bibtex_url
stringlengths 41
50
| bibtext
stringlengths 693
2.88k
| abstract
stringlengths 0
2k
| authors
listlengths 1
45
| title
stringlengths 21
206
| id
stringlengths 7
16
| type
stringclasses 2
values | arxiv_id
stringlengths 9
12
⌀ |
|---|---|---|---|---|---|---|---|
https://aclanthology.org/2024.findings-acl.768.bib
|
@inproceedings{saleva-lignos-2024-language,
title = "Language Model Priors and Data Augmentation Strategies for Low-resource Machine Translation: A Case Study Using {F}innish to {N}orthern {S}{\'a}mi",
author = {S{\"a}lev{\"a}, Jonne and
Lignos, Constantine},
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.768",
pages = "12949--12956",
abstract = "We investigate ways of using monolingual data in both the source and target languages for improving low-resource machine translation. As a case study, we experiment with translation from Finnish to Northern S{\'a}mi.Our experiments show that while conventional backtranslation remains a strong contender, using synthetic target-side data when training backtranslation models can be helpful as well.We also show that monolingual data can be used to train a language model which can act as a regularizer without any augmentation of parallel data.",
}
|
We investigate ways of using monolingual data in both the source and target languages for improving low-resource machine translation. As a case study, we experiment with translation from Finnish to Northern S{\'a}mi.Our experiments show that while conventional backtranslation remains a strong contender, using synthetic target-side data when training backtranslation models can be helpful as well.We also show that monolingual data can be used to train a language model which can act as a regularizer without any augmentation of parallel data.
|
[
"S{\\\"a}lev{\\\"a}, Jonne",
"Lignos, Constantine"
] |
Language Model Priors and Data Augmentation Strategies for Low-resource Machine Translation: A Case Study Using {F}innish to {N}orthern {S}{\'a}mi
|
findings-acl.768
|
Poster
|
1906.04040v1
|
https://aclanthology.org/2024.findings-acl.769.bib
|
@inproceedings{flemings-annavaram-2024-differentially,
title = "Differentially Private Knowledge Distillation via Synthetic Text Generation",
author = "Flemings, James and
Annavaram, Murali",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.769",
pages = "12957--12968",
abstract = "Large Language models (LLMs) are achieving state-of-the-art performance in many different downstream tasks. However, the increasing urgency of data privacy puts pressure on practitioners to train LLMs with Differential Privacy (DP) on private data. Concurrently, the exponential growth in parameter size of LLMs necessitates model compression before deployment of LLMs on resource-constrained devices or latency-sensitive applications. Differential privacy and model compression generally must trade off utility loss to achieve their objectives. Moreover, simultaneously applying both schemes can compound the utility degradation. To this end, we propose DistilDP: a novel differentially private knowledge distillation algorithm that exploits synthetic data generated by a differentially private teacher LLM. The knowledge of a teacher LLM is transferred onto the student in two ways: one way from the synthetic data itself{--} the hard labels, and the other way by the output distribution of the teacher evaluated on the synthetic data{--} the soft labels. Furthermore, if the teacher and student share a similar architectural structure, we can further distill knowledge by aligning the hidden representations between both. Our experimental results demonstrate that DistilDP can substantially improve the utility over existing baselines, at least 9.0 PPL on the Big Patent dataset, with strong privacy parameters, $\epsilon=2$. These promising results progress privacy-preserving compression of autoregressive LLMs. Our code can be accessed here: https://github.com/james-flemings/dp{\_}compress.",
}
|
Large Language models (LLMs) are achieving state-of-the-art performance in many different downstream tasks. However, the increasing urgency of data privacy puts pressure on practitioners to train LLMs with Differential Privacy (DP) on private data. Concurrently, the exponential growth in parameter size of LLMs necessitates model compression before deployment of LLMs on resource-constrained devices or latency-sensitive applications. Differential privacy and model compression generally must trade off utility loss to achieve their objectives. Moreover, simultaneously applying both schemes can compound the utility degradation. To this end, we propose DistilDP: a novel differentially private knowledge distillation algorithm that exploits synthetic data generated by a differentially private teacher LLM. The knowledge of a teacher LLM is transferred onto the student in two ways: one way from the synthetic data itself{--} the hard labels, and the other way by the output distribution of the teacher evaluated on the synthetic data{--} the soft labels. Furthermore, if the teacher and student share a similar architectural structure, we can further distill knowledge by aligning the hidden representations between both. Our experimental results demonstrate that DistilDP can substantially improve the utility over existing baselines, at least 9.0 PPL on the Big Patent dataset, with strong privacy parameters, $\epsilon=2$. These promising results progress privacy-preserving compression of autoregressive LLMs. Our code can be accessed here: https://github.com/james-flemings/dp{\_}compress.
|
[
"Flemings, James",
"Annavaram, Murali"
] |
Differentially Private Knowledge Distillation via Synthetic Text Generation
|
findings-acl.769
|
Poster
|
2403.00932v2
|
https://aclanthology.org/2024.findings-acl.770.bib
|
@inproceedings{xu-etal-2024-kiwi,
title = "{KIWI}: A Dataset of Knowledge-Intensive Writing Instructions for Answering Research Questions",
author = "Xu, Fangyuan and
Lo, Kyle and
Soldaini, Luca and
Kuehl, Bailey and
Choi, Eunsol and
Wadden, David",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.770",
pages = "12969--12990",
abstract = "Large language models (LLMs) adapted to follow user instructions are now widely deployed as conversational agents. In this work, we examine one increasingly common instruction-following task: providing writing assistance to compose a long-form answer. To evaluate the capabilities of current LLMs on this task, we construct KIWI, a dataset of knowledge-intensive writing instructions in the scientific domain. Given a research question, an initial model-generated answer and a set of relevant papers, an expert annotator iteratively issues instructions for the model to revise and improve its answer. We collect 1,260 interaction turns from 234 interaction sessions with three state-of-the-art LLMs. Each turn includes a user instruction, a model response, and a human evaluation of the model response. Through a detailed analysis of the collected responses, we find that all models struggle to incorporate new information into an existing answer, and to perform precise and unambiguous edits. Further, we find that models struggle to judge whether their outputs successfully followed user instructions, with accuracy at least 10 points short of human agreement. Our findings indicate that KIWI will be a valuable resource to measure progress and improve LLMs{'} instruction-following capabilities for knowledge intensive writing tasks.",
}
|
Large language models (LLMs) adapted to follow user instructions are now widely deployed as conversational agents. In this work, we examine one increasingly common instruction-following task: providing writing assistance to compose a long-form answer. To evaluate the capabilities of current LLMs on this task, we construct KIWI, a dataset of knowledge-intensive writing instructions in the scientific domain. Given a research question, an initial model-generated answer and a set of relevant papers, an expert annotator iteratively issues instructions for the model to revise and improve its answer. We collect 1,260 interaction turns from 234 interaction sessions with three state-of-the-art LLMs. Each turn includes a user instruction, a model response, and a human evaluation of the model response. Through a detailed analysis of the collected responses, we find that all models struggle to incorporate new information into an existing answer, and to perform precise and unambiguous edits. Further, we find that models struggle to judge whether their outputs successfully followed user instructions, with accuracy at least 10 points short of human agreement. Our findings indicate that KIWI will be a valuable resource to measure progress and improve LLMs{'} instruction-following capabilities for knowledge intensive writing tasks.
