific causal relationships. To overcome this limitation, we propose the incorporation of \textit{i.i.d.} noise terms into the conversation process, thereby constructing a structural causal model (SCM). It explores how distinct causal relationships of fitted embeddings can be discerned through independent conditions. To facilitate the implementation of deep learning, we introduce the cogn frameworks to handle unstructured conversation data, and employ an autoencoder architecture to regard the unobservable noise as learnable {``}implicit causes.{''} Moreover, we curate a synthetic dataset that includes i.i.d. noise. Through comprehensive experiments, we validate the effectiveness and interpretability of our approach. Our code is available in https://github.com/Zodiark-ch/mater-of-our-EMNLP2023-paper.",
}
| Our investigation into the Affective Reasoning in Conversation (ARC) task highlights the challenge of causal discrimination. Almost all existing models, including large language models (LLMs), excel at capturing semantic correlations within utterance embeddings but fall short in determining the specific causal relationships. To overcome this limitation, we propose the incorporation of \textit{i.i.d.} noise terms into the conversation process, thereby constructing a structural causal model (SCM). It explores how distinct causal relationships of fitted embeddings can be discerned through independent conditions. To facilitate the implementation of deep learning, we introduce the cogn frameworks to handle unstructured conversation data, and employ an autoencoder architecture to regard the unobservable noise as learnable {``}implicit causes.{''} Moreover, we curate a synthetic dataset that includes i.i.d. noise. Through comprehensive experiments, we validate the effectiveness and interpretability of our approach. Our code is available in https://github.com/Zodiark-ch/mater-of-our-EMNLP2023-paper. | [
"Chen, Hang",
"Yang, Xinyu",
"Luo, Jing",
"Zhu, Wenjing"
] | How to Enhance Causal Discrimination of Utterances: A Case on Affective Reasoning | emnlp-main.33 | 2305.02615 | [
"https://github.com/zodiark-ch/mater-of-our-emnlp2023-paper"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.34.bib | https://aclanthology.org/2023.emnlp-main.34/ | @inproceedings{si-etal-2023-compressing,
title = "Compressing and Debiasing Vision-Language Pre-Trained Models for Visual Question Answering",
author = "Si, Qingyi and
Liu, Yuanxin and
Lin, Zheng and
Fu, Peng and
Cao, Yanan and
Wang, Weiping",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.34",
doi = "10.18653/v1/2023.emnlp-main.34",
pages = "513--529",
abstract = "Despite the excellent performance of vision-language pre-trained models (VLPs) on conventional VQA task, they still suffer from two problems: First, VLPs tend to rely on language biases in datasets and fail to generalize to out-of-distribution (OOD) data. Second, they are inefficient in terms of memory footprint and computation. Although promising progress has been made in both problems, most existing works tackle them independently. To facilitate the application of VLP to VQA tasks, it is imperative to jointly study VLP compression and OOD robustness, which, however, has not yet been explored. This paper investigates whether a VLP can be compressed and debiased simultaneously by searching sparse and robust subnetworks. To this end, we systematically study the design of a training and compression pipeline to search the subnetworks, as well as the assignment of sparsity to different modality-specific modules. Our experiments involve 2 VLPs, 2 compression methods, 4 training methods, 2 datasets and a range of sparsity levels. Our results show that there indeed exist sparse and robust subnetworks, which are competitive with the debiased full VLP and clearly outperform the debiasing SoTAs with fewer parameters on OOD datasets VQA-CP v2 and VQA-VS. The codes can be found at https://github.com/PhoebusSi/Compress-Robust-VQA.",
}
| Despite the excellent performance of vision-language pre-trained models (VLPs) on conventional VQA task, they still suffer from two problems: First, VLPs tend to rely on language biases in datasets and fail to generalize to out-of-distribution (OOD) data. Second, they are inefficient in terms of memory footprint and computation. Although promising progress has been made in both problems, most existing works tackle them independently. To facilitate the application of VLP to VQA tasks, it is imperative to jointly study VLP compression and OOD robustness, which, however, has not yet been explored. This paper investigates whether a VLP can be compressed and debiased simultaneously by searching sparse and robust subnetworks. To this end, we systematically study the design of a training and compression pipeline to search the subnetworks, as well as the assignment of sparsity to different modality-specific modules. Our experiments involve 2 VLPs, 2 compression methods, 4 training methods, 2 datasets and a range of sparsity levels. Our results show that there indeed exist sparse and robust subnetworks, which are competitive with the debiased full VLP and clearly outperform the debiasing SoTAs with fewer parameters on OOD datasets VQA-CP v2 and VQA-VS. The codes can be found at https://github.com/PhoebusSi/Compress-Robust-VQA. | [
"Si, Qingyi",
"Liu, Yuanxin",
"Lin, Zheng",
"Fu, Peng",
"Cao, Yanan",
"Wang, Weiping"
] | Compressing and Debiasing Vision-Language Pre-Trained Models for Visual Question Answering | emnlp-main.34 | 2210.14558 | [
"https://github.com/phoebussi/compress-robust-vqa"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.35.bib | https://aclanthology.org/2023.emnlp-main.35/ | @inproceedings{cole-etal-2023-selectively,
title = "Selectively Answering Ambiguous Questions",
author = "Cole, Jeremy and
Zhang, Michael and
Gillick, Daniel and
Eisenschlos, Julian and
Dhingra, Bhuwan and
Eisenstein, Jacob",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.35",
doi = "10.18653/v1/2023.emnlp-main.35",
pages = "530--543",
abstract = "Trustworthy language models should abstain from answering questions when they do not know the answer. However, the answer to a question can be unknown for a variety of reasons. Prior research has focused on the case in which the question is clear and the answer is unambiguous but possibly unknown. However, the answer to a question can also be unclear due to uncertainty of the questioner{'}s intent or context. We investigate question answering from this perspective, focusing on answering a subset of questions with a high degree of accuracy, from a set of questions in which many are inherently ambiguous. In this setting, we find that the most reliable approach to calibration involves quantifying repetition within a set of sampled model outputs, rather than the model{'}s likelihood or self-verification as used in prior work. We find this to be the case across different types of uncertainty, varying model scales and both with or without instruction tuning. Our results suggest that sampling-based confidence scores help calibrate answers to relatively unambiguous questions, with more dramatic improvements on ambiguous questions.",
}
| Trustworthy language models should abstain from answering questions when they do not know the answer. However, the answer to a question can be unknown for a variety of reasons. Prior research has focused on the case in which the question is clear and the answer is unambiguous but possibly unknown. However, the answer to a question can also be unclear due to uncertainty of the questioner{'}s intent or context. We investigate question answering from this perspective, focusing on answering a subset of questions with a high degree of accuracy, from a set of questions in which many are inherently ambiguous. In this setting, we find that the most reliable approach to calibration involves quantifying repetition within a set of sampled model outputs, rather than the model{'}s likelihood or self-verification as used in prior work. We find this to be the case across different types of uncertainty, varying model scales and both with or without instruction tuning. Our results suggest that sampling-based confidence scores help calibrate answers to relatively unambiguous questions, with more dramatic improvements on ambiguous questions. | [
"Cole, Jeremy",
"Zhang, Michael",
"Gillick, Daniel",
"Eisenschlos, Julian",
"Dhingra, Bhuwan",
"Eisenstein, Jacob"
] | Selectively Answering Ambiguous Questions | emnlp-main.35 | 2305.14613 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.36.bib | https://aclanthology.org/2023.emnlp-main.36/ | @inproceedings{lee-etal-2023-temporal,
title = "Temporal Knowledge Graph Forecasting Without Knowledge Using In-Context Learning",
author = "Lee, Dong-Ho and
Ahrabian, Kian and
Jin, Woojeong and
Morstatter, Fred and
Pujara, Jay",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.36",
doi = "10.18653/v1/2023.emnlp-main.36",
pages = "544--557",
abstract = "Temporal knowledge graph (TKG) forecasting benchmarks challenge models to predict future facts using knowledge of past facts. In this paper, we develop an approach to use in-context learning (ICL) with large language models (LLMs) for TKG forecasting. Our extensive evaluation compares diverse baselines, including both simple heuristics and state-of-the-art (SOTA) supervised models, against pre-trained LLMs across several popular benchmarks and experimental settings. We observe that naive LLMs perform on par with SOTA models, which employ carefully designed architectures and supervised training for the forecasting task, falling within the (-3.6{\%}, +1.5{\%}) Hits@1 margin relative to the median performance. To better understand the strengths of LLMs for forecasting, we explore different approaches for selecting historical facts, constructing prompts, controlling information propagation, and parsing outputs into a probability distribution. A surprising finding from our experiments is that LLM performance endures ($\pm$0.4{\%} Hit@1) even when semantic information is removed by mapping entities/relations to arbitrary numbers, suggesting that prior semantic knowledge is unnecessary; rather, LLMs can leverage the symbolic patterns in the context to achieve such a strong performance. Our analysis also reveals that ICL enables LLMs to learn irregular patterns from the historical context, going beyond frequency and recency biases",
}
| Temporal knowledge graph (TKG) forecasting benchmarks challenge models to predict future facts using knowledge of past facts. In this paper, we develop an approach to use in-context learning (ICL) with large language models (LLMs) for TKG forecasting. Our extensive evaluation compares diverse baselines, including both simple heuristics and state-of-the-art (SOTA) supervised models, against pre-trained LLMs across several popular benchmarks and experimental settings. We observe that naive LLMs perform on par with SOTA models, which employ carefully designed architectures and supervised training for the forecasting task, falling within the (-3.6{\%}, +1.5{\%}) Hits@1 margin relative to the median performance. To better understand the strengths of LLMs for forecasting, we explore different approaches for selecting historical facts, constructing prompts, controlling information propagation, and parsing outputs into a probability distribution. A surprising finding from our experiments is that LLM performance endures ($\pm$0.4{\%} Hit@1) even when semantic information is removed by mapping entities/relations to arbitrary numbers, suggesting that prior semantic knowledge is unnecessary; rather, LLMs can leverage the symbolic patterns in the context to achieve such a strong performance. Our analysis also reveals that ICL enables LLMs to learn irregular patterns from the historical context, going beyond frequency and recency biases | [
"Lee, Dong-Ho",
"Ahrabian, Kian",
"Jin, Woojeong",
"Morstatter, Fred",
"Pujara, Jay"
] | Temporal Knowledge Graph Forecasting Without Knowledge Using In-Context Learning | emnlp-main.36 | 2305.10613 | [
"https://github.com/usc-isi-i2/isi-tkg-icl"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.37.bib | https://aclanthology.org/2023.emnlp-main.37/ | @inproceedings{hwang-etal-2023-knowledge,
title = "Knowledge Graph Compression Enhances Diverse Commonsense Generation",
author = "Hwang, EunJeong and
Thost, Veronika and
Shwartz, Vered and
Ma, Tengfei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.37",
doi = "10.18653/v1/2023.emnlp-main.37",
pages = "558--572",
abstract = "Generating commonsense explanations requires reasoning about commonsense knowledge beyond what is explicitly mentioned in the context. Existing models use commonsense knowledge graphs such as ConceptNet to extract a subgraph of relevant knowledge pertaining to concepts in the input. However, due to the large coverage and, consequently, vast scale of ConceptNet, the extracted subgraphs may contain loosely related, redundant and irrelevant information, which can introduce noise into the model. We propose to address this by applying a differentiable graph compression algorithm that focuses on the relevant knowledge for the task. The compressed subgraphs yield considerably more diverse outputs when incorporated into models for the tasks of generating commonsense and abductive explanations. Moreover, our model achieves better quality-diversity tradeoff than a large language model with 100 times the number of parameters. Our generic approach can be applied to additional NLP tasks that can benefit from incorporating external knowledge.",
}
| Generating commonsense explanations requires reasoning about commonsense knowledge beyond what is explicitly mentioned in the context. Existing models use commonsense knowledge graphs such as ConceptNet to extract a subgraph of relevant knowledge pertaining to concepts in the input. However, due to the large coverage and, consequently, vast scale of ConceptNet, the extracted subgraphs may contain loosely related, redundant and irrelevant information, which can introduce noise into the model. We propose to address this by applying a differentiable graph compression algorithm that focuses on the relevant knowledge for the task. The compressed subgraphs yield considerably more diverse outputs when incorporated into models for the tasks of generating commonsense and abductive explanations. Moreover, our model achieves better quality-diversity tradeoff than a large language model with 100 times the number of parameters. Our generic approach can be applied to additional NLP tasks that can benefit from incorporating external knowledge. | [
"Hwang, EunJeong",
"Thost, Veronika",
"Shwartz, Vered",
"Ma, Tengfei"
] | Knowledge Graph Compression Enhances Diverse Commonsense Generation | emnlp-main.37 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.38.bib | https://aclanthology.org/2023.emnlp-main.38/ | @inproceedings{li-etal-2023-pragmatic,
title = "Pragmatic Reasoning Unlocks Quantifier Semantics for Foundation Models",
author = "Li, Yiyuan and
Menon, Rakesh and
Ghosh, Sayan and
Srivastava, Shashank",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.38",
doi = "10.18653/v1/2023.emnlp-main.38",
pages = "573--591",
abstract = "Generalized quantifiers (e.g., $\textit{few}$, $\textit{most}$) are used to indicate the proportions predicates satisfy (for example, $\textit{some}$ apples are red). One way to interpret quantifier semantics is to explicitly bind these satisfactions with percentage scopes (e.g., 30{\%}-40{\%} of apples are red). This approach can be helpful for tasks like logic formalization and surface-form quantitative reasoning (Gordon and Schubert, 2010; Roy et al., 2015). However, it remains unclear if recent foundation models (Bommasani et al., 2021) possess this ability due to the absence of direct training signals. To explore this, we introduce QuRe, a crowd-sourced dataset of human-annotated generalized quantifiers in Wikipedia sentences featuring percentage-equipped predicates. We explore quantifier comprehension using PRESQUE, a framework that combines natural language inference and the Rational Speech Acts framework. Experimental results on the HVD dataset (Herbelot and Vecchi, 2015) and QuRe demonstrate PRESQUE{'}s superiority over a literal listener baseline, showing a 20{\%} relative improvement in F1 in predicting percentage scopes for quantifiers, even with no additional training.",
}
| Generalized quantifiers (e.g., $\textit{few}$, $\textit{most}$) are used to indicate the proportions predicates satisfy (for example, $\textit{some}$ apples are red). One way to interpret quantifier semantics is to explicitly bind these satisfactions with percentage scopes (e.g., 30{\%}-40{\%} of apples are red). This approach can be helpful for tasks like logic formalization and surface-form quantitative reasoning (Gordon and Schubert, 2010; Roy et al., 2015). However, it remains unclear if recent foundation models (Bommasani et al., 2021) possess this ability due to the absence of direct training signals. To explore this, we introduce QuRe, a crowd-sourced dataset of human-annotated generalized quantifiers in Wikipedia sentences featuring percentage-equipped predicates. We explore quantifier comprehension using PRESQUE, a framework that combines natural language inference and the Rational Speech Acts framework. Experimental results on the HVD dataset (Herbelot and Vecchi, 2015) and QuRe demonstrate PRESQUE{'}s superiority over a literal listener baseline, showing a 20{\%} relative improvement in F1 in predicting percentage scopes for quantifiers, even with no additional training. | [
"Li, Yiyuan",
"Menon, Rakesh",
"Ghosh, Sayan",
"Srivastava, Shashank"
] | Pragmatic Reasoning Unlocks Quantifier Semantics for Foundation Models | emnlp-main.38 | 2311.04659 | [
"https://github.com/nativeatom/presque"
] | https://huggingface.co/papers/2311.04659 | 0 | 0 | 0 | 4 | [] | [
"billli/QuRe"
] | [] | 1 | Oral |
https://aclanthology.org/2023.emnlp-main.39.bib | https://aclanthology.org/2023.emnlp-main.39/ | @inproceedings{liu-etal-2023-llm,
title = "{LLM}-{FP}4: 4-Bit Floating-Point Quantized Transformers",
author = "Liu, Shih-yang and
Liu, Zechun and
Huang, Xijie and
Dong, Pingcheng and
Cheng, Kwang-Ting",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.39",
doi = "10.18653/v1/2023.emnlp-main.39",
pages = "592--605",
abstract = "We propose LLM-FP4 for quantizing both weights and activations in large language models (LLMs) down to 4-bit floating-point values, in a post-training manner. Existing post-training quantization (PTQ) solutions are primarily integer-based and struggle with bit widths below 8 bits. Compared to integer quantization, floating-point (FP) quantization is more flexible and can better handle long-tail or bell-shaped distributions, and it has emerged as a default choice in many hardware platforms. One characteristic of FP quantization is that its performance largely depends on the choice of exponent bits and clipping range. In this regard, we construct a strong FP-PTQ baseline by searching for the optimal quantization parameters. Furthermore, we observe a high inter-channel variance and low intra-channel variance pattern in activation distributions, which adds activation quantization difficulty. We recognize this pattern to be consistent across a spectrum of transformer models designed for diverse tasks such as LLMs, BERT, and Vision Transformer models. To tackle this, we propose per-channel activation quantization and show that these additional scaling factors can be reparameterized as exponential biases of weights, incurring a negligible cost. Our method, for the first time, can quantize both weights and activations in the LLaMA-13B to only 4-bit and achieves an average score of 63.1 on the common sense zero-shot reasoning tasks, which is only 5.8 lower than the full-precision model, significantly outperforming the previous state-of-the-art by 12.7 points. Code is available at: https://github.com/nbasyl/LLM-FP4.",
}
| We propose LLM-FP4 for quantizing both weights and activations in large language models (LLMs) down to 4-bit floating-point values, in a post-training manner. Existing post-training quantization (PTQ) solutions are primarily integer-based and struggle with bit widths below 8 bits. Compared to integer quantization, floating-point (FP) quantization is more flexible and can better handle long-tail or bell-shaped distributions, and it has emerged as a default choice in many hardware platforms. One characteristic of FP quantization is that its performance largely depends on the choice of exponent bits and clipping range. In this regard, we construct a strong FP-PTQ baseline by searching for the optimal quantization parameters. Furthermore, we observe a high inter-channel variance and low intra-channel variance pattern in activation distributions, which adds activation quantization difficulty. We recognize this pattern to be consistent across a spectrum of transformer models designed for diverse tasks such as LLMs, BERT, and Vision Transformer models. To tackle this, we propose per-channel activation quantization and show that these additional scaling factors can be reparameterized as exponential biases of weights, incurring a negligible cost. Our method, for the first time, can quantize both weights and activations in the LLaMA-13B to only 4-bit and achieves an average score of 63.1 on the common sense zero-shot reasoning tasks, which is only 5.8 lower than the full-precision model, significantly outperforming the previous state-of-the-art by 12.7 points. Code is available at: https://github.com/nbasyl/LLM-FP4. | [
"Liu, Shih-yang",
"Liu, Zechun",
"Huang, Xijie",
"Dong, Pingcheng",
"Cheng, Kwang-Ting"
] | LLM-FP4: 4-Bit Floating-Point Quantized Transformers | emnlp-main.39 | 2310.16836 | [
"https://github.com/nbasyl/llm-fp4"
] | https://huggingface.co/papers/2310.16836 | 3 | 13 | 0 | 5 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.40.bib | https://aclanthology.org/2023.emnlp-main.40/ | @inproceedings{tang-etal-2023-improving,
title = "Improving Biomedical Abstractive Summarisation with Knowledge Aggregation from Citation Papers",
author = "Tang, Chen and
Wang, Shun and
Goldsack, Tomas and
Lin, Chenghua",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.40",
doi = "10.18653/v1/2023.emnlp-main.40",
pages = "606--618",
abstract = "Abstracts derived from biomedical literature possess distinct domain-specific characteristics, including specialised writing styles and biomedical terminologies, which necessitate a deep understanding of the related literature. As a result, existing language models struggle to generate technical summaries that are on par with those produced by biomedical experts, given the absence of domain-specific background knowledge. This paper aims to enhance the performance of language models in biomedical abstractive summarisation by aggregating knowledge from external papers cited within the source article. We propose a novel attention-based citation aggregation model that integrates domain-specific knowledge from citation papers, allowing neural networks to generate summaries by leveraging both the paper content and relevant knowledge from citation papers. Furthermore, we construct and release a large-scale biomedical summarisation dataset that serves as a foundation for our research. Extensive experiments demonstrate that our model outperforms state-of-the-art approaches and achieves substantial improvements in abstractive biomedical text summarisation.",
}
| Abstracts derived from biomedical literature possess distinct domain-specific characteristics, including specialised writing styles and biomedical terminologies, which necessitate a deep understanding of the related literature. As a result, existing language models struggle to generate technical summaries that are on par with those produced by biomedical experts, given the absence of domain-specific background knowledge. This paper aims to enhance the performance of language models in biomedical abstractive summarisation by aggregating knowledge from external papers cited within the source article. We propose a novel attention-based citation aggregation model that integrates domain-specific knowledge from citation papers, allowing neural networks to generate summaries by leveraging both the paper content and relevant knowledge from citation papers. Furthermore, we construct and release a large-scale biomedical summarisation dataset that serves as a foundation for our research. Extensive experiments demonstrate that our model outperforms state-of-the-art approaches and achieves substantial improvements in abstractive biomedical text summarisation. | [
"Tang, Chen",
"Wang, Shun",
"Goldsack, Tomas",
"Lin, Chenghua"
] | Improving Biomedical Abstractive Summarisation with Knowledge Aggregation from Citation Papers | emnlp-main.40 | 2310.15684 | [
"https://github.com/tangg555/biomed-sum"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.41.bib | https://aclanthology.org/2023.emnlp-main.41/ | @inproceedings{ye-durrett-2023-explanation,
title = "Explanation Selection Using Unlabeled Data for Chain-of-Thought Prompting",
author = "Ye, Xi and
Durrett, Greg",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.41",
doi = "10.18653/v1/2023.emnlp-main.41",
pages = "619--637",