|
[
"Xu, Fangyuan",
"Lo, Kyle",
"Soldaini, Luca",
"Kuehl, Bailey",
"Choi, Eunsol",
"Wadden, David"
] |
{KIWI}: A Dataset of Knowledge-Intensive Writing Instructions for Answering Research Questions
|
findings-acl.770
|
Poster
|
2403.03866v1
|
https://aclanthology.org/2024.findings-acl.771.bib
|
@inproceedings{shohan-etal-2024-xl,
title = "{XL}-{H}ead{T}ags: Leveraging Multimodal Retrieval Augmentation for the Multilingual Generation of News Headlines and Tags",
author = "Shohan, Faisal and
Nayeem, Mir Tafseer and
Islam, Samsul and
Akash, Abu Ubaida and
Joty, Shafiq",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.771",
pages = "12991--13024",
abstract = "Millions of news articles published online daily can overwhelm readers. Headlines and entity (topic) tags are essential for guiding readers to decide if the content is worth their time. While headline generation has been extensively studied, tag generation remains largely unexplored, yet it offers readers better access to topics of interest. The need for conciseness in capturing readers{'} attention necessitates improved content selection strategies for identifying salient and relevant segments within lengthy articles, thereby guiding language models effectively. To address this, we propose to leverage auxiliary information such as images and captions embedded in the articles to retrieve relevant sentences and utilize instruction tuning with variations to generate both headlines and tags for news articles in a multilingual context. To make use of the auxiliary information, we have compiled a dataset named XL-HeadTags, which includes 20 languages across 6 diverse language families. Through extensive evaluation, we demonstrate the effectiveness of our plug-and-play multimodal-multilingual retrievers for both tasks. Additionally, we have developed a suite of tools for processing and evaluating multilingual texts, significantly contributing to the research community by enabling more accurate and efficient analysis across languages.",
}
|
Millions of news articles published online daily can overwhelm readers. Headlines and entity (topic) tags are essential for guiding readers to decide if the content is worth their time. While headline generation has been extensively studied, tag generation remains largely unexplored, yet it offers readers better access to topics of interest. The need for conciseness in capturing readers{'} attention necessitates improved content selection strategies for identifying salient and relevant segments within lengthy articles, thereby guiding language models effectively. To address this, we propose to leverage auxiliary information such as images and captions embedded in the articles to retrieve relevant sentences and utilize instruction tuning with variations to generate both headlines and tags for news articles in a multilingual context. To make use of the auxiliary information, we have compiled a dataset named XL-HeadTags, which includes 20 languages across 6 diverse language families. Through extensive evaluation, we demonstrate the effectiveness of our plug-and-play multimodal-multilingual retrievers for both tasks. Additionally, we have developed a suite of tools for processing and evaluating multilingual texts, significantly contributing to the research community by enabling more accurate and efficient analysis across languages.
|
[
"Shohan, Faisal",
"Nayeem, Mir Tafseer",
"Islam, Samsul",
"Akash, Abu Ubaida",
"Joty, Shafiq"
] |
{XL}-{H}ead{T}ags: Leveraging Multimodal Retrieval Augmentation for the Multilingual Generation of News Headlines and Tags
|
findings-acl.771
|
Poster
|
2406.03776v2
|
https://aclanthology.org/2024.findings-acl.772.bib
|
@inproceedings{qin-etal-2024-infobench,
title = "{I}n{F}o{B}ench: Evaluating Instruction Following Ability in Large Language Models",
author = "Qin, Yiwei and
Song, Kaiqiang and
Hu, Yebowen and
Yao, Wenlin and
Cho, Sangwoo and
Wang, Xiaoyang and
Wu, Xuansheng and
Liu, Fei and
Liu, Pengfei and
Yu, Dong",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.772",
pages = "13025--13048",
abstract = "This paper introduces the Decomposed Requirements Following Ratio (DRFR), a new metric for evaluating Large Language Models{'} (LLMs) ability to follow instructions. Addressing a gap in current methodologies, DRFR breaks down complex instructions into simpler criteria, facilitating a detailed analysis of LLMs{'} compliance with various aspects of tasks. Alongside this metric, we present InFoBench, a benchmark comprising 500 diverse instructions and 2,250 decomposed questions across multiple constraint categories. Our experiments compare DRFR with traditional scoring methods and explore annotation sources, including human experts, crowd-sourced workers, and GPT-4. The findings demonstrate DRFR{'}s higher reliability and the effectiveness of using GPT-4 as a cost-efficient annotator. The evaluation of several advanced LLMs using this framework reveals their strengths and areas needing improvement, particularly in complex instruction-following. This study contributes a novel metric and benchmark, offering insights for future LLM development and evaluation.",
}
|
This paper introduces the Decomposed Requirements Following Ratio (DRFR), a new metric for evaluating Large Language Models{'} (LLMs) ability to follow instructions. Addressing a gap in current methodologies, DRFR breaks down complex instructions into simpler criteria, facilitating a detailed analysis of LLMs{'} compliance with various aspects of tasks. Alongside this metric, we present InFoBench, a benchmark comprising 500 diverse instructions and 2,250 decomposed questions across multiple constraint categories. Our experiments compare DRFR with traditional scoring methods and explore annotation sources, including human experts, crowd-sourced workers, and GPT-4. The findings demonstrate DRFR{'}s higher reliability and the effectiveness of using GPT-4 as a cost-efficient annotator. The evaluation of several advanced LLMs using this framework reveals their strengths and areas needing improvement, particularly in complex instruction-following. This study contributes a novel metric and benchmark, offering insights for future LLM development and evaluation.
|
[
"Qin, Yiwei",
"Song, Kaiqiang",
"Hu, Yebowen",
"Yao, Wenlin",
"Cho, Sangwoo",
"Wang, Xiaoyang",
"Wu, Xuansheng",
"Liu, Fei",
"Liu, Pengfei",
"Yu, Dong"
] |
{I}n{F}o{B}ench: Evaluating Instruction Following Ability in Large Language Models
|
findings-acl.772
|
Poster
|
2406.19999v1
|
https://aclanthology.org/2024.findings-acl.773.bib
|
@inproceedings{rashid-etal-2024-ecorank,
title = "{E}co{R}ank: Budget-Constrained Text Re-ranking Using Large Language Models",
author = "Rashid, Muhammad and
Meem, Jannat and
Dong, Yue and
Hristidis, Vagelis",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.773",
pages = "13049--13063",
abstract = "Large Language Models (LLMs) have achieved state-of-the-art performance in text re-ranking. This process includes queries and candidate passages in the prompts, utilizing pointwise, listwise, and pairwise prompting strategies. A limitation of these ranking strategies with LLMs is their cost: the process can become expensive due to API charges, which are based on the number of input and output tokens. We study how to maximize the re-ranking performance given a budget, by navigating the vast search spaces of prompt choices, LLM APIs, and budget splits. We propose a suite of budget-constrained methods to perform text re-ranking using a set of LLM APIs. Our most efficient method, called EcoRank, is a two-layered pipeline that jointly optimizes decisions regarding budget allocation across prompt strategies and LLM APIs. Our experimental results on four popular QA and passage reranking datasets show that EcoRank outperforms other budget-aware supervised and unsupervised baselines.",
}
|
Large Language Models (LLMs) have achieved state-of-the-art performance in text re-ranking. This process includes queries and candidate passages in the prompts, utilizing pointwise, listwise, and pairwise prompting strategies. A limitation of these ranking strategies with LLMs is their cost: the process can become expensive due to API charges, which are based on the number of input and output tokens. We study how to maximize the re-ranking performance given a budget, by navigating the vast search spaces of prompt choices, LLM APIs, and budget splits. We propose a suite of budget-constrained methods to perform text re-ranking using a set of LLM APIs. Our most efficient method, called EcoRank, is a two-layered pipeline that jointly optimizes decisions regarding budget allocation across prompt strategies and LLM APIs. Our experimental results on four popular QA and passage reranking datasets show that EcoRank outperforms other budget-aware supervised and unsupervised baselines.
|
[
"Rashid, Muhammad",
"Meem, Jannat",
"Dong, Yue",
"Hristidis, Vagelis"
] |
{E}co{R}ank: Budget-Constrained Text Re-ranking Using Large Language Models
|
findings-acl.773
|
Poster
|
2405.01122v1
|
https://aclanthology.org/2024.findings-acl.774.bib
|
@inproceedings{bhatia-etal-2024-fintral,
title = "{F}in{T}ral: A Family of {GPT}-4 Level Multimodal Financial Large Language Models",
author = "Bhatia, Gagan and
Nagoudi, El Moatez Billah and
Cavusoglu, Hasan and
Abdul-Mageed, Muhammad",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.774",
pages = "13064--13087",
abstract = "We introduce FinTral, a suite of state-of-the-art multimodal large language models (LLMs) built upon the Mistral-7b model and tailored for financial analysis. FinTral integrates textual, numerical, tabular, and image data. We enhance FinTral with domain-specific pretraining, instruction fine-tuning, and RLAIF training by exploiting a large collection of textual and visual datasets we curate for this work. We also introduce an extensive benchmark featuring nine tasks and 25 datasets for evaluation, including hallucinations in the financial domain. Our FinTral model trained with direct preference optimization employing advanced Tools and Retrieval methods, dubbed FinTral-DPO-T{\&}R, demonstrates an exceptional zero-shot performance. It outperforms ChatGPT-3.5 in all tasks and surpasses GPT-4 in five out of nine tasks, marking a significant advancement in AI-driven financial technology. We also demonstrate that FinTral has the potential to excel in real-time analysis and decision-making in diverse financial contexts.",
}
|
We introduce FinTral, a suite of state-of-the-art multimodal large language models (LLMs) built upon the Mistral-7b model and tailored for financial analysis. FinTral integrates textual, numerical, tabular, and image data. We enhance FinTral with domain-specific pretraining, instruction fine-tuning, and RLAIF training by exploiting a large collection of textual and visual datasets we curate for this work. We also introduce an extensive benchmark featuring nine tasks and 25 datasets for evaluation, including hallucinations in the financial domain. Our FinTral model trained with direct preference optimization employing advanced Tools and Retrieval methods, dubbed FinTral-DPO-T{\&}R, demonstrates an exceptional zero-shot performance. It outperforms ChatGPT-3.5 in all tasks and surpasses GPT-4 in five out of nine tasks, marking a significant advancement in AI-driven financial technology. We also demonstrate that FinTral has the potential to excel in real-time analysis and decision-making in diverse financial contexts.
|
[
"Bhatia, Gagan",
"Nagoudi, El Moatez Billah",
"Cavusoglu, Hasan",
"Abdul-Mageed, Muhammad"
] |
{F}in{T}ral: A Family of {GPT}-4 Level Multimodal Financial Large Language Models
|
findings-acl.774
|
Poster
|
2402.10986v3
|
https://aclanthology.org/2024.findings-acl.775.bib
|
@inproceedings{sun-etal-2024-aligning,
title = "Aligning Large Multimodal Models with Factually Augmented {RLHF}",
author = "Sun, Zhiqing and
Shen, Sheng and
Cao, Shengcao and
Liu, Haotian and
Li, Chunyuan and
Shen, Yikang and
Gan, Chuang and
Gui, Liangyan and
Wang, Yu-Xiong and
Yang, Yiming and
Keutzer, Kurt and
Darrell, Trevor",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.775",
pages = "13088--13110",
abstract = "Large Multimodal Models (LMM) are built across modalities and the misalignment between two modalities can result in {``}hallucination{''}, generating textual outputs that are not grounded by the multimodal information in context. To address the multimodal misalignment issue, we adapt the Reinforcement Learning from Human Feedback (RLHF) from the text domain to the vision-language alignment, where human annotators are asked to compare two responses and pinpoint the more hallucinated one, and the vision-language model is trained to maximize the simulated human rewards. We propose a new alignment algorithm called Factually Augmented RLHF that augments the reward model with additional factual information such as image captions and ground-truth multi-choice options, which alleviates the reward hacking phenomenon in RLHF and further improves the performance. We also enhance the GPT-4-generated training data (for vision instruction tuning) with previously available human-written image-text pairs to improve the general capabilities of our model. To evaluate the proposed approach in real-world scenarios, we develop a new evaluation benchmark MMHAL-BENCH with a special focus on penalizing hallucinations. As the first LMM trained with RLHF, our approach achieves remarkable improvement on the LLaVA-Bench dataset with the 96{\%} performance level of the text-only GPT-4 (while previous best methods can only achieve the 87{\%} level), and an improvement of 60{\%} on MMHAL-BENCH over other baselines.",
}
|
Large Multimodal Models (LMM) are built across modalities and the misalignment between two modalities can result in {``}hallucination{''}, generating textual outputs that are not grounded by the multimodal information in context. To address the multimodal misalignment issue, we adapt the Reinforcement Learning from Human Feedback (RLHF) from the text domain to the vision-language alignment, where human annotators are asked to compare two responses and pinpoint the more hallucinated one, and the vision-language model is trained to maximize the simulated human rewards. We propose a new alignment algorithm called Factually Augmented RLHF that augments the reward model with additional factual information such as image captions and ground-truth multi-choice options, which alleviates the reward hacking phenomenon in RLHF and further improves the performance. We also enhance the GPT-4-generated training data (for vision instruction tuning) with previously available human-written image-text pairs to improve the general capabilities of our model. To evaluate the proposed approach in real-world scenarios, we develop a new evaluation benchmark MMHAL-BENCH with a special focus on penalizing hallucinations. As the first LMM trained with RLHF, our approach achieves remarkable improvement on the LLaVA-Bench dataset with the 96{\%} performance level of the text-only GPT-4 (while previous best methods can only achieve the 87{\%} level), and an improvement of 60{\%} on MMHAL-BENCH over other baselines.
|
[
"Sun, Zhiqing",
"Shen, Sheng",
"Cao, Shengcao",
"Liu, Haotian",
"Li, Chunyuan",
"Shen, Yikang",
"Gan, Chuang",
"Gui, Liangyan",
"Wang, Yu-Xiong",
"Yang, Yiming",
"Keutzer, Kurt",
"Darrell, Trevor"
] |
Aligning Large Multimodal Models with Factually Augmented {RLHF}
|
findings-acl.775
|
Poster
|
2309.14525v1
|
https://aclanthology.org/2024.findings-acl.776.bib
|
@inproceedings{varshney-etal-2024-art,
title = "The Art of Defending: A Systematic Evaluation and Analysis of {LLM} Defense Strategies on Safety and Over-Defensiveness",
author = "Varshney, Neeraj and
Dolin, Pavel and
Seth, Agastya and
Baral, Chitta",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.776",
pages = "13111--13128",
abstract = "As Large Language Models (LLMs) play an increasingly pivotal role in natural language processing applications, their safety concerns become critical areas of NLP research. This has resulted in the development of various LLM defense strategies. Unfortunately, despite the shared goal of improving the safety of LLMs, the evaluation suites across various research works are disjoint and lack diverse inputs to ensure accurate and precise evaluation estimates. Furthermore, the important factor of {`}over-defensiveness{'} on the safe inputs has largely remained overlooked. Addressing these limitations, this paper presents a systematic evaluation, comparison, and analysis of various LLM defense strategies over both {`}safety{'} and {`}over-defensiveness{'}. To this end, we compile a large and diverse collection of safe and unsafe prompts, design precise evaluation methodology, and study the efficacy of various LLM defense strategies on multiple state-of-the-art LLMs. Our work reveals a number of crucial findings that we believe will pave the way and also facilitate further research in the critical area of improving the safety of LLMs.",
}
|
As Large Language Models (LLMs) play an increasingly pivotal role in natural language processing applications, their safety concerns become critical areas of NLP research. This has resulted in the development of various LLM defense strategies. Unfortunately, despite the shared goal of improving the safety of LLMs, the evaluation suites across various research works are disjoint and lack diverse inputs to ensure accurate and precise evaluation estimates. Furthermore, the important factor of {`}over-defensiveness{'} on the safe inputs has largely remained overlooked. Addressing these limitations, this paper presents a systematic evaluation, comparison, and analysis of various LLM defense strategies over both {`}safety{'} and {`}over-defensiveness{'}. To this end, we compile a large and diverse collection of safe and unsafe prompts, design precise evaluation methodology, and study the efficacy of various LLM defense strategies on multiple state-of-the-art LLMs. Our work reveals a number of crucial findings that we believe will pave the way and also facilitate further research in the critical area of improving the safety of LLMs.