abstract = "Recent work has shown how to prompt large language models with explanations to obtain strong performance on textual reasoning tasks, i.e., the chain-of-thought paradigm. However, subtly different explanations can yield widely varying downstream task accuracy. Explanations that have not been {``}tuned{''} for a task, such as off-the-shelf explanations written by non-experts, may lead to mediocre performance. This paper tackles the problem of how to optimize explanation-infused prompts in a blackbox fashion. We first generate sets of candidate explanations for each example in the prompt using a leave-one-out scheme, then find an effective combination of these explanations with a two-stage framework. We first evaluate explanations for each in-context example in isolation according to two proxy metrics, log likelihood and accuracy on new examples. Then, we search over combinations of explanations to find one that yields high performance against a silver-labeled development set. Across four textual reasoning tasks spanning question answering, mathematical reasoning, and natural language inference, results show that our proxy metrics correlate with ground truth accuracy and our overall method can effectively improve prompts over crowdworker annotations and naive search strategies",
}
| Recent work has shown how to prompt large language models with explanations to obtain strong performance on textual reasoning tasks, i.e., the chain-of-thought paradigm. However, subtly different explanations can yield widely varying downstream task accuracy. Explanations that have not been {``}tuned{''} for a task, such as off-the-shelf explanations written by non-experts, may lead to mediocre performance. This paper tackles the problem of how to optimize explanation-infused prompts in a blackbox fashion. We first generate sets of candidate explanations for each example in the prompt using a leave-one-out scheme, then find an effective combination of these explanations with a two-stage framework. We first evaluate explanations for each in-context example in isolation according to two proxy metrics, log likelihood and accuracy on new examples. Then, we search over combinations of explanations to find one that yields high performance against a silver-labeled development set. Across four textual reasoning tasks spanning question answering, mathematical reasoning, and natural language inference, results show that our proxy metrics correlate with ground truth accuracy and our overall method can effectively improve prompts over crowdworker annotations and naive search strategies | [
"Ye, Xi",
"Durrett, Greg"
] | Explanation Selection Using Unlabeled Data for Chain-of-Thought Prompting | emnlp-main.41 | 2302.04813 | [
"https://github.com/xiye17/explselection"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.42.bib | https://aclanthology.org/2023.emnlp-main.42/ | @inproceedings{dale-etal-2023-halomi,
title = "{H}al{O}mi: A Manually Annotated Benchmark for Multilingual Hallucination and Omission Detection in Machine Translation",
author = "Dale, David and
Voita, Elena and
Lam, Janice and
Hansanti, Prangthip and
Ropers, Christophe and
Kalbassi, Elahe and
Gao, Cynthia and
Barrault, Loic and
Costa-juss{\`a}, Marta",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.42",
doi = "10.18653/v1/2023.emnlp-main.42",
pages = "638--653",
abstract = "Hallucinations in machine translation are translations that contain information completely unrelated to the input. Omissions are translations that do not include some of the input information. While both cases tend to be catastrophic errors undermining user trust, annotated data with these types of pathologies is extremely scarce and is limited to a few high-resource languages. In this work, we release an annotated dataset for the hallucination and omission phenomena covering 18 translation directions with varying resource levels and scripts. Our annotation covers different levels of partial and full hallucinations as well as omissions both at the sentence and at the word level. Additionally, we revisit previous methods for hallucination and omission detection, show that conclusions made based on a single language pair largely do not hold for a large-scale evaluation, and establish new solid baselines.",
}
| Hallucinations in machine translation are translations that contain information completely unrelated to the input. Omissions are translations that do not include some of the input information. While both cases tend to be catastrophic errors undermining user trust, annotated data with these types of pathologies is extremely scarce and is limited to a few high-resource languages. In this work, we release an annotated dataset for the hallucination and omission phenomena covering 18 translation directions with varying resource levels and scripts. Our annotation covers different levels of partial and full hallucinations as well as omissions both at the sentence and at the word level. Additionally, we revisit previous methods for hallucination and omission detection, show that conclusions made based on a single language pair largely do not hold for a large-scale evaluation, and establish new solid baselines. | [
"Dale, David",
"Voita, Elena",
"Lam, Janice",
"Hansanti, Prangthip",
"Ropers, Christophe",
"Kalbassi, Elahe",
"Gao, Cynthia",
"Barrault, Loic",
"Costa-juss{\\`a}, Marta"
] | HalOmi: A Manually Annotated Benchmark for Multilingual Hallucination and Omission Detection in Machine Translation | emnlp-main.42 | 2305.11746 | [
"https://github.com/facebookresearch/stopes"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.43.bib | https://aclanthology.org/2023.emnlp-main.43/ | @inproceedings{he-etal-2023-gradient,
title = "Gradient-based Gradual Pruning for Language-Specific Multilingual Neural Machine Translation",
author = "He, Dan and
Pham, Minh-Quang and
Ha, Thanh-Le and
Turchi, Marco",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.43",
doi = "10.18653/v1/2023.emnlp-main.43",
pages = "654--670",
abstract = "Multilingual neural machine translation (MNMT) offers the convenience of translating between multiple languages with a single model. However, MNMT often suffers from performance degradation in high-resource languages compared to bilingual counterparts. This degradation is commonly attributed to parameter interference, which occurs when parameters are fully shared across all language pairs. In this work, to tackle this issue we propose a gradient-based gradual pruning technique for MNMT. Our approach aims to identify an optimal sub-network for each language pair within the multilingual model by leveraging gradient-based information as pruning criterion and gradually increasing the pruning ratio as schedule. Our approach allows for partial parameter sharing across language pairs to alleviate interference, and each pair preserves its unique parameters to capture language-specific information. Comprehensive experiments on IWSLT and WMT datasets show that our approach yields a notable performance gain on both datasets.",
}
| Multilingual neural machine translation (MNMT) offers the convenience of translating between multiple languages with a single model. However, MNMT often suffers from performance degradation in high-resource languages compared to bilingual counterparts. This degradation is commonly attributed to parameter interference, which occurs when parameters are fully shared across all language pairs. In this work, to tackle this issue we propose a gradient-based gradual pruning technique for MNMT. Our approach aims to identify an optimal sub-network for each language pair within the multilingual model by leveraging gradient-based information as pruning criterion and gradually increasing the pruning ratio as schedule. Our approach allows for partial parameter sharing across language pairs to alleviate interference, and each pair preserves its unique parameters to capture language-specific information. Comprehensive experiments on IWSLT and WMT datasets show that our approach yields a notable performance gain on both datasets. | [
"He, Dan",
"Pham, Minh-Quang",
"Ha, Thanh-Le",
"Turchi, Marco"
] | Gradient-based Gradual Pruning for Language-Specific Multilingual Neural Machine Translation | emnlp-main.43 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.44.bib | https://aclanthology.org/2023.emnlp-main.44/ | @inproceedings{whitehouse-etal-2023-llm,
title = "{LLM}-powered Data Augmentation for Enhanced Cross-lingual Performance",
author = "Whitehouse, Chenxi and
Choudhury, Monojit and
Aji, Alham Fikri",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.44",
doi = "10.18653/v1/2023.emnlp-main.44",
pages = "671--686",
abstract = "This paper explores the potential of leveraging Large Language Models (LLMs) for data augmentation in multilingual commonsense reasoning datasets where the available training data is extremely limited. To achieve this, we utilise several LLMs, namely Dolly-v2, StableVicuna, ChatGPT, and GPT-4, to augment three datasets: XCOPA, XWinograd, and XStoryCloze. Subsequently, we evaluate the effectiveness of fine-tuning smaller multilingual models, mBERT and XLMR, using the synthesised data. We compare the performance of training with data generated in English and target languages, as well as translated English-generated data, revealing the overall advantages of incorporating data generated by LLMs, e.g. a notable 13.4 accuracy score improvement for the best case. Furthermore, we conduct a human evaluation by asking native speakers to assess the naturalness and logical coherence of the generated examples across different languages. The results of the evaluation indicate that LLMs such as ChatGPT and GPT-4 excel at producing natural and coherent text in most languages, however, they struggle to generate meaningful text in certain languages like Tamil. We also observe that ChatGPT falls short in generating plausible alternatives compared to the original dataset, whereas examples from GPT-4 exhibit competitive logical consistency.",
}
| This paper explores the potential of leveraging Large Language Models (LLMs) for data augmentation in multilingual commonsense reasoning datasets where the available training data is extremely limited. To achieve this, we utilise several LLMs, namely Dolly-v2, StableVicuna, ChatGPT, and GPT-4, to augment three datasets: XCOPA, XWinograd, and XStoryCloze. Subsequently, we evaluate the effectiveness of fine-tuning smaller multilingual models, mBERT and XLMR, using the synthesised data. We compare the performance of training with data generated in English and target languages, as well as translated English-generated data, revealing the overall advantages of incorporating data generated by LLMs, e.g. a notable 13.4 accuracy score improvement for the best case. Furthermore, we conduct a human evaluation by asking native speakers to assess the naturalness and logical coherence of the generated examples across different languages. The results of the evaluation indicate that LLMs such as ChatGPT and GPT-4 excel at producing natural and coherent text in most languages, however, they struggle to generate meaningful text in certain languages like Tamil. We also observe that ChatGPT falls short in generating plausible alternatives compared to the original dataset, whereas examples from GPT-4 exhibit competitive logical consistency. | [
"Whitehouse, Chenxi",
"Choudhury, Monojit",
"Aji, Alham Fikri"
] | LLM-powered Data Augmentation for Enhanced Cross-lingual Performance | emnlp-main.44 | 2305.14288 | [
"https://github.com/mbzuai-nlp/gen-X"
] | https://huggingface.co/papers/2305.14288 | 2 | 0 | 0 | 3 | [] | [
"coref-data/gen_winograd_raw"
] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.45.bib | https://aclanthology.org/2023.emnlp-main.45/ | @inproceedings{wang-etal-2023-prompt-based,
title = "Prompt-based Logical Semantics Enhancement for Implicit Discourse Relation Recognition",
author = "Wang, Chenxu and
Jian, Ping and
Huang, Mu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.45",
doi = "10.18653/v1/2023.emnlp-main.45",
pages = "687--699",
abstract = "Implicit Discourse Relation Recognition (IDRR), which infers discourse relations without the help of explicit connectives, is still a crucial and challenging task for discourse parsing. Recent works tend to exploit the hierarchical structure information from the annotated senses, which demonstrate enhanced discourse relation representations can be obtained by integrating sense hierarchy. Nevertheless, the performance and robustness for IDRR are significantly constrained by the availability of annotated data. Fortunately, there is a wealth of unannotated utterances with explicit connectives, that can be utilized to acquire enriched discourse relation features. In light of such motivation, we propose a $\textbf{P}$rompt-based $\textbf{L}$ogical $\textbf{S}$emantics $\textbf{E}$nhancement (PLSE) method for IDRR. Essentially, our method seamlessly injects knowledge relevant to discourse relation into pre-trained language models through prompt-based connective prediction. Furthermore, considering the prompt-based connective prediction exhibits local dependencies due to the deficiency of masked language model (MLM) in capturing global semantics, we design a novel self-supervised learning objective based on mutual information maximization to derive enhanced representations of logical semantics for IDRR. Experimental results on PDTB 2.0 and CoNLL16 datasets demonstrate that our method achieves outstanding and consistent performance against the current state-of-the-art models.",
}
| Implicit Discourse Relation Recognition (IDRR), which infers discourse relations without the help of explicit connectives, is still a crucial and challenging task for discourse parsing. Recent works tend to exploit the hierarchical structure information from the annotated senses, which demonstrate enhanced discourse relation representations can be obtained by integrating sense hierarchy. Nevertheless, the performance and robustness for IDRR are significantly constrained by the availability of annotated data. Fortunately, there is a wealth of unannotated utterances with explicit connectives, that can be utilized to acquire enriched discourse relation features. In light of such motivation, we propose a $\textbf{P}$rompt-based $\textbf{L}$ogical $\textbf{S}$emantics $\textbf{E}$nhancement (PLSE) method for IDRR. Essentially, our method seamlessly injects knowledge relevant to discourse relation into pre-trained language models through prompt-based connective prediction. Furthermore, considering the prompt-based connective prediction exhibits local dependencies due to the deficiency of masked language model (MLM) in capturing global semantics, we design a novel self-supervised learning objective based on mutual information maximization to derive enhanced representations of logical semantics for IDRR. Experimental results on PDTB 2.0 and CoNLL16 datasets demonstrate that our method achieves outstanding and consistent performance against the current state-of-the-art models. | [
"Wang, Chenxu",
"Jian, Ping",
"Huang, Mu"
] | Prompt-based Logical Semantics Enhancement for Implicit Discourse Relation Recognition | emnlp-main.45 | 2311.00367 | [
"https://github.com/lalalamdbf/plse_idrr"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.46.bib | https://aclanthology.org/2023.emnlp-main.46/ | @inproceedings{chung-yu-2023-vlis,
title = "{VLIS}: Unimodal Language Models Guide Multimodal Language Generation",
author = "Chung, Jiwan and
Yu, Youngjae",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.46",
doi = "10.18653/v1/2023.emnlp-main.46",
pages = "700--721",
abstract = "Multimodal language generation, which leverages the synergy of language and vision, is a rapidly expanding field. However, existing vision-language models face challenges in tasks that require complex linguistic understanding. To address this issue, we introduce Visual-Language models as Importance Sampling weights (VLIS), a novel framework that combines the visual conditioning capability of vision-language models with the language understanding of unimodal text-only language models without further training. It extracts pointwise mutual information of each image and text from a visual-language model and uses the value as an importance sampling weight to adjust the token likelihood from a text-only model. VLIS improves vision-language models on diverse tasks, including commonsense understanding (WHOOPS, OK-VQA, and ScienceQA) and complex text generation (Concadia, Image Paragraph Captioning, and ROCStories). Our results suggest that VLIS represents a promising new direction for multimodal language generation.",
}
| Multimodal language generation, which leverages the synergy of language and vision, is a rapidly expanding field. However, existing vision-language models face challenges in tasks that require complex linguistic understanding. To address this issue, we introduce Visual-Language models as Importance Sampling weights (VLIS), a novel framework that combines the visual conditioning capability of vision-language models with the language understanding of unimodal text-only language models without further training. It extracts pointwise mutual information of each image and text from a visual-language model and uses the value as an importance sampling weight to adjust the token likelihood from a text-only model. VLIS improves vision-language models on diverse tasks, including commonsense understanding (WHOOPS, OK-VQA, and ScienceQA) and complex text generation (Concadia, Image Paragraph Captioning, and ROCStories). Our results suggest that VLIS represents a promising new direction for multimodal language generation. | [
"Chung, Jiwan",
"Yu, Youngjae"
] | VLIS: Unimodal Language Models Guide Multimodal Language Generation | emnlp-main.46 | 2310.09767 | [
"https://github.com/jiwanchung/vlis"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.47.bib | https://aclanthology.org/2023.emnlp-main.47/ | @inproceedings{suresh-etal-2023-conceptual,
title = "Conceptual structure coheres in human cognition but not in large language models",
author = "Suresh, Siddharth and
Mukherjee, Kushin and
Yu, Xizheng and
Huang, Wei-Chun and
Padua, Lisa and
Rogers, Timothy",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.47",
doi = "10.18653/v1/2023.emnlp-main.47",
pages = "722--738",
abstract = "Neural network models of language have long been used as a tool for developing hypotheses about conceptual representation in the mind and brain. For many years, such use involved extracting vector-space representations of words and using distances among these to predict or understand human behavior in various semantic tasks. In contemporary language models, however, it is possible to interrogate the latent structure of conceptual representations using methods nearly identical to those commonly used with human participants. The current work uses three common techniques borrowed from cognitive psychology to estimate and compare lexical-semantic structure in both humans and a well-known large language model, the DaVinci variant of GPT-3. In humans, we show that conceptual structure is robust to differences in culture, language, and method of estimation. Structures estimated from the LLM behavior, while individually fairly consistent with those estimated from human behavior, depend much more upon the particular task used to generate behavior responses{--}responses generated by the very same model in the three tasks yield estimates of conceptual structure that cohere less with one another than do human structure estimates. The results suggest one important way that knowledge inhering in contemporary LLMs can differ from human cognition.",
}
| Neural network models of language have long been used as a tool for developing hypotheses about conceptual representation in the mind and brain. For many years, such use involved extracting vector-space representations of words and using distances among these to predict or understand human behavior in various semantic tasks. In contemporary language models, however, it is possible to interrogate the latent structure of conceptual representations using methods nearly identical to those commonly used with human participants. The current work uses three common techniques borrowed from cognitive psychology to estimate and compare lexical-semantic structure in both humans and a well-known large language model, the DaVinci variant of GPT-3. In humans, we show that conceptual structure is robust to differences in culture, language, and method of estimation. Structures estimated from the LLM behavior, while individually fairly consistent with those estimated from human behavior, depend much more upon the particular task used to generate behavior responses{--}responses generated by the very same model in the three tasks yield estimates of conceptual structure that cohere less with one another than do human structure estimates. The results suggest one important way that knowledge inhering in contemporary LLMs can differ from human cognition. | [
"Suresh, Siddharth",
"Mukherjee, Kushin",
"Yu, Xizheng",
"Huang, Wei-Chun",
"Padua, Lisa",
"Rogers, Timothy"
] | Conceptual structure coheres in human cognition but not in large language models | emnlp-main.47 | 2304.02754 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.48.bib | https://aclanthology.org/2023.emnlp-main.48/ | @inproceedings{feng-etal-2023-towards,
title = "Towards {LLM}-driven Dialogue State Tracking",
author = "Feng, Yujie and
Lu, Zexin and
Liu, Bo and
Zhan, Liming and
Wu, Xiao-Ming",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.48",
doi = "10.18653/v1/2023.emnlp-main.48",
pages = "739--755",
abstract = "Dialogue State Tracking (DST) is of paramount importance in ensuring accurate tracking of user goals and system actions within task-oriented dialogue systems. The emergence of large language models (LLMs) such as GPT3 and ChatGPT has sparked considerable interest in assessing their efficacy across diverse applications. In this study, we conduct an initial examination of ChatGPT{'}s capabilities in DST. Our evaluation uncovers the exceptional performance of ChatGPT in this task, offering valuable insights to researchers regarding its capabilities and providing useful directions for designing and enhancing dialogue systems. Despite its impressive performance, ChatGPT has significant limitations including its closed-source nature, request restrictions, raising data privacy concerns, and lacking local deployment capabilities. To address these concerns, we present LDST, an LLM-driven DST framework based on smaller, open-source foundation models. By utilizing a novel domain-slot instruction tuning method, LDST achieves performance on par with ChatGPT. Comprehensive evaluations across three distinct experimental settings, we find that LDST exhibits remarkable performance improvements in both zero-shot and few-shot setting compared to previous SOTA methods. The source code is provided for reproducibility.",
}
| Dialogue State Tracking (DST) is of paramount importance in ensuring accurate tracking of user goals and system actions within task-oriented dialogue systems. The emergence of large language models (LLMs) such as GPT3 and ChatGPT has sparked considerable interest in assessing their efficacy across diverse applications. In this study, we conduct an initial examination of ChatGPT{'}s capabilities in DST. Our evaluation uncovers the exceptional performance of ChatGPT in this task, offering valuable insights to researchers regarding its capabilities and providing useful directions for designing and enhancing dialogue systems. Despite its impressive performance, ChatGPT has significant limitations including its closed-source nature, request restrictions, raising data privacy concerns, and lacking local deployment capabilities. To address these concerns, we present LDST, an LLM-driven DST framework based on smaller, open-source foundation models. By utilizing a novel domain-slot instruction tuning method, LDST achieves performance on par with ChatGPT. Comprehensive evaluations across three distinct experimental settings, we find that LDST exhibits remarkable performance improvements in both zero-shot and few-shot setting compared to previous SOTA methods. The source code is provided for reproducibility. | [
"Feng, Yujie",
"Lu, Zexin",
"Liu, Bo",
"Zhan, Liming",
"Wu, Xiao-Ming"
] | Towards LLM-driven Dialogue State Tracking | emnlp-main.48 | 2310.14970 | [
"https://github.com/woodscene/ldst"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.49.bib | https://aclanthology.org/2023.emnlp-main.49/ | @inproceedings{zhang-etal-2023-learning-language,
title = "Learning Language-guided Adaptive Hyper-modality Representation for Multimodal Sentiment Analysis",
author = "Zhang, Haoyu and
Wang, Yu and
Yin, Guanghao and
Liu, Kejun and
Liu, Yuanyuan and
Yu, Tianshu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.49",
doi = "10.18653/v1/2023.emnlp-main.49",
pages = "756--767",
abstract = "Though Multimodal Sentiment Analysis (MSA) proves effective by utilizing rich information from multiple sources (*e.g.,* language, video, and audio), the potential sentiment-irrelevant and conflicting information across modalities may hinder the performance from being further improved. To alleviate this, we present Adaptive Language-guided Multimodal Transformer (ALMT), which incorporates an Adaptive Hyper-modality Learning (AHL) module to learn an irrelevance/conflict-suppressing representation from visual and audio features under the guidance of language features at different scales. With the obtained hyper-modality representation, the model can obtain a complementary and joint representation through multimodal fusion for effective MSA. In practice, ALMT achieves state-of-the-art performance on several popular datasets (*e.g.,* MOSI, MOSEI and CH-SIMS) and an abundance of ablation demonstrates the validity and necessity of our irrelevance/conflict suppression mechanism.",
}
| Though Multimodal Sentiment Analysis (MSA) proves effective by utilizing rich information from multiple sources (*e.g.,* language, video, and audio), the potential sentiment-irrelevant and conflicting information across modalities may hinder the performance from being further improved. To alleviate this, we present Adaptive Language-guided Multimodal Transformer (ALMT), which incorporates an Adaptive Hyper-modality Learning (AHL) module to learn an irrelevance/conflict-suppressing representation from visual and audio features under the guidance of language features at different scales. With the obtained hyper-modality representation, the model can obtain a complementary and joint representation through multimodal fusion for effective MSA. In practice, ALMT achieves state-of-the-art performance on several popular datasets (*e.g.,* MOSI, MOSEI and CH-SIMS) and an abundance of ablation demonstrates the validity and necessity of our irrelevance/conflict suppression mechanism. | [
"Zhang, Haoyu",
"Wang, Yu",
"Yin, Guanghao",
"Liu, Kejun",
"Liu, Yuanyuan",
"Yu, Tianshu"
] | Learning Language-guided Adaptive Hyper-modality Representation for Multimodal Sentiment Analysis | emnlp-main.49 | 2310.05804 | [
"https://github.com/Haoyu-ha/ALMT"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.50.bib | https://aclanthology.org/2023.emnlp-main.50/ | @inproceedings{pantazopoulos-etal-2023-multitask,
title = "Multitask Multimodal Prompted Training for Interactive Embodied Task Completion",
author = "Pantazopoulos, Georgios and
Nikandrou, Malvina and
Parekh, Amit and
Hemanthage, Bhathiya and
Eshghi, Arash and
Konstas, Ioannis and
Rieser, Verena and
Lemon, Oliver and
Suglia, Alessandro",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.50",
doi = "10.18653/v1/2023.emnlp-main.50",
pages = "768--789",
abstract = "Interactive and embodied tasks pose at least two fundamental challenges to existing Vision {\&} Language (VL) models, including 1) grounding language in trajectories of actions and observations, and 2) referential disambiguation. To tackle these challenges, we propose an Embodied MultiModal Agent (EMMA): a unified encoder-decoder model that reasons over images and trajectories, and casts action prediction as multimodal text generation. By unifying all tasks as text generation, EMMA learns a language of actions which facilitates transfer across tasks. Different to previous modular approaches with independently trained components, we use a single multitask model where each task contributes to goal completion. EMMA performs on par with similar models on several VL benchmarks and sets a new state-of-the-art performance (36.81{\%} success rate) on the Dialog-guided Task Completion (DTC), a benchmark to evaluate dialog-guided agents in the Alexa Arena.",
}
| Interactive and embodied tasks pose at least two fundamental challenges to existing Vision {\&} Language (VL) models, including 1) grounding language in trajectories of actions and observations, and 2) referential disambiguation. To tackle these challenges, we propose an Embodied MultiModal Agent (EMMA): a unified encoder-decoder model that reasons over images and trajectories, and casts action prediction as multimodal text generation. By unifying all tasks as text generation, EMMA learns a language of actions which facilitates transfer across tasks. Different to previous modular approaches with independently trained components, we use a single multitask model where each task contributes to goal completion. EMMA performs on par with similar models on several VL benchmarks and sets a new state-of-the-art performance (36.81{\%} success rate) on the Dialog-guided Task Completion (DTC), a benchmark to evaluate dialog-guided agents in the Alexa Arena. | [
"Pantazopoulos, Georgios",
"Nik",
"rou, Malvina",
"Parekh, Amit",
"Hemanthage, Bhathiya",
"Eshghi, Arash",
"Konstas, Ioannis",
"Rieser, Verena",
"Lemon, Oliver",
"Suglia, Aless",
"ro"
] | Multitask Multimodal Prompted Training for Interactive Embodied Task Completion | emnlp-main.50 | 2311.04067 | [
""
] | https://huggingface.co/papers/2311.04067 | 1 | 1 | 0 | 9 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.51.bib | https://aclanthology.org/2023.emnlp-main.51/ | @inproceedings{liu-etal-2023-afraid,
title = "We{'}re Afraid Language Models Aren{'}t Modeling Ambiguity",
author = "Liu, Alisa and
Wu, Zhaofeng and
Michael, Julian and
Suhr, Alane and
West, Peter and
Koller, Alexander and
Swayamdipta, Swabha and
Smith, Noah and
Choi, Yejin",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.51",
doi = "10.18653/v1/2023.emnlp-main.51",
pages = "790--807",
abstract = "Ambiguity is an intrinsic feature of natural language. Managing ambiguity is a key part of human language understanding, allowing us to anticipate misunderstanding as communicators and revise our interpretations as listeners. As language models are increasingly employed as dialogue interfaces and writing aids, handling ambiguous language is critical to their success. We capture ambiguity in a sentence through its effect on entailment relations with another sentence, and collect AmbiEnt, a linguist-annotated benchmark of 1,645 examples with diverse kinds of ambiguity. We design a suite of tests based on AmbiEnt, presenting the first evaluation of pretrained LMs to recognize ambiguity and disentangle possible meanings. We find that the task remains extremely challenging, including for GPT-4, whose generated disambiguations are considered correct only 32{\%} of the time in crowdworker evaluation, compared to 90{\%} for disambiguations in our dataset. Finally, to illustrate the value of ambiguity-sensitive tools, we show that a multilabel NLI model can flag political claims in the wild that are misleading due to ambiguity. We encourage the field to rediscover the importance of ambiguity for NLP.",
}
| Ambiguity is an intrinsic feature of natural language. Managing ambiguity is a key part of human language understanding, allowing us to anticipate misunderstanding as communicators and revise our interpretations as listeners. As language models are increasingly employed as dialogue interfaces and writing aids, handling ambiguous language is critical to their success. We capture ambiguity in a sentence through its effect on entailment relations with another sentence, and collect AmbiEnt, a linguist-annotated benchmark of 1,645 examples with diverse kinds of ambiguity. We design a suite of tests based on AmbiEnt, presenting the first evaluation of pretrained LMs to recognize ambiguity and disentangle possible meanings. We find that the task remains extremely challenging, including for GPT-4, whose generated disambiguations are considered correct only 32{\%} of the time in crowdworker evaluation, compared to 90{\%} for disambiguations in our dataset. Finally, to illustrate the value of ambiguity-sensitive tools, we show that a multilabel NLI model can flag political claims in the wild that are misleading due to ambiguity. We encourage the field to rediscover the importance of ambiguity for NLP. | [
"Liu, Alisa",
"Wu, Zhaofeng",
"Michael, Julian",
"Suhr, Alane",
"West, Peter",
"Koller, Alex",
"er",
"Swayamdipta, Swabha",
"Smith, Noah",
"Choi, Yejin"
] | We're Afraid Language Models Aren't Modeling Ambiguity | emnlp-main.51 | 2304.14399 | [
"https://github.com/alisawuffles/ambient"
] | https://huggingface.co/papers/2304.14399 | 1 | 0 | 0 | 9 | [] | [
"metaeval/ambient"
] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.52.bib | https://aclanthology.org/2023.emnlp-main.52/ | @inproceedings{liu-etal-2023-linear,
title = "Linear-Time Modeling of Linguistic Structure: An Order-Theoretic Perspective",
author = "Liu, Tianyu and
Amini, Afra and
Sachan, Mrinmaya and
Cotterell, Ryan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.52",
doi = "10.18653/v1/2023.emnlp-main.52",
pages = "808--830",
abstract = "Tasks that model the relation between pairs of tokens in a string are a vital part of understanding natural language. Such tasks, in general, require exhaustive pair-wise comparisons of tokens, thus having a quadratic runtime complexity in the length of the string. We show that these exhaustive comparisons can be avoided, and, moreover, the complexity of such tasks can be reduced to linear by casting the relation between tokens as a partial order over the string. Our method predicts real numbers for each token in a string in parallel and sorts the tokens accordingly, resulting in total orders of the tokens in the string. Each total order implies a set of arcs oriented from smaller to greater tokens, sorted by their predicted numbers. The intersection of total orders results in a partial order over the set of tokens in the string, which is then decoded into a directed graph representing the desired linguistic structure. Our experiments on dependency parsing and coreference resolution show that our method achieves state-of-the-art or comparable performance. Moreover, the linear complexity and parallelism of our method double the speed of graph-based coreference resolution models, and bring a 10-times speed-up over graph-based dependency parsers.",
}
| Tasks that model the relation between pairs of tokens in a string are a vital part of understanding natural language. Such tasks, in general, require exhaustive pair-wise comparisons of tokens, thus having a quadratic runtime complexity in the length of the string. We show that these exhaustive comparisons can be avoided, and, moreover, the complexity of such tasks can be reduced to linear by casting the relation between tokens as a partial order over the string. Our method predicts real numbers for each token in a string in parallel and sorts the tokens accordingly, resulting in total orders of the tokens in the string. Each total order implies a set of arcs oriented from smaller to greater tokens, sorted by their predicted numbers. The intersection of total orders results in a partial order over the set of tokens in the string, which is then decoded into a directed graph representing the desired linguistic structure. Our experiments on dependency parsing and coreference resolution show that our method achieves state-of-the-art or comparable performance. Moreover, the linear complexity and parallelism of our method double the speed of graph-based coreference resolution models, and bring a 10-times speed-up over graph-based dependency parsers. | [
"Liu, Tianyu",
"Amini, Afra",
"Sachan, Mrinmaya",
"Cotterell, Ryan"
] | Linear-Time Modeling of Linguistic Structure: An Order-Theoretic Perspective | emnlp-main.52 | 2305.15057 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.53.bib | https://aclanthology.org/2023.emnlp-main.53/ | @inproceedings{bao-etal-2023-gemini,
title = "{GEMINI}: Controlling The Sentence-Level Summary Style in Abstractive Text Summarization",
author = "Bao, Guangsheng and
Ou, Zebin and
Zhang, Yue",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.53",
doi = "10.18653/v1/2023.emnlp-main.53",
pages = "831--842",
abstract = "Human experts write summaries using different techniques, including extracting a sentence from the document and rewriting it, or fusing various information from the document to abstract it. These techniques are flexible and thus difficult to be imitated by any single method. To address this issue, we propose an adaptive model, GEMINI, that integrates a rewriter and a generator to mimic the sentence rewriting and abstracting techniques, respectively. GEMINI adaptively chooses to rewrite a specific document sentence or generate a summary sentence from scratch. Experiments demonstrate that our adaptive approach outperforms the pure abstractive and rewriting baselines on three benchmark datasets, achieving the best results on WikiHow. Interestingly, empirical results show that the human summary styles of summary sentences are consistently predictable given their context. We release our code and model at \url{https://github.com/baoguangsheng/gemini}.",
}
| Human experts write summaries using different techniques, including extracting a sentence from the document and rewriting it, or fusing various information from the document to abstract it. These techniques are flexible and thus difficult to be imitated by any single method. To address this issue, we propose an adaptive model, GEMINI, that integrates a rewriter and a generator to mimic the sentence rewriting and abstracting techniques, respectively. GEMINI adaptively chooses to rewrite a specific document sentence or generate a summary sentence from scratch. Experiments demonstrate that our adaptive approach outperforms the pure abstractive and rewriting baselines on three benchmark datasets, achieving the best results on WikiHow. Interestingly, empirical results show that the human summary styles of summary sentences are consistently predictable given their context. We release our code and model at \url{https://github.com/baoguangsheng/gemini}. | [
"Bao, Guangsheng",
"Ou, Zebin",
"Zhang, Yue"
] | GEMINI: Controlling The Sentence-Level Summary Style in Abstractive Text Summarization | emnlp-main.53 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.54.bib | https://aclanthology.org/2023.emnlp-main.54/ | @inproceedings{chen-etal-2023-fidelity,
title = "Fidelity-Enriched Contrastive Search: Reconciling the Faithfulness-Diversity Trade-Off in Text Generation",
author = "Chen, Wei-Lin and
Wu, Cheng-Kuang and
Chen, Hsin-Hsi and
Chen, Chung-Chi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.54",
doi = "10.18653/v1/2023.emnlp-main.54",
pages = "843--851",
abstract = "In this paper, we address the hallucination problem commonly found in natural language generation tasks. Language models often generate fluent and convincing content but can lack consistency with the provided source, resulting in potential inaccuracies. We propose a new decoding method called Fidelity-Enriched Contrastive Search (FECS), which augments the contrastive search framework with context-aware regularization terms. FECS promotes tokens that are semantically similar to the provided source while penalizing repetitiveness in the generated text. We demonstrate its effectiveness across two tasks prone to hallucination: abstractive summarization and dialogue generation. Results show that FECS consistently enhances faithfulness across various language model sizes while maintaining output diversity comparable to well-performing decoding algorithms.",
}
| In this paper, we address the hallucination problem commonly found in natural language generation tasks. Language models often generate fluent and convincing content but can lack consistency with the provided source, resulting in potential inaccuracies. We propose a new decoding method called Fidelity-Enriched Contrastive Search (FECS), which augments the contrastive search framework with context-aware regularization terms. FECS promotes tokens that are semantically similar to the provided source while penalizing repetitiveness in the generated text. We demonstrate its effectiveness across two tasks prone to hallucination: abstractive summarization and dialogue generation. Results show that FECS consistently enhances faithfulness across various language model sizes while maintaining output diversity comparable to well-performing decoding algorithms. | [
"Chen, Wei-Lin",
"Wu, Cheng-Kuang",
"Chen, Hsin-Hsi",
"Chen, Chung-Chi"
] | Fidelity-Enriched Contrastive Search: Reconciling the Faithfulness-Diversity Trade-Off in Text Generation | emnlp-main.54 | 2310.14981 | [
"https://github.com/ntunlplab/fecs"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.55.bib | https://aclanthology.org/2023.emnlp-main.55/ | @inproceedings{moon-etal-2023-analyzing,
title = "Analyzing Norm Violations in Live-Stream Chat",
author = "Moon, Jihyung and
Lee, Dong-Ho and
Cho, Hyundong and
Jin, Woojeong and
Park, Chan and
Kim, Minwoo and
May, Jonathan and
Pujara, Jay and
Park, Sungjoon",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.55",
doi = "10.18653/v1/2023.emnlp-main.55",
pages = "852--868",
abstract = "Toxic language, such as hate speech, can deter users from participating in online communities and enjoying popular platforms. Previous approaches to detecting toxic language and norm violations have been primarily concerned with conversations from online forums and social media, such as Reddit and Twitter. These approaches are less effective when applied to conversations on live-streaming platforms, such as Twitch and YouTube Live, as each comment is only visible for a limited time and lacks a thread structure that establishes its relationship with other comments. In this work, we share the first NLP study dedicated to detecting norm violations in conversations on live-streaming platforms. We define norm violation categories in live-stream chats and annotate 4,583 moderated comments from Twitch. We articulate several facets of live-stream data that differ from other forums, and demonstrate that existing models perform poorly in this setting. By conducting a user study, we identify the informational context humans use in live-stream moderation, and train models leveraging context to identify norm violations. Our results show that appropriate contextual information can boost moderation performance by 35{\%}.",
}
| Toxic language, such as hate speech, can deter users from participating in online communities and enjoying popular platforms. Previous approaches to detecting toxic language and norm violations have been primarily concerned with conversations from online forums and social media, such as Reddit and Twitter. These approaches are less effective when applied to conversations on live-streaming platforms, such as Twitch and YouTube Live, as each comment is only visible for a limited time and lacks a thread structure that establishes its relationship with other comments. In this work, we share the first NLP study dedicated to detecting norm violations in conversations on live-streaming platforms. We define norm violation categories in live-stream chats and annotate 4,583 moderated comments from Twitch. We articulate several facets of live-stream data that differ from other forums, and demonstrate that existing models perform poorly in this setting. By conducting a user study, we identify the informational context humans use in live-stream moderation, and train models leveraging context to identify norm violations. Our results show that appropriate contextual information can boost moderation performance by 35{\%}. | [
"Moon, Jihyung",
"Lee, Dong-Ho",
"Cho, Hyundong",
"Jin, Woojeong",
"Park, Chan",
"Kim, Minwoo",
"May, Jonathan",
"Pujara, Jay",
"Park, Sungjoon"
] | Analyzing Norm Violations in Live-Stream Chat | emnlp-main.55 | 2305.10731 | [
""
] | https://huggingface.co/papers/2305.10731 | 0 | 0 | 0 | 9 | [] | [] | [] | 1 | Oral |
https://aclanthology.org/2023.emnlp-main.56.bib | https://aclanthology.org/2023.emnlp-main.56/ | @inproceedings{singh-etal-2023-coarse,
title = "Coarse-to-Fine Contrastive Learning in Image-Text-Graph Space for Improved Vision-Language Compositionality",
author = "Singh, Harman and
Zhang, Pengchuan and
Wang, Qifan and
Wang, Mengjiao and
Xiong, Wenhan and
Du, Jingfei and
Chen, Yu",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.56",
doi = "10.18653/v1/2023.emnlp-main.56",
pages = "869--893",
abstract = "Contrastively trained vision-language models have achieved remarkable progress in vision and language representation learning. However, recent research has highlighted severe limitations of these models in their ability to perform compositional reasoning over objects, attributes, and relations. Scene graphs have emerged as an effective way to understand images compositionally. These are graph-structured semantic representations of images that contain objects, their attributes, and relations with other objects in a scene. In this work, we consider the scene graph parsed from text as a proxy for the image scene graph and propose a graph decomposition and augmentation framework along with a coarse-to-fine contrastive learning objective between images and text that aligns sentences of various complexities to the same image. We also introduce novel negative mining techniques in the scene graph space for improving attribute binding and relation understanding. Through extensive experiments, we demonstrate the effectiveness of our approach that significantly improves attribute binding, relation understanding, systematic generalization, and productivity on multiple recently proposed benchmarks (For example, improvements up to $\mathbf{18}${\%} for systematic generalization, $\mathbf{16.5}${\%} for relation understanding over a strong baseline), while achieving similar or better performance than CLIP on various general multimodal tasks.",
}
| Contrastively trained vision-language models have achieved remarkable progress in vision and language representation learning. However, recent research has highlighted severe limitations of these models in their ability to perform compositional reasoning over objects, attributes, and relations. Scene graphs have emerged as an effective way to understand images compositionally. These are graph-structured semantic representations of images that contain objects, their attributes, and relations with other objects in a scene. In this work, we consider the scene graph parsed from text as a proxy for the image scene graph and propose a graph decomposition and augmentation framework along with a coarse-to-fine contrastive learning objective between images and text that aligns sentences of various complexities to the same image. We also introduce novel negative mining techniques in the scene graph space for improving attribute binding and relation understanding. Through extensive experiments, we demonstrate the effectiveness of our approach that significantly improves attribute binding, relation understanding, systematic generalization, and productivity on multiple recently proposed benchmarks (For example, improvements up to $\mathbf{18}${\%} for systematic generalization, $\mathbf{16.5}${\%} for relation understanding over a strong baseline), while achieving similar or better performance than CLIP on various general multimodal tasks. | [
"Singh, Harman",
"Zhang, Pengchuan",
"Wang, Qifan",
"Wang, Mengjiao",
"Xiong, Wenhan",
"Du, Jingfei",
"Chen, Yu"
] | Coarse-to-Fine Contrastive Learning in Image-Text-Graph Space for Improved Vision-Language Compositionality | emnlp-main.56 | 2305.13812 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.57.bib | https://aclanthology.org/2023.emnlp-main.57/ | @inproceedings{han-etal-2023-reading,
title = "Reading Books is Great, But Not if You Are Driving! Visually Grounded Reasoning about Defeasible Commonsense Norms",
author = "Han, Seungju and
Kim, Junhyeok and
Hessel, Jack and
Jiang, Liwei and
Chung, Jiwan and
Son, Yejin and
Choi, Yejin and
Yu, Youngjae",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.57",
doi = "10.18653/v1/2023.emnlp-main.57",
pages = "894--914",
abstract = "Commonsense norms are defeasible by context: reading books is usually great, but not when driving a car. While contexts can be explicitly described in language, in embodied scenarios, contexts are often provided visually. This type of visually grounded reasoning about defeasible commonsense norms is generally easy for humans, but (as we show) poses a challenge for machines, as it necessitates both visual understanding and reasoning about commonsense norms. We construct a new multimodal benchmark for studying commonsense norms: NormLens. NormLens consists of 10K human judgments accompanied by free-form explanations covering 2K multimodal situations, and serves as a probe to address two questions: (1) to what extent can models align with average human judgment? and (2) how well can models explain their predicted judgments? We find that state-of-the-art model judgments and explanations are not well-aligned with human annotation. Additionally, we present a simple yet effective approach to better align models with humans by distilling social commonsense knowledge from large language models. The data and code will be released.",
}
| Commonsense norms are defeasible by context: reading books is usually great, but not when driving a car. While contexts can be explicitly described in language, in embodied scenarios, contexts are often provided visually. This type of visually grounded reasoning about defeasible commonsense norms is generally easy for humans, but (as we show) poses a challenge for machines, as it necessitates both visual understanding and reasoning about commonsense norms. We construct a new multimodal benchmark for studying commonsense norms: NormLens. NormLens consists of 10K human judgments accompanied by free-form explanations covering 2K multimodal situations, and serves as a probe to address two questions: (1) to what extent can models align with average human judgment? and (2) how well can models explain their predicted judgments? We find that state-of-the-art model judgments and explanations are not well-aligned with human annotation. Additionally, we present a simple yet effective approach to better align models with humans by distilling social commonsense knowledge from large language models. The data and code will be released. | [
"Han, Seungju",
"Kim, Junhyeok",
"Hessel, Jack",
"Jiang, Liwei",
"Chung, Jiwan",
"Son, Yejin",
"Choi, Yejin",
"Yu, Youngjae"
] | Reading Books is Great, But Not if You Are Driving! Visually Grounded Reasoning about Defeasible Commonsense Norms | emnlp-main.57 | 2310.10418 | [
"https://github.com/wade3han/normlens"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.58.bib | https://aclanthology.org/2023.emnlp-main.58/ | @inproceedings{zhang-etal-2023-enhancing-uncertainty,
title = "Enhancing Uncertainty-Based Hallucination Detection with Stronger Focus",
author = "Zhang, Tianhang and
Qiu, Lin and
Guo, Qipeng and
Deng, Cheng and
Zhang, Yue and
Zhang, Zheng and
Zhou, Chenghu and
Wang, Xinbing and
Fu, Luoyi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.58",
doi = "10.18653/v1/2023.emnlp-main.58",
pages = "915--932",