|
[
"Varshney, Neeraj",
"Dolin, Pavel",
"Seth, Agastya",
"Baral, Chitta"
] |
The Art of Defending: A Systematic Evaluation and Analysis of {LLM} Defense Strategies on Safety and Over-Defensiveness
|
findings-acl.776
|
Poster
|
2401.00287v1
|
https://aclanthology.org/2024.findings-acl.777.bib
|
@inproceedings{meem-etal-2024-pat,
title = "{PAT}-Questions: A Self-Updating Benchmark for Present-Anchored Temporal Question-Answering",
author = "Meem, Jannat and
Rashid, Muhammad and
Dong, Yue and
Hristidis, Vagelis",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.777",
pages = "13129--13148",
abstract = "Existing work on Temporal Question Answering (TQA) has predominantly focused on questions anchored to specific timestamps or events (e.g. {`}Who was the US president in 1970?{'}). Little work has studied questions whose temporal context is relative to the present time (e.g. {`}Who was the previous US president?{'}). We refer to this problem as Present-Anchored Temporal QA (PATQA). PATQA poses unique challenges: (1) large language models (LLMs) may have outdated knowledge, (2) complex temporal relationships (e.g. {`}before{'}, {`}previous{'}) are hard to reason, (3) multi-hop reasoning may be required, and (4) the gold answers of benchmarks must be continuously updated. To address these challenges, we introduce the PAT-Questions benchmark, which includes single and multi-hop temporal questions. The answers in PAT-Questions can be automatically refreshed by re-running SPARQL queries on a knowledge graph, if available. We evaluate several state-of-the-art LLMs and a SOTA temporal reasoning model (TEMPREASON-T5) on PAT-Questions through direct prompting and retrieval-augmented generation (RAG). The results highlight the limitations of existing solutions in PATQA and motivate the need for new methods to improve PATQA reasoning capabilities.",
}
|
Existing work on Temporal Question Answering (TQA) has predominantly focused on questions anchored to specific timestamps or events (e.g. {`}Who was the US president in 1970?{'}). Little work has studied questions whose temporal context is relative to the present time (e.g. {`}Who was the previous US president?{'}). We refer to this problem as Present-Anchored Temporal QA (PATQA). PATQA poses unique challenges: (1) large language models (LLMs) may have outdated knowledge, (2) complex temporal relationships (e.g. {`}before{'}, {`}previous{'}) are hard to reason, (3) multi-hop reasoning may be required, and (4) the gold answers of benchmarks must be continuously updated. To address these challenges, we introduce the PAT-Questions benchmark, which includes single and multi-hop temporal questions. The answers in PAT-Questions can be automatically refreshed by re-running SPARQL queries on a knowledge graph, if available. We evaluate several state-of-the-art LLMs and a SOTA temporal reasoning model (TEMPREASON-T5) on PAT-Questions through direct prompting and retrieval-augmented generation (RAG). The results highlight the limitations of existing solutions in PATQA and motivate the need for new methods to improve PATQA reasoning capabilities.
|
[
"Meem, Jannat",
"Rashid, Muhammad",
"Dong, Yue",
"Hristidis, Vagelis"
] |
{PAT}-Questions: A Self-Updating Benchmark for Present-Anchored Temporal Question-Answering
|
findings-acl.777
|
Poster
|
2403.19913v2
|
https://aclanthology.org/2024.findings-acl.778.bib
|
@inproceedings{gao-etal-2024-360,
title = "$360^\circ${REA}: Towards A Reusable Experience Accumulation with $360^\circ$ Assessment for Multi-Agent System",
author = "Gao, Shen and
Li, Hao and
Shi, Zhengliang and
Huang, Chengrui and
Tu, Quan and
Shang, Shuo and
Tian, Zhiliang and
Huang, Minlie",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.778",
pages = "13149--13162",
abstract = "Large language model agents have demonstrated remarkable advancements across various complex tasks. Recent works focus on optimizing the agent team or employing self-reflection to iteratively solve complex tasks. Since these agents are all based on the same LLM, only conducting self-evaluation or removing underperforming agents does not substantively enhance the capability of the agents. We argue that a comprehensive evaluation and accumulating experience from evaluation feedback is an effective approach to improving system performance. In this paper, we propose Reusable Experience Accumulation with $\mathbf{360^\circ}$ Assessment ($360^\circ$REA), a hierarchical multi-agent framework inspired by corporate organizational practices. The framework employs a novel $360^\circ$ performance assessment method for multi-perspective performance evaluation with fine-grained assessment. To enhance the capability of agents in addressing complex tasks, we introduce dual-level experience pool for agents to accumulate experience through fine-grained assessment. Extensive experiments on complex task datasets demonstrate the effectiveness of $360^\circ$REA.",
}
|
Large language model agents have demonstrated remarkable advancements across various complex tasks. Recent works focus on optimizing the agent team or employing self-reflection to iteratively solve complex tasks. Since these agents are all based on the same LLM, only conducting self-evaluation or removing underperforming agents does not substantively enhance the capability of the agents. We argue that a comprehensive evaluation and accumulating experience from evaluation feedback is an effective approach to improving system performance. In this paper, we propose Reusable Experience Accumulation with $\mathbf{360^\circ}$ Assessment ($360^\circ$REA), a hierarchical multi-agent framework inspired by corporate organizational practices. The framework employs a novel $360^\circ$ performance assessment method for multi-perspective performance evaluation with fine-grained assessment. To enhance the capability of agents in addressing complex tasks, we introduce dual-level experience pool for agents to accumulate experience through fine-grained assessment. Extensive experiments on complex task datasets demonstrate the effectiveness of $360^\circ$REA.