abstract = "Large Language Models (LLMs) have gained significant popularity for their impressive performance across diverse fields. However, LLMs are prone to hallucinate untruthful or nonsensical outputs that fail to meet user expectations in many real-world applications. Existing works for detecting hallucinations in LLMs either rely on external knowledge for reference retrieval or require sampling multiple responses from the LLM for consistency verification, making these methods costly and inefficient. In this paper, we propose a novel reference-free, uncertainty-based method for detecting hallucinations in LLMs. Our approach imitates human focus in factuality checking from three aspects: 1) focus on the most informative and important keywords in the given text; 2) focus on the unreliable tokens in historical context which may lead to a cascade of hallucinations; and 3) focus on the token properties such as token type and token frequency. Experimental results on relevant datasets demonstrate the effectiveness of our proposed method, which achieves state-of-the-art performance across all the evaluation metrics and eliminates the need for additional information.",
}
| Large Language Models (LLMs) have gained significant popularity for their impressive performance across diverse fields. However, LLMs are prone to hallucinate untruthful or nonsensical outputs that fail to meet user expectations in many real-world applications. Existing works for detecting hallucinations in LLMs either rely on external knowledge for reference retrieval or require sampling multiple responses from the LLM for consistency verification, making these methods costly and inefficient. In this paper, we propose a novel reference-free, uncertainty-based method for detecting hallucinations in LLMs. Our approach imitates human focus in factuality checking from three aspects: 1) focus on the most informative and important keywords in the given text; 2) focus on the unreliable tokens in historical context which may lead to a cascade of hallucinations; and 3) focus on the token properties such as token type and token frequency. Experimental results on relevant datasets demonstrate the effectiveness of our proposed method, which achieves state-of-the-art performance across all the evaluation metrics and eliminates the need for additional information. | [
"Zhang, Tianhang",
"Qiu, Lin",
"Guo, Qipeng",
"Deng, Cheng",
"Zhang, Yue",
"Zhang, Zheng",
"Zhou, Chenghu",
"Wang, Xinbing",
"Fu, Luoyi"
] | Enhancing Uncertainty-Based Hallucination Detection with Stronger Focus | emnlp-main.58 | 2311.13230 | [
"https://github.com/zthang/focus"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.59.bib | https://aclanthology.org/2023.emnlp-main.59/ | @inproceedings{feng-etal-2023-factkb,
title = "{F}act{KB}: Generalizable Factuality Evaluation using Language Models Enhanced with Factual Knowledge",
author = "Feng, Shangbin and
Balachandran, Vidhisha and
Bai, Yuyang and
Tsvetkov, Yulia",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.59",
doi = "10.18653/v1/2023.emnlp-main.59",
pages = "933--952",
abstract = "Evaluating the factual consistency of automatically generated summaries is essential for the progress and adoption of reliable summarization systems. Despite recent advances, existing factuality evaluation models are not robust, being especially prone to entity and relation errors in new domains. We propose FactKB{---}a simple new approach to factuality evaluation that is generalizable across domains, in particular with respect to entities and relations. FactKB is based on language models pretrained using facts extracted from external knowledge bases. We introduce three types of complementary factuality pretraining objectives based on entity-specific facts, facts extracted from auxiliary knowledge about entities, and facts constructed compositionally through knowledge base walks. The resulting factuality evaluation model achieves state-of-the-art performance on two in-domain news summarization benchmarks as well as on three out-of-domain scientific literature datasets. Further analysis of FactKB shows improved ability to detect erroneous entities and relations in summaries and is robust and easily generalizable across domains.",
}
| Evaluating the factual consistency of automatically generated summaries is essential for the progress and adoption of reliable summarization systems. Despite recent advances, existing factuality evaluation models are not robust, being especially prone to entity and relation errors in new domains. We propose FactKB{---}a simple new approach to factuality evaluation that is generalizable across domains, in particular with respect to entities and relations. FactKB is based on language models pretrained using facts extracted from external knowledge bases. We introduce three types of complementary factuality pretraining objectives based on entity-specific facts, facts extracted from auxiliary knowledge about entities, and facts constructed compositionally through knowledge base walks. The resulting factuality evaluation model achieves state-of-the-art performance on two in-domain news summarization benchmarks as well as on three out-of-domain scientific literature datasets. Further analysis of FactKB shows improved ability to detect erroneous entities and relations in summaries and is robust and easily generalizable across domains. | [
"Feng, Shangbin",
"Balach",
"ran, Vidhisha",
"Bai, Yuyang",
"Tsvetkov, Yulia"
] | FactKB: Generalizable Factuality Evaluation using Language Models Enhanced with Factual Knowledge | emnlp-main.59 | 2305.08281 | [
"https://github.com/bunsenfeng/factkb"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.60.bib | https://aclanthology.org/2023.emnlp-main.60/ | @inproceedings{he-etal-2023-mitigating,
title = "Mitigating Backdoor Poisoning Attacks through the Lens of Spurious Correlation",
author = "He, Xuanli and
Xu, Qiongkai and
Wang, Jun and
Rubinstein, Benjamin and
Cohn, Trevor",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.60",
doi = "10.18653/v1/2023.emnlp-main.60",
pages = "953--967",
abstract = "Modern NLP models are often trained over large untrusted datasets, raising the potential for a malicious adversary to compromise model behaviour. For instance, backdoors can be implanted through crafting training instances with a specific textual trigger and a target label. This paper posits that backdoor poisoning attacks exhibit a spurious correlation between simple text features and classification labels, and accordingly, proposes methods for mitigating spurious correlation as means of defence. Our empirical study reveals that the malicious triggers are highly correlated to their target labels; therefore such correlations are extremely distinguishable compared to those scores of benign features, and can be used to filter out potentially problematic instances. Compared with several existing defences, our defence method significantly reduces attack success rates across backdoor attacks, and in the case of insertion-based attacks, our method provides a near-perfect defence.",
}
| Modern NLP models are often trained over large untrusted datasets, raising the potential for a malicious adversary to compromise model behaviour. For instance, backdoors can be implanted through crafting training instances with a specific textual trigger and a target label. This paper posits that backdoor poisoning attacks exhibit a spurious correlation between simple text features and classification labels, and accordingly, proposes methods for mitigating spurious correlation as means of defence. Our empirical study reveals that the malicious triggers are highly correlated to their target labels; therefore such correlations are extremely distinguishable compared to those scores of benign features, and can be used to filter out potentially problematic instances. Compared with several existing defences, our defence method significantly reduces attack success rates across backdoor attacks, and in the case of insertion-based attacks, our method provides a near-perfect defence. | [
"He, Xuanli",
"Xu, Qiongkai",
"Wang, Jun",
"Rubinstein, Benjamin",
"Cohn, Trevor"
] | Mitigating Backdoor Poisoning Attacks through the Lens of Spurious Correlation | emnlp-main.60 | 2305.11596 | [
"https://github.com/xlhex/emnlp2023_z-defence"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.61.bib | https://aclanthology.org/2023.emnlp-main.61/ | @inproceedings{wei-etal-2023-symbol,
title = "Symbol tuning improves in-context learning in language models",
author = "Wei, Jerry and
Hou, Le and
Lampinen, Andrew and
Chen, Xiangning and
Huang, Da and
Tay, Yi and
Chen, Xinyun and
Lu, Yifeng and
Zhou, Denny and
Ma, Tengyu and
Le, Quoc",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.61",
doi = "10.18653/v1/2023.emnlp-main.61",
pages = "968--979",
abstract = "We present symbol tuning - finetuning language models on in-context input-label pairs where natural language labels (e.g., {``}positive/negative sentiment{''}) are replaced with arbitrary symbols (e.g., {``}foo/bar{''}). Symbol tuning leverages the intuition that when a model cannot use instructions or natural language labels to figure out a task, it must instead do so by learning the input-label mappings. We experiment with symbol tuning across PaLM models up to 540B parameters and observe benefits across various settings. First, symbol tuning boosts performance on unseen in-context learning tasks and is much more robust to underspecified prompts, such as those without instructions or without natural language labels. Second, symbol-tuned models are much stronger at algorithmic reasoning tasks, with up to 18.2{\%} better performance on the List Functions benchmark and up to 15.3{\%} better performance on the Simple Turing Concepts benchmark. Finally, symbol-tuned models show large improvements in following flipped-labels presented in-context, meaning that they are more capable of using in-context information to override prior knowledge.",
}
| We present symbol tuning - finetuning language models on in-context input-label pairs where natural language labels (e.g., {``}positive/negative sentiment{''}) are replaced with arbitrary symbols (e.g., {``}foo/bar{''}). Symbol tuning leverages the intuition that when a model cannot use instructions or natural language labels to figure out a task, it must instead do so by learning the input-label mappings. We experiment with symbol tuning across PaLM models up to 540B parameters and observe benefits across various settings. First, symbol tuning boosts performance on unseen in-context learning tasks and is much more robust to underspecified prompts, such as those without instructions or without natural language labels. Second, symbol-tuned models are much stronger at algorithmic reasoning tasks, with up to 18.2{\%} better performance on the List Functions benchmark and up to 15.3{\%} better performance on the Simple Turing Concepts benchmark. Finally, symbol-tuned models show large improvements in following flipped-labels presented in-context, meaning that they are more capable of using in-context information to override prior knowledge. | [
"Wei, Jerry",
"Hou, Le",
"Lampinen, Andrew",
"Chen, Xiangning",
"Huang, Da",
"Tay, Yi",
"Chen, Xinyun",
"Lu, Yifeng",
"Zhou, Denny",
"Ma, Tengyu",
"Le, Quoc"
] | Symbol tuning improves in-context learning in language models | emnlp-main.61 | 2305.08298 | [
""
] | https://huggingface.co/papers/2305.08298 | 4 | 3 | 0 | 11 | [] | [
"tasksource/icl-symbol-tuning-instruct",
"euclaise/symtune_mini"
] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.62.bib | https://aclanthology.org/2023.emnlp-main.62/ | @inproceedings{gauthier-levy-2023-neural,
title = "The neural dynamics of word recognition and integration",
author = "Gauthier, Jon and
Levy, Roger",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.62",
doi = "10.18653/v1/2023.emnlp-main.62",
pages = "980--995",
abstract = "Listeners recognize and integrate words in rapid and noisy everyday speech by combining expectations about upcoming content with incremental sensory evidence. We present a computational model of word recognition which formalizes this perceptual process in Bayesian decision theory. We fit this model to explain scalp EEG signals recorded as subjects passively listened to a fictional story, revealing both the dynamics of the online auditory word recognition process and the neural correlates of the recognition and integration of words. The model reveals distinct neural processing of words depending on whether or not they can be quickly recognized. While all words trigger a neural response characteristic of probabilistic integration {---} voltage modulations predicted by a word{'}s surprisal in context {---} these modulations are amplified for words which require more than roughly 150 ms of input to be recognized. We observe no difference in the latency of these neural responses according to words{'} recognition times. Our results support a two-part model of speech comprehension, combining an eager and rapid process of word recognition with a temporally independent process of word integration. However, we also developed alternative models of the scalp EEG signal not incorporating word recognition dynamics which showed similar performance improvements. We discuss potential future modeling steps which may help to separate these hypotheses.",
}
| Listeners recognize and integrate words in rapid and noisy everyday speech by combining expectations about upcoming content with incremental sensory evidence. We present a computational model of word recognition which formalizes this perceptual process in Bayesian decision theory. We fit this model to explain scalp EEG signals recorded as subjects passively listened to a fictional story, revealing both the dynamics of the online auditory word recognition process and the neural correlates of the recognition and integration of words. The model reveals distinct neural processing of words depending on whether or not they can be quickly recognized. While all words trigger a neural response characteristic of probabilistic integration {---} voltage modulations predicted by a word{'}s surprisal in context {---} these modulations are amplified for words which require more than roughly 150 ms of input to be recognized. We observe no difference in the latency of these neural responses according to words{'} recognition times. Our results support a two-part model of speech comprehension, combining an eager and rapid process of word recognition with a temporally independent process of word integration. However, we also developed alternative models of the scalp EEG signal not incorporating word recognition dynamics which showed similar performance improvements. We discuss potential future modeling steps which may help to separate these hypotheses. | [
"Gauthier, Jon",
"Levy, Roger"
] | The neural dynamics of word recognition and integration | emnlp-main.62 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
||
https://aclanthology.org/2023.emnlp-main.63.bib | https://aclanthology.org/2023.emnlp-main.63/ | @inproceedings{kim-etal-2023-tree,
title = "Tree of Clarifications: Answering Ambiguous Questions with Retrieval-Augmented Large Language Models",
author = "Kim, Gangwoo and
Kim, Sungdong and
Jeon, Byeongguk and
Park, Joonsuk and
Kang, Jaewoo",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.63",
doi = "10.18653/v1/2023.emnlp-main.63",
pages = "996--1009",
abstract = "Questions in open-domain question answering are often ambiguous, allowing multiple interpretations. One approach to handling them is to identify all possible interpretations of the ambiguous question (AQ) and to generate a long-form answer addressing them all, as suggested by Stelmakh et al., (2022). While it provides a comprehensive response without bothering the user for clarification, considering multiple dimensions of ambiguity and gathering corresponding knowledge remains a challenge. To cope with the challenge, we propose a novel framework, Tree of Clarifications (ToC): It recursively constructs a tree of disambiguations for the AQ{---}via few-shot prompting leveraging external knowledge{---}and uses it to generate a long-form answer. ToC outperforms existing baselines on ASQA in a few-shot setup across the metrics, while surpassing fully-supervised baselines trained on the whole training set in terms of Disambig-F1 and Disambig-ROUGE. Code is available at https://github.com/gankim/tree-of-clarifications.",
}
| Questions in open-domain question answering are often ambiguous, allowing multiple interpretations. One approach to handling them is to identify all possible interpretations of the ambiguous question (AQ) and to generate a long-form answer addressing them all, as suggested by Stelmakh et al., (2022). While it provides a comprehensive response without bothering the user for clarification, considering multiple dimensions of ambiguity and gathering corresponding knowledge remains a challenge. To cope with the challenge, we propose a novel framework, Tree of Clarifications (ToC): It recursively constructs a tree of disambiguations for the AQ{---}via few-shot prompting leveraging external knowledge{---}and uses it to generate a long-form answer. ToC outperforms existing baselines on ASQA in a few-shot setup across the metrics, while surpassing fully-supervised baselines trained on the whole training set in terms of Disambig-F1 and Disambig-ROUGE. Code is available at https://github.com/gankim/tree-of-clarifications. | [
"Kim, Gangwoo",
"Kim, Sungdong",
"Jeon, Byeongguk",
"Park, Joonsuk",
"Kang, Jaewoo"
] | Tree of Clarifications: Answering Ambiguous Questions with Retrieval-Augmented Large Language Models | emnlp-main.63 | 2310.14696 | [
"https://github.com/gankim/tree-of-clarifications"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.64.bib | https://aclanthology.org/2023.emnlp-main.64/ | @inproceedings{huang-etal-2023-incorporating,
title = "Incorporating Worker Perspectives into {MT}urk Annotation Practices for {NLP}",
author = "Huang, Olivia and
Fleisig, Eve and
Klein, Dan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.64",
doi = "10.18653/v1/2023.emnlp-main.64",
pages = "1010--1028",
abstract = "Current practices regarding data collection for natural language processing on Amazon Mechanical Turk (MTurk) often rely on a combination of studies on data quality and heuristics shared among NLP researchers. However, without considering the perspectives of MTurk workers, these approaches are susceptible to issues regarding workers{'} rights and poor response quality. We conducted a critical literature review and a survey of MTurk workers aimed at addressing open questions regarding best practices for fair payment, worker privacy, data quality, and considering worker incentives. We found that worker preferences are often at odds with received wisdom among NLP researchers. Surveyed workers preferred reliable, reasonable payments over uncertain, very high payments; reported frequently lying on demographic questions; and expressed frustration at having work rejected with no explanation. We also found that workers view some quality control methods, such as requiring minimum response times or Master{'}s qualifications, as biased and largely ineffective. Based on the survey results, we provide recommendations on how future NLP studies may better account for MTurk workers{'} experiences in order to respect workers{'} rights and improve data quality.",
}
| Current practices regarding data collection for natural language processing on Amazon Mechanical Turk (MTurk) often rely on a combination of studies on data quality and heuristics shared among NLP researchers. However, without considering the perspectives of MTurk workers, these approaches are susceptible to issues regarding workers{'} rights and poor response quality. We conducted a critical literature review and a survey of MTurk workers aimed at addressing open questions regarding best practices for fair payment, worker privacy, data quality, and considering worker incentives. We found that worker preferences are often at odds with received wisdom among NLP researchers. Surveyed workers preferred reliable, reasonable payments over uncertain, very high payments; reported frequently lying on demographic questions; and expressed frustration at having work rejected with no explanation. We also found that workers view some quality control methods, such as requiring minimum response times or Master{'}s qualifications, as biased and largely ineffective. Based on the survey results, we provide recommendations on how future NLP studies may better account for MTurk workers{'} experiences in order to respect workers{'} rights and improve data quality. | [
"Huang, Olivia",
"Fleisig, Eve",
"Klein, Dan"
] | Incorporating Worker Perspectives into MTurk Annotation Practices for NLP | emnlp-main.64 | 2311.02802 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.65.bib | https://aclanthology.org/2023.emnlp-main.65/ | @inproceedings{guo-etal-2023-predict,
title = "Predict the Future from the Past? On the Temporal Data Distribution Shift in Financial Sentiment Classifications",
author = "Guo, Yue and
Hu, Chenxi and
Yang, Yi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.65",
doi = "10.18653/v1/2023.emnlp-main.65",
pages = "1029--1038",
abstract = "Temporal data distribution shift is prevalent in the financial text. How can a financial sentiment analysis system be trained in a volatile market environment that can accurately infer sentiment and be robust to temporal data distribution shifts? In this paper, we conduct an empirical study on the financial sentiment analysis system under temporal data distribution shifts using a real-world financial social media dataset that spans three years. We find that the fine-tuned models suffer from general performance degradation in the presence of temporal distribution shifts. Furthermore, motivated by the unique temporal nature of the financial text, we propose a novel method that combines out-of-distribution detection with time series modeling for temporal financial sentiment analysis. Experimental results show that the proposed method enhances the model{'}s capability to adapt to evolving temporal shifts in a volatile financial market.",
}
| Temporal data distribution shift is prevalent in the financial text. How can a financial sentiment analysis system be trained in a volatile market environment that can accurately infer sentiment and be robust to temporal data distribution shifts? In this paper, we conduct an empirical study on the financial sentiment analysis system under temporal data distribution shifts using a real-world financial social media dataset that spans three years. We find that the fine-tuned models suffer from general performance degradation in the presence of temporal distribution shifts. Furthermore, motivated by the unique temporal nature of the financial text, we propose a novel method that combines out-of-distribution detection with time series modeling for temporal financial sentiment analysis. Experimental results show that the proposed method enhances the model{'}s capability to adapt to evolving temporal shifts in a volatile financial market. | [
"Guo, Yue",
"Hu, Chenxi",
"Yang, Yi"
] | Predict the Future from the Past? On the Temporal Data Distribution Shift in Financial Sentiment Classifications | emnlp-main.65 | 2310.12620 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.66.bib | https://aclanthology.org/2023.emnlp-main.66/ | @inproceedings{xu-etal-2023-look,
title = "Look-back Decoding for Open-Ended Text Generation",
author = "Xu, Nan and
Zhou, Chunting and
Celikyilmaz, Asli and
Ma, Xuezhe",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.66",
doi = "10.18653/v1/2023.emnlp-main.66",
pages = "1039--1050",
abstract = "Given a prefix (context), open-ended generation aims to decode texts that are coherent, which do not abruptly drift from previous topics, and informative, which do not suffer from undesired repetitions. In this paper, we propose Look-back, an improved decoding algorithm that leverages the Kullback{--}Leibler divergence to track the distribution distance between current and historical decoding steps. Thus Look-back can automatically predict potential repetitive phrase and topic drift, and remove tokens that may cause the failure modes, restricting the next token probability distribution within a plausible distance to the history. We perform decoding experiments on document continuation and story generation, and demonstrate that Look-back is able to generate more fluent and coherent text, outperforming other strong decoding methods significantly in both automatic and human evaluations.",
}
| Given a prefix (context), open-ended generation aims to decode texts that are coherent, which do not abruptly drift from previous topics, and informative, which do not suffer from undesired repetitions. In this paper, we propose Look-back, an improved decoding algorithm that leverages the Kullback{--}Leibler divergence to track the distribution distance between current and historical decoding steps. Thus Look-back can automatically predict potential repetitive phrase and topic drift, and remove tokens that may cause the failure modes, restricting the next token probability distribution within a plausible distance to the history. We perform decoding experiments on document continuation and story generation, and demonstrate that Look-back is able to generate more fluent and coherent text, outperforming other strong decoding methods significantly in both automatic and human evaluations. | [
"Xu, Nan",
"Zhou, Chunting",
"Celikyilmaz, Asli",
"Ma, Xuezhe"
] | Look-back Decoding for Open-Ended Text Generation | emnlp-main.66 | 2305.13477 | [
"https://github.com/xunannancy/lookbackdecoding"
] | https://huggingface.co/papers/2305.13477 | 1 | 0 | 0 | 4 | [] | [] | [] | 1 | Oral |
https://aclanthology.org/2023.emnlp-main.67.bib | https://aclanthology.org/2023.emnlp-main.67/ | @inproceedings{huang-etal-2023-large,
title = "Large Language Models Can Self-Improve",
author = "Huang, Jiaxin and
Gu, Shixiang and
Hou, Le and
Wu, Yuexin and
Wang, Xuezhi and
Yu, Hongkun and