|
[
"Gao, Shen",
"Li, Hao",
"Shi, Zhengliang",
"Huang, Chengrui",
"Tu, Quan",
"Shang, Shuo",
"Tian, Zhiliang",
"Huang, Minlie"
] |
$360^\circ${REA}: Towards A Reusable Experience Accumulation with $360^\circ$ Assessment for Multi-Agent System
|
findings-acl.778
|
Poster
|
2404.05569
|
https://aclanthology.org/2024.findings-acl.779.bib
|
@inproceedings{khalighinejad-etal-2024-extracting,
title = "Extracting Polymer Nanocomposite Samples from Full-Length Documents",
author = "Khalighinejad, Ghazal and
Circi, Defne and
Brinson, L. and
Dhingra, Bhuwan",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.779",
pages = "13163--13175",
abstract = "This paper investigates the use of large language models (LLMs) for extracting sample lists of polymer nanocomposites (PNCs) from full-length materials science research papers. The challenge lies in the complex nature of PNC samples, which have numerous attributes scattered throughout the text. The complexity of annotating detailed information on PNCs limits the availability of data, making conventional document-level relation extraction techniques impractical due to the challenge in creating comprehensive named entity span annotations.To address this, we introduce a new benchmark and an evaluation technique for this task and explore different prompting strategies in a zero-shot manner. We also incorporate self-consistency to improve the performance. Our findings show that even advanced LLMs struggle to extract all of the samples from an article. Finally, we analyze the errors encountered in this process, categorizing them into three main challenges, and discuss potential strategies for future research to overcome them.",
}
|
This paper investigates the use of large language models (LLMs) for extracting sample lists of polymer nanocomposites (PNCs) from full-length materials science research papers. The challenge lies in the complex nature of PNC samples, which have numerous attributes scattered throughout the text. The complexity of annotating detailed information on PNCs limits the availability of data, making conventional document-level relation extraction techniques impractical due to the challenge in creating comprehensive named entity span annotations.To address this, we introduce a new benchmark and an evaluation technique for this task and explore different prompting strategies in a zero-shot manner. We also incorporate self-consistency to improve the performance. Our findings show that even advanced LLMs struggle to extract all of the samples from an article. Finally, we analyze the errors encountered in this process, categorizing them into three main challenges, and discuss potential strategies for future research to overcome them.
|
[
"Khalighinejad, Ghazal",
"Circi, Defne",
"Brinson, L.",
"Dhingra, Bhuwan"
] |
Extracting Polymer Nanocomposite Samples from Full-Length Documents
|
findings-acl.779
|
Poster
|
2403.00260v1
|
https://aclanthology.org/2024.findings-acl.780.bib
|
@inproceedings{tsai-etal-2024-leveraging,
title = "Leveraging {LLM} Reasoning Enhances Personalized Recommender Systems",
author = "Tsai, Alicia and
Kraft, Adam and
Jin, Long and
Cai, Chenwei and
Hosseini, Anahita and
Xu, Taibai and
Zhang, Zemin and
Hong, Lichan and
Chi, Ed and
Yi, Xinyang",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.780",
pages = "13176--13188",
abstract = "Recent advancements have showcased the potential of Large Language Models (LLMs) in executing reasoning tasks, particularly facilitated by Chain-of-Thought (CoT) prompting. While tasks like arithmetic reasoning involve clear, definitive answers and logical chains of thought, the application of LLM reasoning in recommendation systems (RecSys) presents a distinct challenge. RecSys tasks revolve around subjectivity and personalized preferences, an under-explored domain in utilizing LLMs{'} reasoning capabilities. Our study explores several aspects to better understand reasoning for RecSys and demonstrate how task quality improves by utilizing LLM reasoning for both zero-shot and fine-tuning settings. Additionally, we propose $\textbf{Rec-SAVER}$ ($\textbf{Rec}$ommender $\textbf{S}$ystems $\textbf{A}$utomatic $\textbf{V}$erification and $\textbf{E}$valuation of $\textbf{R}$easoning) to automatically assess the quality of LLM reasoning responses without the requirement of curated gold references or human raters. We show that our framework aligns with real human judgment on the coherence and faithfulness of reasoning responses. Overall, our work shows that incorporating reasoning into RecSys can improve personalized tasks, paving the way for further advancements in recommender system methodologies.",
}
|
Recent advancements have showcased the potential of Large Language Models (LLMs) in executing reasoning tasks, particularly facilitated by Chain-of-Thought (CoT) prompting. While tasks like arithmetic reasoning involve clear, definitive answers and logical chains of thought, the application of LLM reasoning in recommendation systems (RecSys) presents a distinct challenge. RecSys tasks revolve around subjectivity and personalized preferences, an under-explored domain in utilizing LLMs{'} reasoning capabilities. Our study explores several aspects to better understand reasoning for RecSys and demonstrate how task quality improves by utilizing LLM reasoning for both zero-shot and fine-tuning settings. Additionally, we propose $\textbf{Rec-SAVER}$ ($\textbf{Rec}$ommender $\textbf{S}$ystems $\textbf{A}$utomatic $\textbf{V}$erification and $\textbf{E}$valuation of $\textbf{R}$easoning) to automatically assess the quality of LLM reasoning responses without the requirement of curated gold references or human raters. We show that our framework aligns with real human judgment on the coherence and faithfulness of reasoning responses. Overall, our work shows that incorporating reasoning into RecSys can improve personalized tasks, paving the way for further advancements in recommender system methodologies.
|
[
"Tsai, Alicia",
"Kraft, Adam",
"Jin, Long",
"Cai, Chenwei",
"Hosseini, Anahita",
"Xu, Taibai",
"Zhang, Zemin",
"Hong, Lichan",
"Chi, Ed",
"Yi, Xinyang"
] |
Leveraging {LLM} Reasoning Enhances Personalized Recommender Systems
|
findings-acl.780
|
Poster
|
2408.00802v1
|
https://aclanthology.org/2024.findings-acl.781.bib
|
@inproceedings{elmadany-etal-2024-toucan,
title = "Toucan: Many-to-Many Translation for 150 {A}frican Language Pairs",
author = "Elmadany, AbdelRahim and
Adebara, Ife and
Abdul-Mageed, Muhammad",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.781",
pages = "13189--13206",
abstract = "We address a notable gap in Natural Language Processing (NLP) by introducing a collection of resources designed to improve Machine Translation (MT) for low-resource languages, with a specific focus on African languages. First, We introduce two language models (LMs), Cheetah-1.2B and Cheetah-3.7B, with 1.2 billion and 3.7 billion parameters respectively. Next, we finetune the aforementioned models to create Toucan, an Afrocentric machine translation model designed to support 156 African language pairs. To evaluate Toucan, we carefully develop an extensive machine translation benchmark, dubbed Afro-Lingu-MT, tailored for evaluating machine translation. Toucan significantly outperforms other models, showcasing its remarkable performance on MT for African languages. Finally, we train a new model, spBLEU-1K, to enhance translation evaluation metrics, covering 1K languages, including African languages. This work aims to advance the field of NLP, fostering cross-cultural understanding and knowledge exchange, particularly in regions with limited language resources such as Africa.",
}
|
We address a notable gap in Natural Language Processing (NLP) by introducing a collection of resources designed to improve Machine Translation (MT) for low-resource languages, with a specific focus on African languages. First, We introduce two language models (LMs), Cheetah-1.2B and Cheetah-3.7B, with 1.2 billion and 3.7 billion parameters respectively. Next, we finetune the aforementioned models to create Toucan, an Afrocentric machine translation model designed to support 156 African language pairs. To evaluate Toucan, we carefully develop an extensive machine translation benchmark, dubbed Afro-Lingu-MT, tailored for evaluating machine translation. Toucan significantly outperforms other models, showcasing its remarkable performance on MT for African languages. Finally, we train a new model, spBLEU-1K, to enhance translation evaluation metrics, covering 1K languages, including African languages. This work aims to advance the field of NLP, fostering cross-cultural understanding and knowledge exchange, particularly in regions with limited language resources such as Africa.
|
[
"Elmadany, AbdelRahim",
"Adebara, Ife",
"Abdul-Mageed, Muhammad"
] |
Toucan: Many-to-Many Translation for 150 {A}frican Language Pairs
|
findings-acl.781
|
Poster
|
2407.04796v2
|
https://aclanthology.org/2024.findings-acl.782.bib
|
@inproceedings{xie-etal-2024-shot-dialogue,
title = "Few-shot Dialogue Strategy Learning for Motivational Interviewing via Inductive Reasoning",
author = "Xie, Zhouhang and
Majumder, Bodhisattwa Prasad and
Zhao, Mengjie and
Maeda, Yoshinori and
Yamada, Keiichi and
Wakaki, Hiromi and
McAuley, Julian",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.782",
pages = "13207--13219",
abstract = "We consider the task of building a dialogue system that can motivate users to adopt positive lifestyle changes, Motivational Interviewing (MI). Addressing such a task requires a system that could infer \textit{how} to motivate the user effectively. We propose DIIR, a framework that is capable of learning and applying conversation strategies in the form of natural language inductive rules from expert demonstrations. Automatic and human evaluation on instruction-following large language models show natural language strategies descriptions discovered by DIIR can improve active listening skills, reduce unsolicited advice, and promote more collaborative and less authoritative conversations, outperforming in-context demonstrations that are over 50 times longer.",
}
|
We consider the task of building a dialogue system that can motivate users to adopt positive lifestyle changes, Motivational Interviewing (MI). Addressing such a task requires a system that could infer \textit{how} to motivate the user effectively. We propose DIIR, a framework that is capable of learning and applying conversation strategies in the form of natural language inductive rules from expert demonstrations. Automatic and human evaluation on instruction-following large language models show natural language strategies descriptions discovered by DIIR can improve active listening skills, reduce unsolicited advice, and promote more collaborative and less authoritative conversations, outperforming in-context demonstrations that are over 50 times longer.
|
[
"Xie, Zhouhang",
"Majumder, Bodhisattwa Prasad",
"Zhao, Mengjie",
"Maeda, Yoshinori",
"Yamada, Keiichi",
"Wakaki, Hiromi",
"McAuley, Julian"
] |
Few-shot Dialogue Strategy Learning for Motivational Interviewing via Inductive Reasoning
|
findings-acl.782
|
Poster
|
2403.15737v1
|
https://aclanthology.org/2024.findings-acl.783.bib
|
@inproceedings{kumon-etal-2024-evaluating,
title = "Evaluating Structural Generalization in Neural Machine Translation",
author = "Kumon, Ryoma and
Matsuoka, Daiki and
Yanaka, Hitomi",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.783",
pages = "13220--13239",
abstract = "Compositional generalization refers to the ability to generalize to novel combinations of previously observed words and syntactic structures.Since it is regarded as a desired property of neural models, recent work has assessed compositional generalization in machine translation as well as semantic parsing.However, previous evaluations with machine translation have focused mostly on lexical generalization (i.e., generalization to unseen combinations of known words).Thus, it remains unclear to what extent models can translate sentences that require structural generalization (i.e., generalization to different sorts of syntactic structures).To address this question, we construct SGET, a machine translation dataset covering various types of compositional generalization with control of words and sentence structures.We evaluate neural machine translation models on SGET and show that they struggle more in structural generalization than in lexical generalization.We also find different performance trends in semantic parsing and machine translation, which indicates the importance of evaluations across various tasks.",
}
|
Compositional generalization refers to the ability to generalize to novel combinations of previously observed words and syntactic structures.Since it is regarded as a desired property of neural models, recent work has assessed compositional generalization in machine translation as well as semantic parsing.However, previous evaluations with machine translation have focused mostly on lexical generalization (i.e., generalization to unseen combinations of known words).Thus, it remains unclear to what extent models can translate sentences that require structural generalization (i.e., generalization to different sorts of syntactic structures).To address this question, we construct SGET, a machine translation dataset covering various types of compositional generalization with control of words and sentence structures.We evaluate neural machine translation models on SGET and show that they struggle more in structural generalization than in lexical generalization.We also find different performance trends in semantic parsing and machine translation, which indicates the importance of evaluations across various tasks.
|
[
"Kumon, Ryoma",
"Matsuoka, Daiki",
"Yanaka, Hitomi"
] |
Evaluating Structural Generalization in Neural Machine Translation
|
findings-acl.783
|
Poster
|
2406.13363v1
|
https://aclanthology.org/2024.findings-acl.784.bib
|
@inproceedings{katsios-etal-2024-figuratively,
title = "Figuratively Speaking: Authorship Attribution via Multi-Task Figurative Language Modeling",
author = "Katsios, Gregorios and
Sa, Ning and
Strzalkowski, Tomek",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.784",
pages = "13240--13255",
abstract = "The identification of Figurative Language (FL) features in text is crucial for various Natural Language Processing (NLP) tasks, where understanding of the author{'}s intended meaning and its nuances is key for successful communication. At the same time, the use of a specific blend of various FL forms most accurately reflects a writer{'}s style, rather than the use of any single construct, such as just metaphors or irony. Thus, we postulate that FL features could play an important role in Authorship Attribution (AA) tasks. We believe that our is the first computational study of AA based on FL use. Accordingly, we propose a Multi-task Figurative Language Model (MFLM) that learns to detect multiple FL features in text at once. We demonstrate, through detailed evaluation across multiple test sets, that the our model tends to perform equally or outperform specialized binary models in FL detection. Subsequently, we evaluate the predictive capability of joint FL features towards the AA task on three datasets, observing improved AA performance through the integration of MFLM embeddings.",
}
|
The identification of Figurative Language (FL) features in text is crucial for various Natural Language Processing (NLP) tasks, where understanding of the author{'}s intended meaning and its nuances is key for successful communication. At the same time, the use of a specific blend of various FL forms most accurately reflects a writer{'}s style, rather than the use of any single construct, such as just metaphors or irony. Thus, we postulate that FL features could play an important role in Authorship Attribution (AA) tasks. We believe that our is the first computational study of AA based on FL use. Accordingly, we propose a Multi-task Figurative Language Model (MFLM) that learns to detect multiple FL features in text at once. We demonstrate, through detailed evaluation across multiple test sets, that the our model tends to perform equally or outperform specialized binary models in FL detection. Subsequently, we evaluate the predictive capability of joint FL features towards the AA task on three datasets, observing improved AA performance through the integration of MFLM embeddings.
|
[
"Katsios, Gregorios",
"Sa, Ning",
"Strzalkowski, Tomek"
] |
Figuratively Speaking: Authorship Attribution via Multi-Task Figurative Language Modeling
|
findings-acl.784
|
Poster
|
2406.12925v2
|
https://aclanthology.org/2024.findings-acl.785.bib
|
@inproceedings{mao-etal-2024-champ,
title = "{CHAMP}: A Competition-level Dataset for Fine-Grained Analyses of {LLM}s{'} Mathematical Reasoning Capabilities",
author = "Mao, Yujun and
Kim, Yoon and
Zhou, Yilun",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.785",
pages = "13256--13274",
abstract = "Recent large language models (LLMs) have shown indications of mathematical reasoning ability on challenging competition-level problems, especially with self-generated verbalizations of intermediate reasoning steps (i.e., chain-of-thought prompting). However, current evaluations mainly focus on the end-to-end final answer correctness, and it is unclear whether LLMs can make use of helpful side information such as problem-specific hints. In this paper, we propose a challenging benchmark dataset for enabling such analyses. The Concept and Hint-Annotated Math Problems (CHAMP) consists of high school math competition problems, annotated with concepts, or general math facts, and hints, or problem-specific tricks. These annotations allow us to explore the effects of additional information, such as relevant hints, misleading concepts, or related problems. This benchmark is difficult, with the best model only scoring 58.1{\%} in standard settings. With concepts and hints, performance sometimes improves, indicating that some models can make use of such side information. Furthermore, we annotate model-generated solutions for their correctness. Using this corpus, we find that models often arrive at the correct final answer through wrong reasoning steps. In addition, we test whether models are able to verify these solutions, and find that most models struggle.",
}
|
Recent large language models (LLMs) have shown indications of mathematical reasoning ability on challenging competition-level problems, especially with self-generated verbalizations of intermediate reasoning steps (i.e., chain-of-thought prompting). However, current evaluations mainly focus on the end-to-end final answer correctness, and it is unclear whether LLMs can make use of helpful side information such as problem-specific hints. In this paper, we propose a challenging benchmark dataset for enabling such analyses. The Concept and Hint-Annotated Math Problems (CHAMP) consists of high school math competition problems, annotated with concepts, or general math facts, and hints, or problem-specific tricks. These annotations allow us to explore the effects of additional information, such as relevant hints, misleading concepts, or related problems. This benchmark is difficult, with the best model only scoring 58.1{\%} in standard settings. With concepts and hints, performance sometimes improves, indicating that some models can make use of such side information. Furthermore, we annotate model-generated solutions for their correctness. Using this corpus, we find that models often arrive at the correct final answer through wrong reasoning steps. In addition, we test whether models are able to verify these solutions, and find that most models struggle.