Han, Jiawei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.67",
doi = "10.18653/v1/2023.emnlp-main.67",
pages = "1051--1068",
abstract = "Large Language Models (LLMs) have achieved excellent performances in various tasks. However, fine-tuning an LLM requires extensive supervision. Human, on the other hand, may improve their reasoning abilities by self-thinking without external inputs. In this work, we demonstrate that an LLM is also capable of self-improving with only unlabeled datasets. We use a pre-trained LLM to generate {``}high-confidence{''} rationale-augmented answers for unlabeled questions using Chain-of-Though (CoT) prompting and self-consistency, and fine-tune the LLM using those self-generated solutions as target outputs. We show that without any ground truth label, our approach improves the general reasoning ability of a 540B-parameter LLM (74.4{\%}$\rightarrow$82.1{\%} on GSM8K, 90.0{\%}$\rightarrow$94.4{\%} on OpenBookQA, and 63.4{\%}$\rightarrow$67.9{\%} on ANLI-A3) and can also be adapted to extreme low-resource cases where even training questions and CoT prompts are limited. We conduct ablation studies and show that fine-tuning on diverse reasoning paths is critical for self-improvement.",
}
| Large Language Models (LLMs) have achieved excellent performances in various tasks. However, fine-tuning an LLM requires extensive supervision. Human, on the other hand, may improve their reasoning abilities by self-thinking without external inputs. In this work, we demonstrate that an LLM is also capable of self-improving with only unlabeled datasets. We use a pre-trained LLM to generate {``}high-confidence{''} rationale-augmented answers for unlabeled questions using Chain-of-Though (CoT) prompting and self-consistency, and fine-tune the LLM using those self-generated solutions as target outputs. We show that without any ground truth label, our approach improves the general reasoning ability of a 540B-parameter LLM (74.4{\%}$\rightarrow$82.1{\%} on GSM8K, 90.0{\%}$\rightarrow$94.4{\%} on OpenBookQA, and 63.4{\%}$\rightarrow$67.9{\%} on ANLI-A3) and can also be adapted to extreme low-resource cases where even training questions and CoT prompts are limited. We conduct ablation studies and show that fine-tuning on diverse reasoning paths is critical for self-improvement. | [
"Huang, Jiaxin",
"Gu, Shixiang",
"Hou, Le",
"Wu, Yuexin",
"Wang, Xuezhi",
"Yu, Hongkun",
"Han, Jiawei"
] | Large Language Models Can Self-Improve | emnlp-main.67 | 2210.11610 | [
""
] | https://huggingface.co/papers/2405.20309 | 2 | 1 | 0 | 6 | [] | [] | [] | 1 | Oral |
https://aclanthology.org/2023.emnlp-main.68.bib | https://aclanthology.org/2023.emnlp-main.68/ | @inproceedings{wang-etal-2023-codet5,
title = "{C}ode{T}5+: Open Code Large Language Models for Code Understanding and Generation",
author = "Wang, Yue and
Le, Hung and
Gotmare, Akhilesh and
Bui, Nghi and
Li, Junnan and
Hoi, Steven",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.68",
doi = "10.18653/v1/2023.emnlp-main.68",
pages = "1069--1088",
abstract = "Large language models (LLMs) pretrained on vast source code have achieved prominent progress in code intelligence. However, existing code LLMs have two main limitations. First, they often adopt a specific architecture (encoder-only or decoder-only) or rely on a unified encoder-decoder network for different downstream tasks, lacking the flexibility to operate in the optimal architecture for a specific task. Secondly, they often employ a limited set of pretraining objectives which might not be relevant to some tasks and hence result in substantial performance degrade. To address these limitations, we propose {``}CodeT5+{''}, a family of encoder-decoder LLMs for code in which component modules can be flexibly combined to suit a wide range of code tasks. Such flexibility is enabled by our proposed mixture of pretraining objectives, which cover span denoising, contrastive learning, text-code matching, and causal LM pretraining tasks, on both unimodal and bimodal multilingual code corpora. Furthermore, we propose to initialize CodeT5+ with frozen off-the-shelf LLMs without training from scratch to efficiently scale up our models, and explore instruction-tuning to align with natural language instructions. We extensively evaluate CodeT5+ on over 20 code-related benchmarks in different settings, including zero-shot, finetuning, and instruction-tuning. We observe state-of-the-art (SoTA) performance on various code-related tasks, and our instruction-tuned CodeT5+ 16B achieves new SoTA results of 35.0{\%} pass@1 and 54.5{\%} pass@10 on the HumanEval code generation task against other open code LLMs, even surpassing the OpenAI code-cushman-001 model.",
}
| Large language models (LLMs) pretrained on vast source code have achieved prominent progress in code intelligence. However, existing code LLMs have two main limitations. First, they often adopt a specific architecture (encoder-only or decoder-only) or rely on a unified encoder-decoder network for different downstream tasks, lacking the flexibility to operate in the optimal architecture for a specific task. Secondly, they often employ a limited set of pretraining objectives which might not be relevant to some tasks and hence result in substantial performance degrade. To address these limitations, we propose {``}CodeT5+{''}, a family of encoder-decoder LLMs for code in which component modules can be flexibly combined to suit a wide range of code tasks. Such flexibility is enabled by our proposed mixture of pretraining objectives, which cover span denoising, contrastive learning, text-code matching, and causal LM pretraining tasks, on both unimodal and bimodal multilingual code corpora. Furthermore, we propose to initialize CodeT5+ with frozen off-the-shelf LLMs without training from scratch to efficiently scale up our models, and explore instruction-tuning to align with natural language instructions. We extensively evaluate CodeT5+ on over 20 code-related benchmarks in different settings, including zero-shot, finetuning, and instruction-tuning. We observe state-of-the-art (SoTA) performance on various code-related tasks, and our instruction-tuned CodeT5+ 16B achieves new SoTA results of 35.0{\%} pass@1 and 54.5{\%} pass@10 on the HumanEval code generation task against other open code LLMs, even surpassing the OpenAI code-cushman-001 model. | [
"Wang, Yue",
"Le, Hung",
"Gotmare, Akhilesh",
"Bui, Nghi",
"Li, Junnan",
"Hoi, Steven"
] | CodeT5+: Open Code Large Language Models for Code Understanding and Generation | emnlp-main.68 | 2305.07922 | [
"https://github.com/salesforce/codet5"
] | https://huggingface.co/papers/2305.07922 | 3 | 4 | 2 | 6 | [
"Salesforce/codet5p-16b",
"Salesforce/instructcodet5p-16b",
"Salesforce/codet5p-110m-embedding",
"Salesforce/codet5p-2b",
"Salesforce/codet5p-220m",
"Salesforce/codet5p-770m-py",
"Salesforce/codet5p-770m",
"Salesforce/codet5p-6b",
"Salesforce/codet5p-220m-py",
"michaelfeil/ct2fast-codet5p-770m-py",
"michaelfeil/ct2fast-codet5p-770m",
"Salesforce/codet5p-220m-bimodal",
"jncraton/codet5p-220m-py-ct2-int8",
"OpenNMT/codet5p-770m-ct2-int8",
"jncraton/codet5p-770m-py-ct2-int8",
"vodkaslime/test-converter-repo",
"OpenNMT/codet5p-770m-py-ct2-int8",
"burcusu/ct2-codet5p-220m"
] | [] | [
"TIGER-Lab/GenAI-Arena",
"Sharathhebbar24/One-stop-for-Open-source-models",
"ZhangYuhan/3DGen-Arena",
"alKoGolik/codellama-CodeLlama-7b-hf",
"li-qing/FIRE",
"tianleliphoebe/visual-arena",
"Ashmal/MobiLlama",
"jeevavijay10/code-gen",
"alKoGolik/asd",
"K00B404/One-stop-till-you-drop",
"lb1064/Salesforce-codet5p-220m",
"vodkaslime/ctranslate2-converter",
"kasaliyusufoloriegbe/Salesforce-codet5p-220m",
"alexshengzhili/calahealthgpt",
"Bofeee5675/FIRE",
"evelyn-lo/evelyn",
"yuantao-infini-ai/demo_test",
"TseFeng202/Salesforce-instructcodet5p-16b"
] | 1 | Oral |
https://aclanthology.org/2023.emnlp-main.69.bib | https://aclanthology.org/2023.emnlp-main.69/ | @inproceedings{petit-etal-2023-structural,
title = "Structural generalization in {COGS}: Supertagging is (almost) all you need",
author = "Petit, Alban and
Corro, Caio and
Yvon, Fran{\c{c}}ois",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.69",
doi = "10.18653/v1/2023.emnlp-main.69",
pages = "1089--1101",
abstract = "In many Natural Language Processing applications, neural networks have been found to fail to generalize on out-of-distribution examples. In particular, several recent semantic parsing datasets have put forward important limitations of neural networks in cases where compositional generalization is required. In this work, we extend a neural graph-based parsing framework in several ways to alleviate this issue, notably: (1) the introduction of a supertagging step with valency constraints, expressed as an integer linear program; (2) the reduction of the graph prediction problem to the maximum matching problem; (3) the design of an incremental early-stopping training strategy to prevent overfitting. Experimentally, our approach significantly improves results on examples that require structural generalization in the COGS dataset, a known challenging benchmark for compositional generalization. Overall, these results confirm that structural constraints are important for generalization in semantic parsing.",
}
| In many Natural Language Processing applications, neural networks have been found to fail to generalize on out-of-distribution examples. In particular, several recent semantic parsing datasets have put forward important limitations of neural networks in cases where compositional generalization is required. In this work, we extend a neural graph-based parsing framework in several ways to alleviate this issue, notably: (1) the introduction of a supertagging step with valency constraints, expressed as an integer linear program; (2) the reduction of the graph prediction problem to the maximum matching problem; (3) the design of an incremental early-stopping training strategy to prevent overfitting. Experimentally, our approach significantly improves results on examples that require structural generalization in the COGS dataset, a known challenging benchmark for compositional generalization. Overall, these results confirm that structural constraints are important for generalization in semantic parsing. | [
"Petit, Alban",
"Corro, Caio",
"Yvon, Fran{\\c{c}}ois"
] | Structural generalization in COGS: Supertagging is (almost) all you need | emnlp-main.69 | 2310.14124 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.70.bib | https://aclanthology.org/2023.emnlp-main.70/ | @inproceedings{pei-etal-2023-biot5,
title = "{B}io{T}5: Enriching Cross-modal Integration in Biology with Chemical Knowledge and Natural Language Associations",
author = "Pei, Qizhi and
Zhang, Wei and
Zhu, Jinhua and
Wu, Kehan and
Gao, Kaiyuan and
Wu, Lijun and
Xia, Yingce and
Yan, Rui",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.70",
doi = "10.18653/v1/2023.emnlp-main.70",
pages = "1102--1123",
abstract = "Recent advancements in biological research leverage the integration of molecules, proteins, and natural language to enhance drug discovery. However, current models exhibit several limitations, such as the generation of invalid molecular SMILES, underutilization of contextual information, and equal treatment of structured and unstructured knowledge. To address these issues, we propose BioT5, a comprehensive pre-training framework that enriches cross-modal integration in biology with chemical knowledge and natural language associations. BioT5 utilizes SELFIES for 100{\%} robust molecular representations and extracts knowledge from the surrounding context of bio-entities in unstructured biological literature. Furthermore, BioT5 distinguishes between structured and unstructured knowledge, leading to more effective utilization of information. After fine-tuning, BioT5 shows superior performance across a wide range of tasks, demonstrating its strong capability of capturing underlying relations and properties of bio-entities. Our code is available at https://github.com/QizhiPei/BioT5.",
}
| Recent advancements in biological research leverage the integration of molecules, proteins, and natural language to enhance drug discovery. However, current models exhibit several limitations, such as the generation of invalid molecular SMILES, underutilization of contextual information, and equal treatment of structured and unstructured knowledge. To address these issues, we propose BioT5, a comprehensive pre-training framework that enriches cross-modal integration in biology with chemical knowledge and natural language associations. BioT5 utilizes SELFIES for 100{\%} robust molecular representations and extracts knowledge from the surrounding context of bio-entities in unstructured biological literature. Furthermore, BioT5 distinguishes between structured and unstructured knowledge, leading to more effective utilization of information. After fine-tuning, BioT5 shows superior performance across a wide range of tasks, demonstrating its strong capability of capturing underlying relations and properties of bio-entities. Our code is available at https://github.com/QizhiPei/BioT5. | [
"Pei, Qizhi",
"Zhang, Wei",
"Zhu, Jinhua",
"Wu, Kehan",
"Gao, Kaiyuan",
"Wu, Lijun",
"Xia, Yingce",
"Yan, Rui"
] | BioT5: Enriching Cross-modal Integration in Biology with Chemical Knowledge and Natural Language Associations | emnlp-main.70 | 2310.07276 | [
"https://github.com/QizhiPei/BioT5"
] | https://huggingface.co/papers/2310.07276 | 1 | 5 | 0 | 8 | [
"QizhiPei/biot5-base",
"QizhiPei/biot5-base-text2mol",
"QizhiPei/biot5-base-mol2text",
"QizhiPei/biot5-base-peer-solubility",
"QizhiPei/biot5-base-dti-human",
"QizhiPei/biot5-base-dti-biosnap",
"QizhiPei/biot5-base-dti-bindingdb",
"QizhiPei/biot5-base-peer-binloc",
"QizhiPei/biot5-base-peer-human_ppi",
"QizhiPei/biot5-base-peer-yeast_ppi"
] | [
"QizhiPei/BioT5_finetune_dataset"
] | [
"ndhieunguyen/Lang2mol-Diff"
] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.71.bib | https://aclanthology.org/2023.emnlp-main.71/ | @inproceedings{wen-yi-mimno-2023-hyperpolyglot,
title = "Hyperpolyglot {LLM}s: Cross-Lingual Interpretability in Token Embeddings",
author = "Wen-Yi, Andrea W and
Mimno, David",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.71",
doi = "10.18653/v1/2023.emnlp-main.71",
pages = "1124--1131",
abstract = "Cross-lingual transfer learning is an important property of multilingual large language models (LLMs). But how do LLMs represent relationships between languages? Every language model has an input layer that maps tokens to vectors. This ubiquitous layer of language models is often overlooked. We find that similarities between these input embeddings are highly interpretable and that the geometry of these embeddings differs between model families. In one case (XLM-RoBERTa), embeddings encode language: tokens in different writing systems can be linearly separated with an average of 99.2{\%} accuracy. Another family (mT5) represents cross-lingual semantic similarity: the 50 nearest neighbors for any token represent an average of 7.61 writing systems, and are frequently translations. This result is surprising given that there is no explicit parallel cross-lingual training corpora and no explicit incentive for translations in pre-training objectives. Our research opens the door for investigations in 1) The effect of pre-training and model architectures on representations of languages and 2) The applications of cross-lingual representations embedded in language models.",
}
| Cross-lingual transfer learning is an important property of multilingual large language models (LLMs). But how do LLMs represent relationships between languages? Every language model has an input layer that maps tokens to vectors. This ubiquitous layer of language models is often overlooked. We find that similarities between these input embeddings are highly interpretable and that the geometry of these embeddings differs between model families. In one case (XLM-RoBERTa), embeddings encode language: tokens in different writing systems can be linearly separated with an average of 99.2{\%} accuracy. Another family (mT5) represents cross-lingual semantic similarity: the 50 nearest neighbors for any token represent an average of 7.61 writing systems, and are frequently translations. This result is surprising given that there is no explicit parallel cross-lingual training corpora and no explicit incentive for translations in pre-training objectives. Our research opens the door for investigations in 1) The effect of pre-training and model architectures on representations of languages and 2) The applications of cross-lingual representations embedded in language models. | [
"Wen-Yi, Andrea W",
"Mimno, David"
] | Hyperpolyglot LLMs: Cross-Lingual Interpretability in Token Embeddings | emnlp-main.71 | 2311.18034 | [
"https://github.com/andreawwenyi/hyperpolyglot"
] | https://huggingface.co/papers/2311.18034 | 1 | 0 | 0 | 2 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.72.bib | https://aclanthology.org/2023.emnlp-main.72/ | @inproceedings{wang-etal-2023-target,
title = "Target-oriented Proactive Dialogue Systems with Personalization: Problem Formulation and Dataset Curation",
author = "Wang, Jian and
Cheng, Yi and
Lin, Dongding and
Leong, Chak and
Li, Wenjie",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.72",
doi = "10.18653/v1/2023.emnlp-main.72",
pages = "1132--1143",
abstract = "Target-oriented dialogue systems, designed to proactively steer conversations toward predefined targets or accomplish specific system-side goals, are an exciting area in conversational AI. In this work, by formulating a {\textless}dialogue act, topic{\textgreater} pair as the conversation target, we explore a novel problem of personalized target-oriented dialogue by considering personalization during the target accomplishment process. However, there remains an emergent need for high-quality datasets, and building one from scratch requires tremendous human effort. To address this, we propose an automatic dataset curation framework using a role-playing approach. Based on this framework, we construct a large-scale personalized target-oriented dialogue dataset, TopDial, which comprises about 18K multi-turn dialogues. The experimental results show that this dataset is of high quality and could contribute to exploring personalized target-oriented dialogue.",
}
| Target-oriented dialogue systems, designed to proactively steer conversations toward predefined targets or accomplish specific system-side goals, are an exciting area in conversational AI. In this work, by formulating a {\textless}dialogue act, topic{\textgreater} pair as the conversation target, we explore a novel problem of personalized target-oriented dialogue by considering personalization during the target accomplishment process. However, there remains an emergent need for high-quality datasets, and building one from scratch requires tremendous human effort. To address this, we propose an automatic dataset curation framework using a role-playing approach. Based on this framework, we construct a large-scale personalized target-oriented dialogue dataset, TopDial, which comprises about 18K multi-turn dialogues. The experimental results show that this dataset is of high quality and could contribute to exploring personalized target-oriented dialogue. | [
"Wang, Jian",
"Cheng, Yi",
"Lin, Dongding",
"Leong, Chak",
"Li, Wenjie"
] | Target-oriented Proactive Dialogue Systems with Personalization: Problem Formulation and Dataset Curation | emnlp-main.72 | 2310.07397 | [
"https://github.com/iwangjian/topdial"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.73.bib | https://aclanthology.org/2023.emnlp-main.73/ | @inproceedings{wang-etal-2023-seqxgpt,
title = "{S}eq{XGPT}: Sentence-Level {AI}-Generated Text Detection",
author = "Wang, Pengyu and
Li, Linyang and
Ren, Ke and
Jiang, Botian and
Zhang, Dong and
Qiu, Xipeng",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.73",
pages = "1144--1156",
abstract = "Widely applied large language models (LLMs) can generate human-like content, raising concerns about the abuse of LLMs. Therefore, it is important to build strong AI-generated text (AIGT) detectors. Current works only consider document-level AIGT detection, therefore, in this paper, we first introduce a sentence-level detection challenge by synthesizing a dataset that contains documents that are polished with LLMs, that is, the documents contain sentences written by humans and sentences modified by LLMs. Then we propose \textbf{Seq}uence \textbf{X} (Check) \textbf{GPT}, a novel method that utilizes log probability lists from white-box LLMs as features for sentence-level AIGT detection. These features are composed like \textit{waves} in speech processing and cannot be studied by LLMs. Therefore, we build SeqXGPT based on convolution and self-attention networks. We test it in both sentence and document-level detection challenges. Experimental results show that previous methods struggle in solving sentence-level AIGT detection, while our method not only significantly surpasses baseline methods in both sentence and document-level detection challenges but also exhibits strong generalization capabilities.",
}
| Widely applied large language models (LLMs) can generate human-like content, raising concerns about the abuse of LLMs. Therefore, it is important to build strong AI-generated text (AIGT) detectors. Current works only consider document-level AIGT detection, therefore, in this paper, we first introduce a sentence-level detection challenge by synthesizing a dataset that contains documents that are polished with LLMs, that is, the documents contain sentences written by humans and sentences modified by LLMs. Then we propose \textbf{Seq}uence \textbf{X} (Check) \textbf{GPT}, a novel method that utilizes log probability lists from white-box LLMs as features for sentence-level AIGT detection. These features are composed like \textit{waves} in speech processing and cannot be studied by LLMs. Therefore, we build SeqXGPT based on convolution and self-attention networks. We test it in both sentence and document-level detection challenges. Experimental results show that previous methods struggle in solving sentence-level AIGT detection, while our method not only significantly surpasses baseline methods in both sentence and document-level detection challenges but also exhibits strong generalization capabilities. | [
"Wang, Pengyu",
"Li, Linyang",
"Ren, Ke",
"Jiang, Botian",
"Zhang, Dong",
"Qiu, Xipeng"
] | SeqXGPT: Sentence-Level AI-Generated Text Detection | emnlp-main.73 | 2310.08903 | [
"https://github.com/jihuai-wpy/seqxgpt"
] | https://huggingface.co/papers/2310.08903 | 2 | 1 | 0 | 6 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.74.bib | https://aclanthology.org/2023.emnlp-main.74/ | @inproceedings{zhao-etal-2023-qtsumm,
title = "{QTS}umm: Query-Focused Summarization over Tabular Data",
author = "Zhao, Yilun and
Qi, Zhenting and
Nan, Linyong and
Mi, Boyu and
Liu, Yixin and
Zou, Weijin and
Han, Simeng and
Chen, Ruizhe and
Tang, Xiangru and
Xu, Yumo and
Radev, Dragomir and
Cohan, Arman",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.74",
doi = "10.18653/v1/2023.emnlp-main.74",
pages = "1157--1172",
abstract = "People primarily consult tables to conduct data analysis or answer specific questions. Text generation systems that can provide accurate table summaries tailored to users{'} information needs can facilitate more efficient access to relevant data insights. Motivated by this, we define a new query-focused table summarization task, where text generation models have to perform human-like reasoning and analysis over the given table to generate a tailored summary. We introduce a new benchmark named QTSumm for this task, which contains 7,111 human-annotated query-summary pairs over 2,934 tables covering diverse topics. We investigate a set of strong baselines on QTSumm, including text generation, table-to-text generation, and large language models. Experimental results and manual analysis reveal that the new task presents significant challenges in table-to-text generation for future research. Moreover, we propose a new approach named ReFactor, to retrieve and reason over query-relevant information from tabular data to generate several natural language facts. Experimental results demonstrate that ReFactor can bring effective improvements to baselines by concatenating the generated facts to the model input. Our data and code are publicly available at https://github.com/yale-nlp/QTSumm.",
}
| People primarily consult tables to conduct data analysis or answer specific questions. Text generation systems that can provide accurate table summaries tailored to users{'} information needs can facilitate more efficient access to relevant data insights. Motivated by this, we define a new query-focused table summarization task, where text generation models have to perform human-like reasoning and analysis over the given table to generate a tailored summary. We introduce a new benchmark named QTSumm for this task, which contains 7,111 human-annotated query-summary pairs over 2,934 tables covering diverse topics. We investigate a set of strong baselines on QTSumm, including text generation, table-to-text generation, and large language models. Experimental results and manual analysis reveal that the new task presents significant challenges in table-to-text generation for future research. Moreover, we propose a new approach named ReFactor, to retrieve and reason over query-relevant information from tabular data to generate several natural language facts. Experimental results demonstrate that ReFactor can bring effective improvements to baselines by concatenating the generated facts to the model input. Our data and code are publicly available at https://github.com/yale-nlp/QTSumm. | [
"Zhao, Yilun",
"Qi, Zhenting",
"Nan, Linyong",
"Mi, Boyu",
"Liu, Yixin",
"Zou, Weijin",
"Han, Simeng",
"Chen, Ruizhe",
"Tang, Xiangru",
"Xu, Yumo",
"Radev, Dragomir",
"Cohan, Arman"
] | QTSumm: Query-Focused Summarization over Tabular Data | emnlp-main.74 | 2305.14303 | [
"https://github.com/yilunzhao/qtsumm"
] | https://huggingface.co/papers/2305.14303 | 4 | 0 | 0 | 11 | [
"yale-nlp/bart-large-finetuned-qtsumm",
"yale-nlp/flan-t5-large-finetuned-qtsumm",
"yale-nlp/t5-large-finetuned-qtsumm",
"yale-nlp/omnitab-large-finetuned-qtsumm",
"yale-nlp/tapex-large-finetuned-qtsumm",
"yale-nlp/reastap-large-finetuned-qtsumm"
] | [
"yale-nlp/QTSumm",
"faizalbs777/research"
] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.75.bib | https://aclanthology.org/2023.emnlp-main.75/ | @inproceedings{ge-etal-2023-wrong,
title = "From Wrong To Right: A Recursive Approach Towards Vision-Language Explanation",
author = "Ge, Jiaxin and
Subramanian, Sanjay and
Darrell, Trevor and
Li, Boyi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.75",
doi = "10.18653/v1/2023.emnlp-main.75",
pages = "1173--1185",
abstract = "Addressing the challenge of adapting pre-trained vision-language models for generating insightful explanations for visual reasoning tasks with limited annotations, we present ReVisE: a Recursive Visual Explanation algorithm. Our method iteratively computes visual features (conditioned on the text input), an answer, and an explanation, to improve the explanation quality step by step until the answer converges. We find that this multi-step approach guides the model to correct its own answers and outperforms single-step explanation generation. Furthermore, explanations generated by ReVisE also serve as valuable annotations for few-shot self-training. Our approach outperforms previous methods while utilizing merely 5{\%} of the human-annotated explanations across 10 metrics, demonstrating up to a 4.2 and 1.3 increase in BLEU-1 score on the VCR and VQA-X datasets, underscoring the efficacy and data-efficiency of our method.",
}
| Addressing the challenge of adapting pre-trained vision-language models for generating insightful explanations for visual reasoning tasks with limited annotations, we present ReVisE: a Recursive Visual Explanation algorithm. Our method iteratively computes visual features (conditioned on the text input), an answer, and an explanation, to improve the explanation quality step by step until the answer converges. We find that this multi-step approach guides the model to correct its own answers and outperforms single-step explanation generation. Furthermore, explanations generated by ReVisE also serve as valuable annotations for few-shot self-training. Our approach outperforms previous methods while utilizing merely 5{\%} of the human-annotated explanations across 10 metrics, demonstrating up to a 4.2 and 1.3 increase in BLEU-1 score on the VCR and VQA-X datasets, underscoring the efficacy and data-efficiency of our method. | [
"Ge, Jiaxin",
"Subramanian, Sanjay",
"Darrell, Trevor",
"Li, Boyi"
] | From Wrong To Right: A Recursive Approach Towards Vision-Language Explanation | emnlp-main.75 | 2311.12391 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.76.bib | https://aclanthology.org/2023.emnlp-main.76/ | @inproceedings{cardenas-etal-2023-dont,
title = "{`}Don{'}t Get Too Technical with Me{'}: A Discourse Structure-Based Framework for Automatic Science Journalism",
author = "Cardenas, Ronald and
Yao, Bingsheng and
Wang, Dakuo and
Hou, Yufang",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.76",
doi = "10.18653/v1/2023.emnlp-main.76",
pages = "1186--1202",
abstract = "Science journalism refers to the task of reporting technical findings of a scientific paper as a less technical news article to the general public audience. We aim to design an automated system to support this real-world task (i.e., automatic science journalism ) by 1) introducing a newly-constructed and real-world dataset (SciTechNews), with tuples of a publicly-available scientific paper, its corresponding news article, and an expert-written short summary snippet; 2) proposing a novel technical framework that integrates a paper{'}s discourse structure with its metadata to guide generation; and, 3) demonstrating with extensive automatic and human experiments that our model outperforms other baseline methods (e.g. Alpaca and ChatGPT) in elaborating a content plan meaningful for the target audience, simplify the information selected, and produce a coherent final report in a layman{'}s style.",
}
| Science journalism refers to the task of reporting technical findings of a scientific paper as a less technical news article to the general public audience. We aim to design an automated system to support this real-world task (i.e., automatic science journalism ) by 1) introducing a newly-constructed and real-world dataset (SciTechNews), with tuples of a publicly-available scientific paper, its corresponding news article, and an expert-written short summary snippet; 2) proposing a novel technical framework that integrates a paper{'}s discourse structure with its metadata to guide generation; and, 3) demonstrating with extensive automatic and human experiments that our model outperforms other baseline methods (e.g. Alpaca and ChatGPT) in elaborating a content plan meaningful for the target audience, simplify the information selected, and produce a coherent final report in a layman{'}s style. | [
"Cardenas, Ronald",
"Yao, Bingsheng",
"Wang, Dakuo",
"Hou, Yufang"
] | `Don't Get Too Technical with Me': A Discourse Structure-Based Framework for Automatic Science Journalism | emnlp-main.76 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.77.bib | https://aclanthology.org/2023.emnlp-main.77/ | @inproceedings{yang-etal-2023-lacma,
title = "{LACMA}: Language-Aligning Contrastive Learning with Meta-Actions for Embodied Instruction Following",
author = "Yang, Cheng-Fu and
Chen, Yen-Chun and
Yang, Jianwei and
Dai, Xiyang and
Yuan, Lu and
Wang, Yu-Chiang and
Chang, Kai-Wei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.77",
doi = "10.18653/v1/2023.emnlp-main.77",
pages = "1203--1217",
abstract = "End-to-end Transformers have demonstrated an impressive success rate for Embodied Instruction Following when the environment has been seen in training. However, they tend to struggle when deployed in an unseen environment. This lack of generalizability is due to the agent{'}s insensitivity to subtle changes in natural language instructions. To mitigate this issue, we propose explicitly aligning the agent{'}s hidden states with the instructions via contrastive learning. Nevertheless, the semantic gap between high-level language instructions and the agent{'}s low-level action space remains an obstacle. Therefore, we further introduce a novel concept of meta-actions to bridge the gap. Meta-actions are ubiquitous action patterns that can be parsed from the original action sequence. These patterns represent higher-level semantics that are intuitively aligned closer to the instructions. When meta-actions are applied as additional training signals, the agent generalizes better to unseen environments. Compared to a strong multi-modal Transformer baseline, we achieve a significant 4.5{\%} absolute gain in success rate in unseen environments of ALFRED Embodied Instruction Following. Additional analysis shows that the contrastive objective and meta-actions are complementary in achieving the best results, and the resulting agent better aligns its states with corresponding instructions, making it more suitable for real-world embodied agents.",
}
| End-to-end Transformers have demonstrated an impressive success rate for Embodied Instruction Following when the environment has been seen in training. However, they tend to struggle when deployed in an unseen environment. This lack of generalizability is due to the agent{'}s insensitivity to subtle changes in natural language instructions. To mitigate this issue, we propose explicitly aligning the agent{'}s hidden states with the instructions via contrastive learning. Nevertheless, the semantic gap between high-level language instructions and the agent{'}s low-level action space remains an obstacle. Therefore, we further introduce a novel concept of meta-actions to bridge the gap. Meta-actions are ubiquitous action patterns that can be parsed from the original action sequence. These patterns represent higher-level semantics that are intuitively aligned closer to the instructions. When meta-actions are applied as additional training signals, the agent generalizes better to unseen environments. Compared to a strong multi-modal Transformer baseline, we achieve a significant 4.5{\%} absolute gain in success rate in unseen environments of ALFRED Embodied Instruction Following. Additional analysis shows that the contrastive objective and meta-actions are complementary in achieving the best results, and the resulting agent better aligns its states with corresponding instructions, making it more suitable for real-world embodied agents. | [
"Yang, Cheng-Fu",
"Chen, Yen-Chun",
"Yang, Jianwei",
"Dai, Xiyang",
"Yuan, Lu",
"Wang, Yu-Chiang",
"Chang, Kai-Wei"
] | LACMA: Language-Aligning Contrastive Learning with Meta-Actions for Embodied Instruction Following | emnlp-main.77 | 2310.12344 | [
"https://github.com/joeyy5588/lacma"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.78.bib | https://aclanthology.org/2023.emnlp-main.78/ | @inproceedings{zhu-etal-2023-penalty,
title = "Penalty Decoding: Well Suppress the Self-Reinforcement Effect in Open-Ended Text Generation",
author = "Zhu, Wenhong and
Hao, Hongkun and
Wang, Rui",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.78",
doi = "10.18653/v1/2023.emnlp-main.78",
pages = "1218--1228",
abstract = "The decoding algorithm is critical for open-ended text generation, transforming latent representations into coherent and meaningful outputs. This paper investigates the self-reinforcement effect in text generation and the effectiveness of a repetition penalty to mitigate it. However, determining the optimal repetition penalty value is challenging. To tackle this, we propose a forgetting mechanism that disregards distant tokens, reducing the burden of penalty selection. In addition, we introduce a length penalty to address overly short sentences caused by excessive penalties. Our penalty decoding approach incorporating three strategies helps resolve issues with sampling methods deviating from factual information. Experimental results demonstrate the efficacy of our approach in generating high-quality sentences resembling human output.",
}
| The decoding algorithm is critical for open-ended text generation, transforming latent representations into coherent and meaningful outputs. This paper investigates the self-reinforcement effect in text generation and the effectiveness of a repetition penalty to mitigate it. However, determining the optimal repetition penalty value is challenging. To tackle this, we propose a forgetting mechanism that disregards distant tokens, reducing the burden of penalty selection. In addition, we introduce a length penalty to address overly short sentences caused by excessive penalties. Our penalty decoding approach incorporating three strategies helps resolve issues with sampling methods deviating from factual information. Experimental results demonstrate the efficacy of our approach in generating high-quality sentences resembling human output. | [
"Zhu, Wenhong",
"Hao, Hongkun",
"Wang, Rui"
] | Penalty Decoding: Well Suppress the Self-Reinforcement Effect in Open-Ended Text Generation | emnlp-main.78 | 2310.14971 | [
"https://github.com/zwhong714/penalty_decoding"
] | https://huggingface.co/papers/2310.14971 | 0 | 0 | 0 | 3 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.79.bib | https://aclanthology.org/2023.emnlp-main.79/ | @inproceedings{li-etal-2023-towards-robust,
title = "Towards Robust Pruning: An Adaptive Knowledge-Retention Pruning Strategy for Language Models",
author = "Li, Jianwei and
Lei, Qi and
Cheng, Wei and
Xu, Dongkuan",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.79",
doi = "10.18653/v1/2023.emnlp-main.79",
pages = "1229--1247",
abstract = "The pruning objective has recently extended beyond accuracy and sparsity to robustness in language models. Despite this, existing methods struggle to enhance robustness against adversarial attacks when continually increasing model sparsity and require a retraining process. As humans step into the era of large language models, these issues become increasingly prominent. This paper proposes that the robustness of language models is proportional to the extent of pre-trained knowledge they encompass. Accordingly, we introduce a post-training pruning strategy designed to faithfully replicate the embedding space and feature space of dense language models, aiming to conserve more pre-trained knowledge during the pruning process. In this setup, each layer{'}s reconstruction error not only originates from itself but also includes cumulative error from preceding layers, followed by an adaptive rectification. Compared to other state-of-art baselines, our approach demonstrates a superior balance between accuracy, sparsity, robustness, and pruning cost with BERT on datasets SST2, IMDB, and AGNews, marking a significant stride towards robust pruning in language models.",
}
| The pruning objective has recently extended beyond accuracy and sparsity to robustness in language models. Despite this, existing methods struggle to enhance robustness against adversarial attacks when continually increasing model sparsity and require a retraining process. As humans step into the era of large language models, these issues become increasingly prominent. This paper proposes that the robustness of language models is proportional to the extent of pre-trained knowledge they encompass. Accordingly, we introduce a post-training pruning strategy designed to faithfully replicate the embedding space and feature space of dense language models, aiming to conserve more pre-trained knowledge during the pruning process. In this setup, each layer{'}s reconstruction error not only originates from itself but also includes cumulative error from preceding layers, followed by an adaptive rectification. Compared to other state-of-art baselines, our approach demonstrates a superior balance between accuracy, sparsity, robustness, and pruning cost with BERT on datasets SST2, IMDB, and AGNews, marking a significant stride towards robust pruning in language models. | [
"Li, Jianwei",
"Lei, Qi",
"Cheng, Wei",
"Xu, Dongkuan"
] | Towards Robust Pruning: An Adaptive Knowledge-Retention Pruning Strategy for Language Models | emnlp-main.79 | 2310.13191 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.80.bib | https://aclanthology.org/2023.emnlp-main.80/ | @inproceedings{makhervaks-etal-2023-clinical,
title = "Clinical Contradiction Detection",
author = "Makhervaks, Dave and
Gillis, Plia and
Radinsky, Kira",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.80",
doi = "10.18653/v1/2023.emnlp-main.80",
pages = "1248--1263",
abstract = "Detecting contradictions in text is essential in determining the validity of the literature and sources that we consume. Medical corpora are riddled with conflicting statements. This is due to the large throughput of new studies and the difficulty in replicating experiments, such as clinical trials. Detecting contradictions in this domain is hard since it requires clinical expertise. We present a distant supervision approach that leverages a medical ontology to build a seed of potential clinical contradictions over 22 million medical abstracts. We automatically build a labeled training dataset consisting of paired clinical sentences that are grounded in an ontology and represent potential medical contradiction. The dataset is used to weakly-supervise state-of-the-art deep learning models showing significant empirical improvements across multiple medical contradiction datasets.",
}
| Detecting contradictions in text is essential in determining the validity of the literature and sources that we consume. Medical corpora are riddled with conflicting statements. This is due to the large throughput of new studies and the difficulty in replicating experiments, such as clinical trials. Detecting contradictions in this domain is hard since it requires clinical expertise. We present a distant supervision approach that leverages a medical ontology to build a seed of potential clinical contradictions over 22 million medical abstracts. We automatically build a labeled training dataset consisting of paired clinical sentences that are grounded in an ontology and represent potential medical contradiction. The dataset is used to weakly-supervise state-of-the-art deep learning models showing significant empirical improvements across multiple medical contradiction datasets. | [
"Makhervaks, Dave",
"Gillis, Plia",
"Radinsky, Kira"
] | Clinical Contradiction Detection | emnlp-main.80 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
||
https://aclanthology.org/2023.emnlp-main.81.bib | https://aclanthology.org/2023.emnlp-main.81/ | @inproceedings{liu-etal-2023-vera,
title = "Vera: A General-Purpose Plausibility Estimation Model for Commonsense Statements",
author = "Liu, Jiacheng and
Wang, Wenya and
Wang, Dianzhuo and
Smith, Noah and
Choi, Yejin and
Hajishirzi, Hannaneh",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.81",
doi = "10.18653/v1/2023.emnlp-main.81",
pages = "1264--1287",
abstract = "Today{'}s language models can be remarkably intelligent yet still produce text that contains trivial commonsense errors. Therefore, we seek a retrospective verification approach that can reflect on the commonsense plausibility of the machine text, and introduce Vera, a general-purpose model that learns to estimate the commonsense plausibility of declarative statements. To support diverse commonsense domains, Vera is trained on {\textasciitilde}7M commonsense statements that are automatically converted from 19 QA datasets and two commonsense knowledge bases, and using a combination of three training objectives. When applied to solving commonsense problems in the verification format, Vera substantially outperforms existing models that can be repurposed for commonsense verification, even including GPT-3.5/ChatGPT/GPT-4, and it further exhibits generalization capabilities to unseen tasks and provides well-calibrated outputs. We find that Vera excels at filtering machine-generated commonsense knowledge and is useful in detecting erroneous commonsense statements generated by models like ChatGPT in real-world settings.",
}
| Today{'}s language models can be remarkably intelligent yet still produce text that contains trivial commonsense errors. Therefore, we seek a retrospective verification approach that can reflect on the commonsense plausibility of the machine text, and introduce Vera, a general-purpose model that learns to estimate the commonsense plausibility of declarative statements. To support diverse commonsense domains, Vera is trained on {\textasciitilde}7M commonsense statements that are automatically converted from 19 QA datasets and two commonsense knowledge bases, and using a combination of three training objectives. When applied to solving commonsense problems in the verification format, Vera substantially outperforms existing models that can be repurposed for commonsense verification, even including GPT-3.5/ChatGPT/GPT-4, and it further exhibits generalization capabilities to unseen tasks and provides well-calibrated outputs. We find that Vera excels at filtering machine-generated commonsense knowledge and is useful in detecting erroneous commonsense statements generated by models like ChatGPT in real-world settings. | [
"Liu, Jiacheng",
"Wang, Wenya",
"Wang, Dianzhuo",
"Smith, Noah",
"Choi, Yejin",
"Hajishirzi, Hannaneh"
] | Vera: A General-Purpose Plausibility Estimation Model for Commonsense Statements | emnlp-main.81 | 2305.03695 | [
"https://github.com/liujch1998/vera"
] | https://huggingface.co/papers/2305.03695 | 3 | 3 | 0 | 6 | [
"liujch1998/vera"
] | [
"liujch1998/vera_contrib"
] | [
"liujch1998/vera"
] | 1 | Oral |
https://aclanthology.org/2023.emnlp-main.82.bib | https://aclanthology.org/2023.emnlp-main.82/ | @inproceedings{lin-etal-2023-text,
title = "Text-Transport: Toward Learning Causal Effects of Natural Language",
author = "Lin, Victoria and
Morency, Louis-Philippe and
Ben-Michael, Eli",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.82",
doi = "10.18653/v1/2023.emnlp-main.82",
pages = "1288--1304",
abstract = "As language technologies gain prominence in real-world settings, it is important to understand *how* changes to language affect reader perceptions. This can be formalized as the *causal effect* of varying a linguistic attribute (e.g., sentiment) on a reader{'}s response to the text. In this paper, we introduce Text-Transport, a method for estimation of causal effects from natural language under any text distribution. Current approaches for valid causal effect estimation require strong assumptions about the data, meaning the data from which one *can* estimate valid causal effects often is not representative of the actual target domain of interest. To address this issue, we leverage the notion of distribution shift to describe an estimator that *transports* causal effects between domains, bypassing the need for strong assumptions in the target domain. We derive statistical guarantees on the uncertainty of this estimator, and we report empirical results and analyses that support the validity of Text-Transport across data settings. Finally, we use Text-Transport to study a realistic setting{---}hate speech on social media{---}in which causal effects do shift significantly between text domains, demonstrating the necessity of transport when conducting causal inference on natural language.",
}
| As language technologies gain prominence in real-world settings, it is important to understand *how* changes to language affect reader perceptions. This can be formalized as the *causal effect* of varying a linguistic attribute (e.g., sentiment) on a reader{'}s response to the text. In this paper, we introduce Text-Transport, a method for estimation of causal effects from natural language under any text distribution. Current approaches for valid causal effect estimation require strong assumptions about the data, meaning the data from which one *can* estimate valid causal effects often is not representative of the actual target domain of interest. To address this issue, we leverage the notion of distribution shift to describe an estimator that *transports* causal effects between domains, bypassing the need for strong assumptions in the target domain. We derive statistical guarantees on the uncertainty of this estimator, and we report empirical results and analyses that support the validity of Text-Transport across data settings. Finally, we use Text-Transport to study a realistic setting{---}hate speech on social media{---}in which causal effects do shift significantly between text domains, demonstrating the necessity of transport when conducting causal inference on natural language. | [
"Lin, Victoria",
"Morency, Louis-Philippe",
"Ben-Michael, Eli"
] | Text-Transport: Toward Learning Causal Effects of Natural Language | emnlp-main.82 | 2310.20697 | [
"https://github.com/torylin/text-transport"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.83.bib | https://aclanthology.org/2023.emnlp-main.83/ | @inproceedings{pradeep-etal-2023-generative,
title = "How Does Generative Retrieval Scale to Millions of Passages?",
author = "Pradeep, Ronak and
Hui, Kai and
Gupta, Jai and
Lelkes, Adam and
Zhuang, Honglei and
Lin, Jimmy and
Metzler, Donald and
Tran, Vinh",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.83",
doi = "10.18653/v1/2023.emnlp-main.83",
pages = "1305--1321",
abstract = "The emerging paradigm of generative retrieval re-frames the classic information retrieval problem into a sequence-to-sequence modeling task, forgoing external indices and encoding an entire document corpus within a single Transformer. Although many different approaches have been proposed to improve the effectiveness of generative retrieval, they have only been evaluated on document corpora on the order of 100K in size. We conduct the first empirical study of generative retrieval techniques across various corpus scales, ultimately scaling up to the entire MS MARCO passage ranking task with a corpus of 8.8M passages and evaluating model sizes up to 11B parameters. We uncover several findings about scaling generative retrieval to millions of passages; notably, the central importance of using synthetic queries as document representations during indexing, the ineffectiveness of existing proposed architecture modifications when accounting for compute cost, and the limits of naively scaling model parameters with respect to retrieval performance. While we find that generative retrieval is competitive with state-of-the-art dual encoders on small corpora, scaling to millions of passages remains an important and unsolved challenge. We believe these findings will be valuable for the community to clarify the current state of generative retrieval, highlight the unique challenges, and inspire new research directions.",
}