|
[
"Mao, Yujun",
"Kim, Yoon",
"Zhou, Yilun"
] |
{CHAMP}: A Competition-level Dataset for Fine-Grained Analyses of {LLM}s{'} Mathematical Reasoning Capabilities
|
findings-acl.785
|
Poster
|
2403.07747v1
|
https://aclanthology.org/2024.findings-acl.786.bib
|
@inproceedings{zeng-etal-2024-improving,
title = "Improving Machine Translation with Large Language Models: A Preliminary Study with Cooperative Decoding",
author = "Zeng, Jiali and
Meng, Fandong and
Yin, Yongjing and
Zhou, Jie",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.786",
pages = "13275--13288",
abstract = "Contemporary translation engines based on the encoder-decoder framework have made significant strides in development.However, the emergence of Large Language Models (LLMs) has disrupted their position by presenting the potential for achieving superior translation quality.To uncover the circumstances in which LLMs excel and explore how their strengths can be harnessed to enhance translation quality,we first conduct a comprehensive analysis to assess the strengths and limitations of various commercial NMT systems and MT-oriented LLMs. Our findings indicate that neither NMT nor MT-oriented LLMs alone can effectively address all the translation issues, but MT-oriented LLMs show promise as a complementary solution to NMT systems.Building upon these insights, we propose Cooperative Decoding (CoDec), which treats NMT systems as a pretranslation model and MT-oriented LLMs as a supplemental solution to handle complex scenarios beyond the capability of NMT alone.Experimental results on the WMT22 test sets and a newly collected test set WebCrawl demonstrate the effectiveness and efficiency of CoDec, highlighting its potential as a robust solution for combining NMT systems with MT-oriented LLMs in the field of machine translation.",
}
|
Contemporary translation engines based on the encoder-decoder framework have made significant strides in development.However, the emergence of Large Language Models (LLMs) has disrupted their position by presenting the potential for achieving superior translation quality.To uncover the circumstances in which LLMs excel and explore how their strengths can be harnessed to enhance translation quality,we first conduct a comprehensive analysis to assess the strengths and limitations of various commercial NMT systems and MT-oriented LLMs. Our findings indicate that neither NMT nor MT-oriented LLMs alone can effectively address all the translation issues, but MT-oriented LLMs show promise as a complementary solution to NMT systems.Building upon these insights, we propose Cooperative Decoding (CoDec), which treats NMT systems as a pretranslation model and MT-oriented LLMs as a supplemental solution to handle complex scenarios beyond the capability of NMT alone.Experimental results on the WMT22 test sets and a newly collected test set WebCrawl demonstrate the effectiveness and efficiency of CoDec, highlighting its potential as a robust solution for combining NMT systems with MT-oriented LLMs in the field of machine translation.
|
[
"Zeng, Jiali",
"Meng, F",
"ong",
"Yin, Yongjing",
"Zhou, Jie"
] |
Improving Machine Translation with Large Language Models: A Preliminary Study with Cooperative Decoding
|
findings-acl.786
|
Poster
|
2311.02851v2
|
https://aclanthology.org/2024.findings-acl.787.bib
|
@inproceedings{hono-etal-2024-integrating,
title = "Integrating Pre-Trained Speech and Language Models for End-to-End Speech Recognition",
author = "Hono, Yukiya and
Mitsuda, Koh and
Zhao, Tianyu and
Mitsui, Kentaro and
Wakatsuki, Toshiaki and
Sawada, Kei",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.787",
pages = "13289--13305",
abstract = "Advances in machine learning have made it possible to perform various text and speech processing tasks, such as automatic speech recognition (ASR), in an end-to-end (E2E) manner. E2E approaches utilizing pre-trained models are gaining attention for conserving training data and resources. However, most of their applications in ASR involve only one of either a pre-trained speech or a language model. This paper proposes integrating a pre-trained speech representation model and a large language model (LLM) for E2E ASR. The proposed model enables the optimization of the entire ASR process, including acoustic feature extraction and acoustic and language modeling, by combining pre-trained models with a bridge network and also enables the application of remarkable developments in LLM utilization, such as parameter-efficient domain adaptation and inference optimization. Experimental results demonstrate that the proposed model achieves a performance comparable to that of modern E2E ASR models by utilizing powerful pre-training models with the proposed integrated approach.",
}
|
Advances in machine learning have made it possible to perform various text and speech processing tasks, such as automatic speech recognition (ASR), in an end-to-end (E2E) manner. E2E approaches utilizing pre-trained models are gaining attention for conserving training data and resources. However, most of their applications in ASR involve only one of either a pre-trained speech or a language model. This paper proposes integrating a pre-trained speech representation model and a large language model (LLM) for E2E ASR. The proposed model enables the optimization of the entire ASR process, including acoustic feature extraction and acoustic and language modeling, by combining pre-trained models with a bridge network and also enables the application of remarkable developments in LLM utilization, such as parameter-efficient domain adaptation and inference optimization. Experimental results demonstrate that the proposed model achieves a performance comparable to that of modern E2E ASR models by utilizing powerful pre-training models with the proposed integrated approach.