| The emerging paradigm of generative retrieval re-frames the classic information retrieval problem into a sequence-to-sequence modeling task, forgoing external indices and encoding an entire document corpus within a single Transformer. Although many different approaches have been proposed to improve the effectiveness of generative retrieval, they have only been evaluated on document corpora on the order of 100K in size. We conduct the first empirical study of generative retrieval techniques across various corpus scales, ultimately scaling up to the entire MS MARCO passage ranking task with a corpus of 8.8M passages and evaluating model sizes up to 11B parameters. We uncover several findings about scaling generative retrieval to millions of passages; notably, the central importance of using synthetic queries as document representations during indexing, the ineffectiveness of existing proposed architecture modifications when accounting for compute cost, and the limits of naively scaling model parameters with respect to retrieval performance. While we find that generative retrieval is competitive with state-of-the-art dual encoders on small corpora, scaling to millions of passages remains an important and unsolved challenge. We believe these findings will be valuable for the community to clarify the current state of generative retrieval, highlight the unique challenges, and inspire new research directions. | [
"Pradeep, Ronak",
"Hui, Kai",
"Gupta, Jai",
"Lelkes, Adam",
"Zhuang, Honglei",
"Lin, Jimmy",
"Metzler, Donald",
"Tran, Vinh"
] | How Does Generative Retrieval Scale to Millions of Passages? | emnlp-main.83 | 2305.11841 | [
""
] | https://huggingface.co/papers/2305.11841 | 1 | 3 | 0 | 8 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.84.bib | https://aclanthology.org/2023.emnlp-main.84/ | @inproceedings{wen-etal-2023-unveiling,
title = "Unveiling the Implicit Toxicity in Large Language Models",
author = "Wen, Jiaxin and
Ke, Pei and
Sun, Hao and
Zhang, Zhexin and
Li, Chengfei and
Bai, Jinfeng and
Huang, Minlie",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.84",
doi = "10.18653/v1/2023.emnlp-main.84",
pages = "1322--1338",
abstract = "The open-endedness of large language models (LLMs) combined with their impressive capabilities may lead to new safety issues when being exploited for malicious use. While recent studies primarily focus on probing toxic outputs that can be easily detected with existing toxicity classifiers, we show that LLMs can generate diverse implicit toxic outputs that are exceptionally difficult to detect via simply zero-shot prompting. Moreover, we propose a reinforcement learning (RL) based attacking method to further induce the implicit toxicity in LLMs. Specifically, we optimize the language model with a reward that prefers implicit toxic outputs to explicit toxic and non-toxic ones. Experiments on five widely-adopted toxicity classifiers demonstrate that the attack success rate can be significantly improved through RL fine-tuning. For instance, the RL-finetuned LLaMA-13B model achieves an attack success rate of 90.04{\%} on BAD and 62.85{\%} on Davinci003. Our findings suggest that LLMs pose a significant threat in generating undetectable implicit toxic outputs. We further show that fine-tuning toxicity classifiers on the annotated examples from our attacking method can effectively enhance their ability to detect LLM-generated implicit toxic language.",
}
| The open-endedness of large language models (LLMs) combined with their impressive capabilities may lead to new safety issues when being exploited for malicious use. While recent studies primarily focus on probing toxic outputs that can be easily detected with existing toxicity classifiers, we show that LLMs can generate diverse implicit toxic outputs that are exceptionally difficult to detect via simply zero-shot prompting. Moreover, we propose a reinforcement learning (RL) based attacking method to further induce the implicit toxicity in LLMs. Specifically, we optimize the language model with a reward that prefers implicit toxic outputs to explicit toxic and non-toxic ones. Experiments on five widely-adopted toxicity classifiers demonstrate that the attack success rate can be significantly improved through RL fine-tuning. For instance, the RL-finetuned LLaMA-13B model achieves an attack success rate of 90.04{\%} on BAD and 62.85{\%} on Davinci003. Our findings suggest that LLMs pose a significant threat in generating undetectable implicit toxic outputs. We further show that fine-tuning toxicity classifiers on the annotated examples from our attacking method can effectively enhance their ability to detect LLM-generated implicit toxic language. | [
"Wen, Jiaxin",
"Ke, Pei",
"Sun, Hao",
"Zhang, Zhexin",
"Li, Chengfei",
"Bai, Jinfeng",
"Huang, Minlie"
] | Unveiling the Implicit Toxicity in Large Language Models | emnlp-main.84 | 2311.17391 | [
"https://github.com/thu-coai/implicit-toxicity"
] | https://huggingface.co/papers/2311.17391 | 0 | 0 | 0 | 7 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.85.bib | https://aclanthology.org/2023.emnlp-main.85/ | @inproceedings{qin-etal-2023-chatgpt,
title = "Is {C}hat{GPT} a General-Purpose Natural Language Processing Task Solver?",
author = "Qin, Chengwei and
Zhang, Aston and
Zhang, Zhuosheng and
Chen, Jiaao and
Yasunaga, Michihiro and
Yang, Diyi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.85",
doi = "10.18653/v1/2023.emnlp-main.85",
pages = "1339--1384",
abstract = "Spurred by advancements in scale, large language models (LLMs) have demonstrated the ability to perform a variety of natural language processing (NLP) tasks zero-shot{---}i.e., without adaptation on downstream data. Recently, the debut of ChatGPT has drawn a great deal of attention from the natural language processing (NLP) community due to the fact that it can generate high-quality responses to human input and self-correct previous mistakes based on subsequent conversations. However, it is not yet known whether ChatGPT can serve as a generalist model that can perform many NLP tasks zero-shot. In this work, we empirically analyze the zero-shot learning ability of ChatGPT by evaluating it on 20 popular NLP datasets covering 7 representative task categories. With extensive empirical studies, we demonstrate both the effectiveness and limitations of the current version of ChatGPT. We find that ChatGPT performs well on many tasks favoring reasoning capabilities (e.g., arithmetic reasoning) while it still faces challenges when solving specific tasks such as sequence tagging. We additionally provide in-depth analysis through qualitative case studies.",
}
| Spurred by advancements in scale, large language models (LLMs) have demonstrated the ability to perform a variety of natural language processing (NLP) tasks zero-shot{---}i.e., without adaptation on downstream data. Recently, the debut of ChatGPT has drawn a great deal of attention from the natural language processing (NLP) community due to the fact that it can generate high-quality responses to human input and self-correct previous mistakes based on subsequent conversations. However, it is not yet known whether ChatGPT can serve as a generalist model that can perform many NLP tasks zero-shot. In this work, we empirically analyze the zero-shot learning ability of ChatGPT by evaluating it on 20 popular NLP datasets covering 7 representative task categories. With extensive empirical studies, we demonstrate both the effectiveness and limitations of the current version of ChatGPT. We find that ChatGPT performs well on many tasks favoring reasoning capabilities (e.g., arithmetic reasoning) while it still faces challenges when solving specific tasks such as sequence tagging. We additionally provide in-depth analysis through qualitative case studies. | [
"Qin, Chengwei",
"Zhang, Aston",
"Zhang, Zhuosheng",
"Chen, Jiaao",
"Yasunaga, Michihiro",
"Yang, Diyi"
] | Is ChatGPT a General-Purpose Natural Language Processing Task Solver? | emnlp-main.85 | 2302.06476 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.86.bib | https://aclanthology.org/2023.emnlp-main.86/ | @inproceedings{xiao-etal-2023-length,
title = "Length is a Curse and a Blessing for Document-level Semantics",
author = "Xiao, Chenghao and
Li, Yizhi and
Hudson, G and
Lin, Chenghua and
Al Moubayed, Noura",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.86",
doi = "10.18653/v1/2023.emnlp-main.86",
pages = "1385--1396",
abstract = "In recent years, contrastive learning (CL) has been extensively utilized to recover sentence and document-level encoding capability from pre-trained language models. In this work, we question the length generalizability of CL-based models, i.e., their vulnerability towards length-induced semantic shift. We verify not only that length vulnerability is a significant yet overlooked research gap, but we can devise unsupervised CL methods solely depending on the semantic signal provided by document length. We first derive the theoretical foundations underlying length attacks, showing that elongating a document would intensify the high intra-document similarity that is already brought by CL. Moreover, we found that isotropy promised by CL is highly dependent on the length range of text exposed in training. Inspired by these findings, we introduce a simple yet universal document representation learning framework, **LA(SER)$^3$**: length-agnostic self-reference for semantically robust sentence representation learning, achieving state-of-the-art unsupervised performance on the standard information retrieval benchmark. [Our code is publicly available.](https://github.com/gowitheflow-1998/LA-SER-cubed)",
}
| In recent years, contrastive learning (CL) has been extensively utilized to recover sentence and document-level encoding capability from pre-trained language models. In this work, we question the length generalizability of CL-based models, i.e., their vulnerability towards length-induced semantic shift. We verify not only that length vulnerability is a significant yet overlooked research gap, but we can devise unsupervised CL methods solely depending on the semantic signal provided by document length. We first derive the theoretical foundations underlying length attacks, showing that elongating a document would intensify the high intra-document similarity that is already brought by CL. Moreover, we found that isotropy promised by CL is highly dependent on the length range of text exposed in training. Inspired by these findings, we introduce a simple yet universal document representation learning framework, **LA(SER)$^3$**: length-agnostic self-reference for semantically robust sentence representation learning, achieving state-of-the-art unsupervised performance on the standard information retrieval benchmark. [Our code is publicly available.](https://github.com/gowitheflow-1998/LA-SER-cubed) | [
"Xiao, Chenghao",
"Li, Yizhi",
"Hudson, G",
"Lin, Chenghua",
"Al Moubayed, Noura"
] | Length is a Curse and a Blessing for Document-level Semantics | emnlp-main.86 | 2310.16193 | [
"https://github.com/gowitheflow-1998/la-ser-cubed"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.87.bib | https://aclanthology.org/2023.emnlp-main.87/ | @inproceedings{yin-etal-2023-alcuna,
title = "{ALCUNA}: Large Language Models Meet New Knowledge",
author = "Yin, Xunjian and
Huang, Baizhou and
Wan, Xiaojun",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.87",
doi = "10.18653/v1/2023.emnlp-main.87",
pages = "1397--1414",
abstract = "With the rapid development of NLP, large-scale language models (LLMs) excel in various tasks across multiple domains now. However, existing benchmarks may not adequately measure these models{'} capabilities, especially when faced with new knowledge. In this paper, we address the lack of benchmarks to evaluate LLMs{'} ability to handle new knowledge, an important and challenging aspect in the rapidly evolving world. We propose an approach called KnowGen that generates new knowledge by altering existing entity attributes and relationships, resulting in artificial entities that are distinct from real-world entities. With KnowGen, we introduce a benchmark named ALCUNA to assess LLMs{'} abilities in knowledge understanding, differentiation, and association. We benchmark several LLMs, reveals that their performance in face of new knowledge is not satisfactory, particularly in reasoning between new and internal knowledge. We also explore the impact of entity similarity on the model{'}s understanding of entity knowledge and the influence of contextual entities. We appeal to the need for caution when using LLMs in new scenarios or with new knowledge, and hope that our benchmarks can help drive the development of LLMs in face of new knowledge.",
}
| With the rapid development of NLP, large-scale language models (LLMs) excel in various tasks across multiple domains now. However, existing benchmarks may not adequately measure these models{'} capabilities, especially when faced with new knowledge. In this paper, we address the lack of benchmarks to evaluate LLMs{'} ability to handle new knowledge, an important and challenging aspect in the rapidly evolving world. We propose an approach called KnowGen that generates new knowledge by altering existing entity attributes and relationships, resulting in artificial entities that are distinct from real-world entities. With KnowGen, we introduce a benchmark named ALCUNA to assess LLMs{'} abilities in knowledge understanding, differentiation, and association. We benchmark several LLMs, reveals that their performance in face of new knowledge is not satisfactory, particularly in reasoning between new and internal knowledge. We also explore the impact of entity similarity on the model{'}s understanding of entity knowledge and the influence of contextual entities. We appeal to the need for caution when using LLMs in new scenarios or with new knowledge, and hope that our benchmarks can help drive the development of LLMs in face of new knowledge. | [
"Yin, Xunjian",
"Huang, Baizhou",
"Wan, Xiaojun"
] | ALCUNA: Large Language Models Meet New Knowledge | emnlp-main.87 | 2310.14820 | [
"https://github.com/arvid-pku/alcuna"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.88.bib | https://aclanthology.org/2023.emnlp-main.88/ | @inproceedings{suwono-etal-2023-location,
title = "Location-Aware Visual Question Generation with Lightweight Models",
author = "Suwono, Nicholas and
Chen, Justin and
Hung, Tun and
Huang, Ting-Hao and
Liao, I-Bin and
Li, Yung-Hui and
Ku, Lun-Wei and
Sun, Shao-Hua",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.88",
doi = "10.18653/v1/2023.emnlp-main.88",
pages = "1415--1432",
abstract = "This work introduces a novel task, location-aware visual question generation (LocaVQG), which aims to generate engaging questions from data relevant to a particular geographical location. Specifically, we represent such location-aware information with surrounding images and a GPS coordinate. To tackle this task, we present a dataset generation pipeline that leverages GPT-4 to produce diverse and sophisticated questions. Then, we aim to learn a lightweight model that can address the LocaVQG task and fit on an edge device, such as a mobile phone. To this end, we propose a method which can reliably generate engaging questions from location-aware information. Our proposed method outperforms baselines regarding human evaluation (e.g., engagement, grounding, coherence) and automatic evaluation metrics (e.g., BERTScore, ROUGE-2). Moreover, we conduct extensive ablation studies to justify our proposed techniques for both generating the dataset and solving the task.",
}
| This work introduces a novel task, location-aware visual question generation (LocaVQG), which aims to generate engaging questions from data relevant to a particular geographical location. Specifically, we represent such location-aware information with surrounding images and a GPS coordinate. To tackle this task, we present a dataset generation pipeline that leverages GPT-4 to produce diverse and sophisticated questions. Then, we aim to learn a lightweight model that can address the LocaVQG task and fit on an edge device, such as a mobile phone. To this end, we propose a method which can reliably generate engaging questions from location-aware information. Our proposed method outperforms baselines regarding human evaluation (e.g., engagement, grounding, coherence) and automatic evaluation metrics (e.g., BERTScore, ROUGE-2). Moreover, we conduct extensive ablation studies to justify our proposed techniques for both generating the dataset and solving the task. | [
"Suwono, Nicholas",
"Chen, Justin",
"Hung, Tun",
"Huang, Ting-Hao",
"Liao, I-Bin",
"Li, Yung-Hui",
"Ku, Lun-Wei",
"Sun, Shao-Hua"
] | Location-Aware Visual Question Generation with Lightweight Models | emnlp-main.88 | 2310.15129 | [
"https://github.com/academiasinicanlplab/locavqg"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.89.bib | https://aclanthology.org/2023.emnlp-main.89/ | @inproceedings{hwang-shwartz-2023-memecap,
title = "{M}eme{C}ap: A Dataset for Captioning and Interpreting Memes",
author = "Hwang, EunJeong and
Shwartz, Vered",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.89",
doi = "10.18653/v1/2023.emnlp-main.89",
pages = "1433--1445",
abstract = "Memes are a widely popular tool for web users to express their thoughts using visual metaphors. Understanding memes requires recognizing and interpreting visual metaphors with respect to the text inside or around the meme, often while employing background knowledge and reasoning abilities. We present the task of meme captioning and release a new dataset, MemeCap. Our dataset contains 6.3K memes along with the title of the post containing the meme, the meme captions, the literal image caption, and the visual metaphors. Despite the recent success of vision and language (VL) models on tasks such as image captioning and visual question answering, our extensive experiments using state-of-the-art VL models show that they still struggle with visual metaphors, and perform substantially worse than humans.",
}
| Memes are a widely popular tool for web users to express their thoughts using visual metaphors. Understanding memes requires recognizing and interpreting visual metaphors with respect to the text inside or around the meme, often while employing background knowledge and reasoning abilities. We present the task of meme captioning and release a new dataset, MemeCap. Our dataset contains 6.3K memes along with the title of the post containing the meme, the meme captions, the literal image caption, and the visual metaphors. Despite the recent success of vision and language (VL) models on tasks such as image captioning and visual question answering, our extensive experiments using state-of-the-art VL models show that they still struggle with visual metaphors, and perform substantially worse than humans. | [
"Hwang, EunJeong",
"Shwartz, Vered"
] | MemeCap: A Dataset for Captioning and Interpreting Memes | emnlp-main.89 | 2305.13703 | [
"https://github.com/eujhwang/meme-cap"
] | https://huggingface.co/papers/2305.13703 | 0 | 0 | 0 | 2 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.90.bib | https://aclanthology.org/2023.emnlp-main.90/ | @inproceedings{choshen-etal-2023-start,
title = "Where to start? Analyzing the potential value of intermediate models",
author = "Choshen, Leshem and
Venezian, Elad and
Don-Yehiya, Shachar and
Slonim, Noam and
Katz, Yoav",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.90",
doi = "10.18653/v1/2023.emnlp-main.90",
pages = "1446--1470",
abstract = "Previous studies observed that finetuned models may be better base models than the vanilla pretrained model. Such a model, finetuned on some source dataset, may provide a better starting point for a new finetuning process on a desired target dataset. Here, we perform a systematic analysis of this \textit{intertraining} scheme, over a wide range of English classification tasks. Surprisingly, our analysis suggests that the potential intertraining gain can be analyzed \textit{independently} for the target dataset under consideration, and for a base model being considered as a starting point. Hence, a performant model is generally strong, even if its training data was not aligned with the target dataset. Furthermore, we leverage our analysis to propose a practical and efficient approach to determine if and how to select a base model in real-world settings. Last, we release an updating ranking of best models in the HuggingFace hub per architecture.",
}
| Previous studies observed that finetuned models may be better base models than the vanilla pretrained model. Such a model, finetuned on some source dataset, may provide a better starting point for a new finetuning process on a desired target dataset. Here, we perform a systematic analysis of this \textit{intertraining} scheme, over a wide range of English classification tasks. Surprisingly, our analysis suggests that the potential intertraining gain can be analyzed \textit{independently} for the target dataset under consideration, and for a base model being considered as a starting point. Hence, a performant model is generally strong, even if its training data was not aligned with the target dataset. Furthermore, we leverage our analysis to propose a practical and efficient approach to determine if and how to select a base model in real-world settings. Last, we release an updating ranking of best models in the HuggingFace hub per architecture. | [
"Choshen, Leshem",
"Venezian, Elad",
"Don-Yehiya, Shachar",
"Slonim, Noam",
"Katz, Yoav"
] | Where to start? Analyzing the potential value of intermediate models | emnlp-main.90 | 2211.00107 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.91.bib | https://aclanthology.org/2023.emnlp-main.91/ | @inproceedings{tay-etal-2023-transcending,
title = "Transcending Scaling Laws with 0.1{\%} Extra Compute",
author = "Tay, Yi and
Wei, Jason and
Chung, Hyung and
Tran, Vinh and
So, David and
Shakeri, Siamak and
Garcia, Xavier and
Zheng, Steven and
Rao, Jinfeng and
Chowdhery, Aakanksha and
Zhou, Denny and
Metzler, Donald and
Petrov, Slav and
Houlsby, Neil and
Le, Quoc and
Dehghani, Mostafa",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.91",
doi = "10.18653/v1/2023.emnlp-main.91",
pages = "1471--1486",
abstract = "Scaling language models improves performance but comes with significant computational costs. This paper proposes UL2R, a method that substantially improves existing language models and their scaling curves with a relatively tiny amount of extra compute. The key idea is to continue training a state-of-the-art large language model on a few more steps with UL2{'}s mixture-of-denoiser objective. We show that, with almost negligible extra computational costs and no new sources of data, we are able to substantially improve the scaling properties of large language models on downstream metrics. In this paper, we continue training a baseline language model, PaLM, with ULR2, introducing a new set of models at 8B, 62B, and 540B scale which we call U-PaLM. Impressively, at 540B scale, we show an approximately 2x computational savings rate where U-PaLM achieves the same performance as the final PaLM 540B model at around half its computational budget (i.e., saving {\textasciitilde}4.4 million TPUv4 hours). We further show that this improved scaling curve leads to {``}emergent abilities{''} on challenging BIG-Bench tasks{---}for instance, U-PaLM does much better on some tasks or demonstrates better quality at much smaller scale (62B as opposed to 540B). Overall, we show that U-PaLM outperforms PaLM on many few-shot setups, including reasoning tasks with chain-of-thought (e.g., GSM8K), multilingual tasks (MGSM, TydiQA), MMLU and challenging BIG-Bench tasks.",
}
| Scaling language models improves performance but comes with significant computational costs. This paper proposes UL2R, a method that substantially improves existing language models and their scaling curves with a relatively tiny amount of extra compute. The key idea is to continue training a state-of-the-art large language model on a few more steps with UL2{'}s mixture-of-denoiser objective. We show that, with almost negligible extra computational costs and no new sources of data, we are able to substantially improve the scaling properties of large language models on downstream metrics. In this paper, we continue training a baseline language model, PaLM, with ULR2, introducing a new set of models at 8B, 62B, and 540B scale which we call U-PaLM. Impressively, at 540B scale, we show an approximately 2x computational savings rate where U-PaLM achieves the same performance as the final PaLM 540B model at around half its computational budget (i.e., saving {\textasciitilde}4.4 million TPUv4 hours). We further show that this improved scaling curve leads to {``}emergent abilities{''} on challenging BIG-Bench tasks{---}for instance, U-PaLM does much better on some tasks or demonstrates better quality at much smaller scale (62B as opposed to 540B). Overall, we show that U-PaLM outperforms PaLM on many few-shot setups, including reasoning tasks with chain-of-thought (e.g., GSM8K), multilingual tasks (MGSM, TydiQA), MMLU and challenging BIG-Bench tasks. | [
"Tay, Yi",
"Wei, Jason",
"Chung, Hyung",
"Tran, Vinh",
"So, David",
"Shakeri, Siamak",
"Garcia, Xavier",
"Zheng, Steven",
"Rao, Jinfeng",
"Chowdhery, Aakanksha",
"Zhou, Denny",
"Metzler, Donald",
"Petrov, Slav",
"Houlsby, Neil",
"Le, Quoc",
"Dehghani, Mostafa"
] | Transcending Scaling Laws with 0.1% Extra Compute | emnlp-main.91 | 2210.11399 | [
""
] | https://huggingface.co/papers/2210.11399 | 1 | 0 | 0 | 16 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.92.bib | https://aclanthology.org/2023.emnlp-main.92/ | @inproceedings{li-etal-2023-coannotating,
title = "{C}o{A}nnotating: Uncertainty-Guided Work Allocation between Human and Large Language Models for Data Annotation",
author = "Li, Minzhi and
Shi, Taiwei and
Ziems, Caleb and
Kan, Min-Yen and
Chen, Nancy and
Liu, Zhengyuan and
Yang, Diyi",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.92",
doi = "10.18653/v1/2023.emnlp-main.92",
pages = "1487--1505",
abstract = "Annotated data plays a critical role in Natural Language Processing (NLP) in training models and evaluating their performance. Given recent developments in Large Language Models (LLMs), models such as ChatGPT demonstrate zero-shot capability on many text-annotation tasks, comparable with or even exceeding human annotators. Such LLMs can serve as alternatives for manual annotation, due to lower costs and higher scalability. However, limited work has leveraged LLMs as complementary annotators, nor explored how annotation work is best allocated among humans and LLMs to achieve both quality and cost objectives. We propose CoAnnotating, a novel paradigm for Human-LLM co-annotation of unstructured texts at scale. Under this framework, we utilize uncertainty to estimate LLMs{'} annotation capability. Our empirical study shows CoAnnotating to be an effective means to allocate work from results on different datasets, with up to 21{\%} performance improvement over random baseline. For code implementation, see https://github.com/SALT-NLP/CoAnnotating.",
}
| Annotated data plays a critical role in Natural Language Processing (NLP) in training models and evaluating their performance. Given recent developments in Large Language Models (LLMs), models such as ChatGPT demonstrate zero-shot capability on many text-annotation tasks, comparable with or even exceeding human annotators. Such LLMs can serve as alternatives for manual annotation, due to lower costs and higher scalability. However, limited work has leveraged LLMs as complementary annotators, nor explored how annotation work is best allocated among humans and LLMs to achieve both quality and cost objectives. We propose CoAnnotating, a novel paradigm for Human-LLM co-annotation of unstructured texts at scale. Under this framework, we utilize uncertainty to estimate LLMs{'} annotation capability. Our empirical study shows CoAnnotating to be an effective means to allocate work from results on different datasets, with up to 21{\%} performance improvement over random baseline. For code implementation, see https://github.com/SALT-NLP/CoAnnotating. | [
"Li, Minzhi",
"Shi, Taiwei",
"Ziems, Caleb",
"Kan, Min-Yen",
"Chen, Nancy",
"Liu, Zhengyuan",
"Yang, Diyi"
] | CoAnnotating: Uncertainty-Guided Work Allocation between Human and Large Language Models for Data Annotation | emnlp-main.92 | 2310.15638 | [
"https://github.com/salt-nlp/coannotating"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.93.bib | https://aclanthology.org/2023.emnlp-main.93/ | @inproceedings{berchansky-etal-2023-optimizing,
title = "Optimizing Retrieval-augmented Reader Models via Token Elimination",
author = "Berchansky, Moshe and
Izsak, Peter and
Caciularu, Avi and
Dagan, Ido and
Wasserblat, Moshe",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.93",
doi = "10.18653/v1/2023.emnlp-main.93",
pages = "1506--1524",
abstract = "Fusion-in-Decoder (FiD) is an effective retrieval-augmented language model applied across a variety of open-domain tasks, such as question answering, fact checking, etc. In FiD, supporting passages are first retrieved and then processed using a generative model (Reader), which can cause a significant bottleneck in decoding time, particularly with long outputs. In this work, we analyze the contribution and necessity of all the retrieved passages to the performance of reader models, and propose eliminating some of the retrieved information, at the token level, that might not contribute essential information to the answer generation process. We demonstrate that our method can reduce run-time by up to 62.2{\%}, with only a 2{\%} reduction in performance, and in some cases, even improve the performance results.",
}
| Fusion-in-Decoder (FiD) is an effective retrieval-augmented language model applied across a variety of open-domain tasks, such as question answering, fact checking, etc. In FiD, supporting passages are first retrieved and then processed using a generative model (Reader), which can cause a significant bottleneck in decoding time, particularly with long outputs. In this work, we analyze the contribution and necessity of all the retrieved passages to the performance of reader models, and propose eliminating some of the retrieved information, at the token level, that might not contribute essential information to the answer generation process. We demonstrate that our method can reduce run-time by up to 62.2{\%}, with only a 2{\%} reduction in performance, and in some cases, even improve the performance results. | [
"Berchansky, Moshe",
"Izsak, Peter",
"Caciularu, Avi",
"Dagan, Ido",
"Wasserblat, Moshe"
] | Optimizing Retrieval-augmented Reader Models via Token Elimination | emnlp-main.93 | 2310.13682 | [
"https://github.com/mosheber/token_elimination"
] | https://huggingface.co/papers/2310.13682 | 2 | 1 | 0 | 5 | [] | [] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.94.bib | https://aclanthology.org/2023.emnlp-main.94/ | @inproceedings{yang-etal-2023-wsdms,
title = "{WSDMS}: Debunk Fake News via Weakly Supervised Detection of Misinforming Sentences with Contextualized Social Wisdom",
author = "Yang, Ruichao and
Gao, Wei and
Ma, Jing and
Lin, Hongzhan and
Yang, Zhiwei",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.94",
doi = "10.18653/v1/2023.emnlp-main.94",
pages = "1525--1538",
abstract = "Fake news debunking primarily focuses on determining the truthfulness of news articles, which oversimplifies the issue as fake news often combines elements of both truth and falsehood. Thus, it becomes crucial to identify specific instances of misinformation within the articles. In this research, we investigate a novel task in the field of fake news debunking, which involves detecting sentence-level misinformation. One of the major challenges in this task is the absence of a training dataset with sentence-level annotations regarding veracity. Inspired by the Multiple Instance Learning (MIL) approach, we propose a model called Weakly Supervised Detection of Misinforming Sentences (WSDMS). This model only requires bag-level labels for training but is capable of inferring both sentence-level misinformation and article-level veracity, aided by relevant social media conversations that are attentively contextualized with news sentences. We evaluate WSDMS on three real-world benchmarks and demonstrate that it outperforms existing state-of-the-art baselines in debunking fake news at both the sentence and article levels.",
}
| Fake news debunking primarily focuses on determining the truthfulness of news articles, which oversimplifies the issue as fake news often combines elements of both truth and falsehood. Thus, it becomes crucial to identify specific instances of misinformation within the articles. In this research, we investigate a novel task in the field of fake news debunking, which involves detecting sentence-level misinformation. One of the major challenges in this task is the absence of a training dataset with sentence-level annotations regarding veracity. Inspired by the Multiple Instance Learning (MIL) approach, we propose a model called Weakly Supervised Detection of Misinforming Sentences (WSDMS). This model only requires bag-level labels for training but is capable of inferring both sentence-level misinformation and article-level veracity, aided by relevant social media conversations that are attentively contextualized with news sentences. We evaluate WSDMS on three real-world benchmarks and demonstrate that it outperforms existing state-of-the-art baselines in debunking fake news at both the sentence and article levels. | [
"Yang, Ruichao",
"Gao, Wei",
"Ma, Jing",
"Lin, Hongzhan",
"Yang, Zhiwei"
] | WSDMS: Debunk Fake News via Weakly Supervised Detection of Misinforming Sentences with Contextualized Social Wisdom | emnlp-main.94 | 2310.16579 | [
"https://github.com/hkbunlp/wsdms-emnlp2023"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.95.bib | https://aclanthology.org/2023.emnlp-main.95/ | @inproceedings{li-etal-2023-robust,
title = "Robust Prompt Optimization for Large Language Models Against Distribution Shifts",
author = "Li, Moxin and
Wang, Wenjie and
Feng, Fuli and
Cao, Yixin and
Zhang, Jizhi and
Chua, Tat-Seng",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.95",
doi = "10.18653/v1/2023.emnlp-main.95",
pages = "1539--1554",
abstract = "Large Language Model (LLM) has demonstrated significant ability in various Natural Language Processing tasks. However, their effectiveness is highly dependent on the phrasing of the task prompt, leading to research on automatic prompt optimization using labeled task data. We reveal that these prompt optimization techniques are vulnerable to distribution shifts such as subpopulation shifts, which are common for LLMs in real-world scenarios such as customer reviews analysis. In this light, we propose a new problem of robust prompt optimization for LLMs against distribution shifts, which requires the prompt optimized over the labeled source group can simultaneously generalize to an unlabeled target group. To solve this problem, we propose Generalized Prompt Optimization framework , which incorporates the unlabeled data from the target group into prompt optimization. Extensive experimental results demonstrate the effectiveness of the proposed framework with significant performance improvement on the target group and comparable performance on the source group.",
}
| Large Language Model (LLM) has demonstrated significant ability in various Natural Language Processing tasks. However, their effectiveness is highly dependent on the phrasing of the task prompt, leading to research on automatic prompt optimization using labeled task data. We reveal that these prompt optimization techniques are vulnerable to distribution shifts such as subpopulation shifts, which are common for LLMs in real-world scenarios such as customer reviews analysis. In this light, we propose a new problem of robust prompt optimization for LLMs against distribution shifts, which requires the prompt optimized over the labeled source group can simultaneously generalize to an unlabeled target group. To solve this problem, we propose Generalized Prompt Optimization framework , which incorporates the unlabeled data from the target group into prompt optimization. Extensive experimental results demonstrate the effectiveness of the proposed framework with significant performance improvement on the target group and comparable performance on the source group. | [
"Li, Moxin",
"Wang, Wenjie",
"Feng, Fuli",
"Cao, Yixin",
"Zhang, Jizhi",
"Chua, Tat-Seng"
] | Robust Prompt Optimization for Large Language Models Against Distribution Shifts | emnlp-main.95 | 2305.13954 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.96.bib | https://aclanthology.org/2023.emnlp-main.96/ | @inproceedings{josifoski-etal-2023-exploiting,
title = "Exploiting Asymmetry for Synthetic Training Data Generation: {S}ynth{IE} and the Case of Information Extraction",
author = "Josifoski, Martin and
Sakota, Marija and
Peyrard, Maxime and
West, Robert",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.96",
doi = "10.18653/v1/2023.emnlp-main.96",
pages = "1555--1574",
abstract = "Large language models (LLMs) have great potential for synthetic data generation. This work shows that useful data can be synthetically generated even for tasks that cannot be solved directly by LLMs: for problems with structured outputs, it is possible to prompt an LLM to perform the task in the reverse direction, by generating plausible input text for a target output structure. Leveraging this asymmetry in task difficulty makes it possible to produce large-scale, high-quality data for complex tasks. We demonstrate the effectiveness of this approach on closed information extraction, where collecting ground-truth data is challenging, and no satisfactory dataset exists to date. We synthetically generate a dataset of 1.8M data points, establish its superior quality compared to existing datasets in a human evaluation, and use it to finetune small models (220M and 770M parameters), termed SynthIE, that outperform the prior state of the art (with equal model size) by a substantial margin of 57 absolute points in micro-F1 and 79 points in macro-F1. Code, data, and models are available at anonymous.",
}
| Large language models (LLMs) have great potential for synthetic data generation. This work shows that useful data can be synthetically generated even for tasks that cannot be solved directly by LLMs: for problems with structured outputs, it is possible to prompt an LLM to perform the task in the reverse direction, by generating plausible input text for a target output structure. Leveraging this asymmetry in task difficulty makes it possible to produce large-scale, high-quality data for complex tasks. We demonstrate the effectiveness of this approach on closed information extraction, where collecting ground-truth data is challenging, and no satisfactory dataset exists to date. We synthetically generate a dataset of 1.8M data points, establish its superior quality compared to existing datasets in a human evaluation, and use it to finetune small models (220M and 770M parameters), termed SynthIE, that outperform the prior state of the art (with equal model size) by a substantial margin of 57 absolute points in micro-F1 and 79 points in macro-F1. Code, data, and models are available at anonymous. | [
"Josifoski, Martin",
"Sakota, Marija",
"Peyrard, Maxime",
"West, Robert"
] | Exploiting Asymmetry for Synthetic Training Data Generation: SynthIE and the Case of Information Extraction | emnlp-main.96 | 2303.04132 | [
"https://github.com/epfl-dlab/synthie"
] | https://huggingface.co/papers/2303.04132 | 0 | 0 | 0 | 4 | [
"martinjosifoski/SynthIE"
] | [
"martinjosifoski/SynthIE"
] | [] | 1 | Poster |
https://aclanthology.org/2023.emnlp-main.97.bib | https://aclanthology.org/2023.emnlp-main.97/ | @inproceedings{xu-etal-2023-condensing,
title = "Condensing Multilingual Knowledge with Lightweight Language-Specific Modules",
author = "Xu, Haoran and
Tan, Weiting and
Li, Shuyue and
Chen, Yunmo and
Van Durme, Benjamin and
Koehn, Philipp and
Murray, Kenton",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.97",
doi = "10.18653/v1/2023.emnlp-main.97",
pages = "1575--1587",
abstract = "Incorporating language-specific (LS) modules or Mixture-of-Experts (MoE) are proven methods to boost performance in multilingual model performance, but the scalability of these approaches to hundreds of languages or experts tends to be hard to manage. We present Language-specific Matrix Synthesis (LMS), a novel method that addresses the issue. LMS utilizes parameter-efficient and lightweight modules, reducing the number of parameters while outperforming existing methods, e.g., +1.73 BLEU over Switch Transformer on OPUS-100 multilingual translation. Additionally, we introduce Fuse Distillation (FD) to condense multilingual knowledge from multiple LS modules into a single shared module, improving model inference and storage efficiency. Our approach demonstrates superior scalability and performance compared to state-of-the-art methods.",
}
| Incorporating language-specific (LS) modules or Mixture-of-Experts (MoE) are proven methods to boost performance in multilingual model performance, but the scalability of these approaches to hundreds of languages or experts tends to be hard to manage. We present Language-specific Matrix Synthesis (LMS), a novel method that addresses the issue. LMS utilizes parameter-efficient and lightweight modules, reducing the number of parameters while outperforming existing methods, e.g., +1.73 BLEU over Switch Transformer on OPUS-100 multilingual translation. Additionally, we introduce Fuse Distillation (FD) to condense multilingual knowledge from multiple LS modules into a single shared module, improving model inference and storage efficiency. Our approach demonstrates superior scalability and performance compared to state-of-the-art methods. | [
"Xu, Haoran",
"Tan, Weiting",
"Li, Shuyue",
"Chen, Yunmo",
"Van Durme, Benjamin",
"Koehn, Philipp",
"Murray, Kenton"
] | Condensing Multilingual Knowledge with Lightweight Language-Specific Modules | emnlp-main.97 | 2305.13993 | [
"https://github.com/fe1ixxu/lms_fd"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.98.bib | https://aclanthology.org/2023.emnlp-main.98/ | @inproceedings{fernandez-etal-2023-framework,
title = "The Framework Tax: Disparities Between Inference Efficiency in {NLP} Research and Deployment",
author = "Fernandez, Jared and
Kahn, Jacob and
Na, Clara and
Bisk, Yonatan and
Strubell, Emma",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.98",
doi = "10.18653/v1/2023.emnlp-main.98",
pages = "1588--1600",
abstract = "Increased focus on the computational efficiency of systems in natural language processing has motivated the design of efficient model architectures and improvements to underlying hardware accelerators. However, the resulting increases in computational throughput and reductions in floating point operations have not directly translated to improvements in wall-clock inference latency. We demonstrate that these discrepancies can be largely attributed to bottlenecks introduced by deep learning frameworks. We denote this phenomena as the framework tax, and observe that the disparity is growing as hardware speed increases over time. In this work, we examine this phenomena through a series of case studies analyzing the effects of model design decisions, framework paradigms, and hardware platforms on total model latency. Based on our findings, we provide actionable recommendations to researchers and practitioners aimed at narrowing the gap between efficient NLP model research and practice.",
}
| Increased focus on the computational efficiency of systems in natural language processing has motivated the design of efficient model architectures and improvements to underlying hardware accelerators. However, the resulting increases in computational throughput and reductions in floating point operations have not directly translated to improvements in wall-clock inference latency. We demonstrate that these discrepancies can be largely attributed to bottlenecks introduced by deep learning frameworks. We denote this phenomena as the framework tax, and observe that the disparity is growing as hardware speed increases over time. In this work, we examine this phenomena through a series of case studies analyzing the effects of model design decisions, framework paradigms, and hardware platforms on total model latency. Based on our findings, we provide actionable recommendations to researchers and practitioners aimed at narrowing the gap between efficient NLP model research and practice. | [
"Fern",
"ez, Jared",
"Kahn, Jacob",
"Na, Clara",
"Bisk, Yonatan",
"Strubell, Emma"
] | The Framework Tax: Disparities Between Inference Efficiency in NLP Research and Deployment | emnlp-main.98 | 2302.06117 | [
"https://github.com/jaredfern/framework-tax"
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Oral |
|
https://aclanthology.org/2023.emnlp-main.99.bib | https://aclanthology.org/2023.emnlp-main.99/ | @inproceedings{pourreza-rafiei-2023-evaluating,
title = "Evaluating Cross-Domain Text-to-{SQL} Models and Benchmarks",
author = "Pourreza, Mohammadreza and
Rafiei, Davood",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.99",
doi = "10.18653/v1/2023.emnlp-main.99",
pages = "1601--1611",
abstract = "Text-to-SQL benchmarks play a crucial role in evaluating the progress made in the field and the ranking of different models. However, accurately matching a model-generated SQL query to a reference SQL query in a benchmark fails for various reasons, such as underspecified natural language queries, inherent assumptions in both model-generated and reference queries, and the non-deterministic nature of SQL output under certain conditions. In this paper, we conduct an extensive study of several prominent cross-domain text-to-SQL benchmarks and re-evaluate some of the top-performing models within these benchmarks, by both manually evaluating the SQL queries and rewriting them in equivalent expressions. Our evaluation reveals that attaining a perfect performance on these benchmarks is unfeasible due to the multiple interpretations that can be derived from the provided samples. Furthermore, we find that the true performance of the models is underestimated and their relative performance changes after a re-evaluation. Most notably, our evaluation reveals a surprising discovery: a recent GPT4-based model surpasses the gold standard reference queries in the Spider benchmark in our human evaluation. This finding highlights the importance of interpreting benchmark evaluations cautiously, while also acknowledging the critical role of additional independent evaluations in driving advancements in the field.",
}
| Text-to-SQL benchmarks play a crucial role in evaluating the progress made in the field and the ranking of different models. However, accurately matching a model-generated SQL query to a reference SQL query in a benchmark fails for various reasons, such as underspecified natural language queries, inherent assumptions in both model-generated and reference queries, and the non-deterministic nature of SQL output under certain conditions. In this paper, we conduct an extensive study of several prominent cross-domain text-to-SQL benchmarks and re-evaluate some of the top-performing models within these benchmarks, by both manually evaluating the SQL queries and rewriting them in equivalent expressions. Our evaluation reveals that attaining a perfect performance on these benchmarks is unfeasible due to the multiple interpretations that can be derived from the provided samples. Furthermore, we find that the true performance of the models is underestimated and their relative performance changes after a re-evaluation. Most notably, our evaluation reveals a surprising discovery: a recent GPT4-based model surpasses the gold standard reference queries in the Spider benchmark in our human evaluation. This finding highlights the importance of interpreting benchmark evaluations cautiously, while also acknowledging the critical role of additional independent evaluations in driving advancements in the field. | [
"Pourreza, Mohammadreza",
"Rafiei, Davood"
] | Evaluating Cross-Domain Text-to-SQL Models and Benchmarks | emnlp-main.99 | 2310.18538 | [
""
] | -1 | -1 | -1 | -1 | [] | [] | [] | 0 | Poster |
|
https://aclanthology.org/2023.emnlp-main.100.bib | https://aclanthology.org/2023.emnlp-main.100/ | @inproceedings{conia-etal-2023-increasing,
title = "Increasing Coverage and Precision of Textual Information in Multilingual Knowledge Graphs",
author = "Conia, Simone and
Li, Min and
Lee, Daniel and
Minhas, Umar and
Ilyas, Ihab and
Li, Yunyao",
editor = "Bouamor, Houda and
Pino, Juan and
Bali, Kalika",
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
month = dec,
year = "2023",
address = "Singapore",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.emnlp-main.100",
doi = "10.18653/v1/2023.emnlp-main.100",
pages = "1612--1634",
abstract = "Recent work in Natural Language Processing and Computer Vision has been using textual information {--} e.g., entity names and descriptions {--} available in knowledge graphs to ground neural models to high-quality structured data. However, when it comes to non-English languages, the quantity and quality of textual information are comparatively scarce. To address this issue, we introduce the novel task of automatic Knowledge Graph Completion (KGE) and perform a thorough investigation on bridging the gap in both the quantity and quality of textual information between English and non-English languages. More specifically, we: i) bring to light the problem of increasing multilingual coverage and precision of entity names and descriptions in Wikidata; ii) demonstrate that state-of-the-art methods, namely, Machine Translation (MT), Web Search (WS), and Large Language Models (LLMs), struggle with this task; iii) present M-NTA, a novel unsupervised approach that combines MT, WS, and LLMs to generate high-quality textual information; and, iv) study the impact of increasing multilingual coverage and precision of non-English textual information in Entity Linking, Knowledge Graph Completion, and Question Answering. As part of our effort towards better multilingual knowledge graphs, we also introduce WikiKGE-10, the first human-curated benchmark to evaluate KGE approaches in 10 languages across 7 language families.",
}
| Recent work in Natural Language Processing and Computer Vision has been using textual information {--} e.g., entity names and descriptions {--} available in knowledge graphs to ground neural models to high-quality structured data. However, when it comes to non-English languages, the quantity and quality of textual information are comparatively scarce. To address this issue, we introduce the novel task of automatic Knowledge Graph Completion (KGE) and perform a thorough investigation on bridging the gap in both the quantity and quality of textual information between English and non-English languages. More specifically, we: i) bring to light the problem of increasing multilingual coverage and precision of entity names and descriptions in Wikidata; ii) demonstrate that state-of-the-art methods, namely, Machine Translation (MT), Web Search (WS), and Large Language Models (LLMs), struggle with this task; iii) present M-NTA, a novel unsupervised approach that combines MT, WS, and LLMs to generate high-quality textual information; and, iv) study the impact of increasing multilingual coverage and precision of non-English textual information in Entity Linking, Knowledge Graph Completion, and Question Answering. As part of our effort towards better multilingual knowledge graphs, we also introduce WikiKGE-10, the first human-curated benchmark to evaluate KGE approaches in 10 languages across 7 language families. | [
"Conia, Simone",
"Li, Min",
"Lee, Daniel",
"Minhas, Umar",
"Ilyas, Ihab",
"Li, Yunyao"
] | Increasing Coverage and Precision of Textual Information in Multilingual Knowledge Graphs | emnlp-main.100 | 2311.15781 | [
"https://github.com/apple/ml-kge"
] | https://huggingface.co/papers/2311.15781 | 1 | 0 | 0 | 6 | [] | [
"davanstrien/ml-kge"
] | [] | 1 | Poster |