|
[
"Hono, Yukiya",
"Mitsuda, Koh",
"Zhao, Tianyu",
"Mitsui, Kentaro",
"Wakatsuki, Toshiaki",
"Sawada, Kei"
] |
Integrating Pre-Trained Speech and Language Models for End-to-End Speech Recognition
|
findings-acl.787
|
Poster
|
0206037v1
|
https://aclanthology.org/2024.findings-acl.788.bib
|
@inproceedings{wei-etal-2024-proving,
title = "Proving membership in {LLM} pretraining data via data watermarks",
author = "Wei, Johnny and
Wang, Ryan and
Jia, Robin",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.788",
pages = "13306--13320",
abstract = "Detecting whether copyright holders{'} works were used in LLM pretraining is poised to be an important problem. This work proposes using data watermarks to enable principled detection with only black-box model access, provided that the rightholder contributed multiple training documents and watermarked them before public release. By applying a randomly sampled data watermark, detection can be framed as hypothesis testing, which provides guarantees on the false detection rate. We study two watermarks: one that inserts random sequences, and another that randomly substitutes characters with Unicode lookalikes. We first show how three aspects of watermark design - watermark length, number of duplications, and interference - affect the power of the hypothesis test. Next, we study how a watermark{'}s detection strength changes under model and dataset scaling: while increasing the dataset size decreases the strength of the watermark, watermarks remain strong if the model size also increases. Finally, we view SHA hashes as natural watermarks and show that we can robustly detect hashes from BLOOM-176B{'}s training data, as long as they occurred at least 90 times. Together, our results point towards a promising future for data watermarks in real world use.",
}
|
Detecting whether copyright holders{'} works were used in LLM pretraining is poised to be an important problem. This work proposes using data watermarks to enable principled detection with only black-box model access, provided that the rightholder contributed multiple training documents and watermarked them before public release. By applying a randomly sampled data watermark, detection can be framed as hypothesis testing, which provides guarantees on the false detection rate. We study two watermarks: one that inserts random sequences, and another that randomly substitutes characters with Unicode lookalikes. We first show how three aspects of watermark design - watermark length, number of duplications, and interference - affect the power of the hypothesis test. Next, we study how a watermark{'}s detection strength changes under model and dataset scaling: while increasing the dataset size decreases the strength of the watermark, watermarks remain strong if the model size also increases. Finally, we view SHA hashes as natural watermarks and show that we can robustly detect hashes from BLOOM-176B{'}s training data, as long as they occurred at least 90 times. Together, our results point towards a promising future for data watermarks in real world use.
|
[
"Wei, Johnny",
"Wang, Ryan",
"Jia, Robin"
] |
Proving membership in {LLM} pretraining data via data watermarks
|
findings-acl.788
|
Poster
|
2402.10892v2
|
https://aclanthology.org/2024.findings-acl.789.bib
|
@inproceedings{zhang-etal-2024-enhancing-hallucination,
title = "Enhancing Hallucination Detection through Perturbation-Based Synthetic Data Generation in System Responses",
author = "Zhang, Dongxu and
Gangal, Varun and
Lattimer, Barrett and
Yang, Yi",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.789",
pages = "13321--13332",
abstract = "Detecting hallucinations in large language model (LLM) outputs is pivotal, yet traditional fine-tuning for this classification task is impeded by the expensive and quickly outdated annotation process, especially across numerous vertical domains and in the face of rapid LLM advancements. In this study, we introduce an approach that automatically generates both faithful and hallucinated outputs by rewriting system responses. Experimental findings demonstrate that a T5-base model, fine-tuned on our generated dataset, surpasses state-of-the-art zero-shot detectors and existing synthetic generation methods in both accuracy and latency, indicating efficacy of our approach.",
}
|
Detecting hallucinations in large language model (LLM) outputs is pivotal, yet traditional fine-tuning for this classification task is impeded by the expensive and quickly outdated annotation process, especially across numerous vertical domains and in the face of rapid LLM advancements. In this study, we introduce an approach that automatically generates both faithful and hallucinated outputs by rewriting system responses. Experimental findings demonstrate that a T5-base model, fine-tuned on our generated dataset, surpasses state-of-the-art zero-shot detectors and existing synthetic generation methods in both accuracy and latency, indicating efficacy of our approach.
|
[
"Zhang, Dongxu",
"Gangal, Varun",
"Lattimer, Barrett",
"Yang, Yi"
] |
Enhancing Hallucination Detection through Perturbation-Based Synthetic Data Generation in System Responses
|
findings-acl.789
|
Poster
|
2407.05474v1
|
https://aclanthology.org/2024.findings-acl.790.bib
|
@inproceedings{luo-etal-2024-secformer,
title = "{S}ec{F}ormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via {SMPC}",
author = "Luo, Jinglong and
Zhang, Yehong and
Zhang, Zhuo and
Zhang, Jiaqi and
Mu, Xin and
Wang, Hui and
Yu, Yue and
Xu, Zenglin",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.790",
pages = "13333--13348",
abstract = "With the growing use of Transformer models hosted on cloud platforms to offer inference services, privacy concerns are escalating, especially concerning sensitive data like investment plans and bank account details. Secure Multi-Party Computing (SMPC) emerges as a promising solution to protect the privacy of inference data and model parameters. However, the application of SMPC in Privacy-Preserving Inference (PPI) for Transformer models often leads to considerable slowdowns or declines in performance. This is largely due to the multitude of nonlinear operations in the Transformer architecture, which are not well-suited to SMPC and are difficult to circumvent or optimize effectively. To address this concern, we introduce a comprehensive PPI framework called SecFormer to achieve fast and accurate PPI for Transformer models. We successfully eliminate the high-cost exponential and maximum operations in PPI without sacrificing model performance and develop a suite of efficient SMPC protocols by employing suitable numerical computation methods to boost other complex nonlinear functions in PPI, including GeLU, LayerNorm, and a redesigned Softmax. Our extensive experiments reveal that SecFormer outperforms MPCFormer in performance, showing improvements of 3.4{\%} and 24.7{\%} for $BERT_{\text{BASE}}$ and $BERT_{\text{LARGE}}$, respectively. In terms of efficiency, SecFormer is 3.57 and 3.58 times faster than PUMA for $BERT_{\text{BASE}}$ and $BERT_{\text{LARGE}}$, demonstrating its effectiveness and speed.",
}
|
With the growing use of Transformer models hosted on cloud platforms to offer inference services, privacy concerns are escalating, especially concerning sensitive data like investment plans and bank account details. Secure Multi-Party Computing (SMPC) emerges as a promising solution to protect the privacy of inference data and model parameters. However, the application of SMPC in Privacy-Preserving Inference (PPI) for Transformer models often leads to considerable slowdowns or declines in performance. This is largely due to the multitude of nonlinear operations in the Transformer architecture, which are not well-suited to SMPC and are difficult to circumvent or optimize effectively. To address this concern, we introduce a comprehensive PPI framework called SecFormer to achieve fast and accurate PPI for Transformer models. We successfully eliminate the high-cost exponential and maximum operations in PPI without sacrificing model performance and develop a suite of efficient SMPC protocols by employing suitable numerical computation methods to boost other complex nonlinear functions in PPI, including GeLU, LayerNorm, and a redesigned Softmax. Our extensive experiments reveal that SecFormer outperforms MPCFormer in performance, showing improvements of 3.4{\%} and 24.7{\%} for $BERT_{\text{BASE}}$ and $BERT_{\text{LARGE}}$, respectively. In terms of efficiency, SecFormer is 3.57 and 3.58 times faster than PUMA for $BERT_{\text{BASE}}$ and $BERT_{\text{LARGE}}$, demonstrating its effectiveness and speed.
|
[
"Luo, Jinglong",
"Zhang, Yehong",
"Zhang, Zhuo",
"Zhang, Jiaqi",
"Mu, Xin",
"Wang, Hui",
"Yu, Yue",
"Xu, Zenglin"
] |
{S}ec{F}ormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via {SMPC}
|
findings-acl.790
|
Poster
|
2401.00793v3
|
https://aclanthology.org/2024.findings-acl.791.bib
|
@inproceedings{wang-etal-2024-raccoon,
title = "Raccoon: Prompt Extraction Benchmark of {LLM}-Integrated Applications",
author = "Wang, Junlin and
Yang, Tianyi and
Xie, Roy and
Dhingra, Bhuwan",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.791",
pages = "13349--13365",
abstract = "With the proliferation of LLM-integrated applications such as GPT-s, millions are deployed, offering valuable services through proprietary instruction prompts. These systems, however, are prone to prompt extraction attacks through meticulously designed queries. To help mitigate this problem, we introduce the Raccoon benchmark which comprehensively evaluates a model{'}s susceptibility to prompt extraction attacks. Our novel evaluation method assesses models under both defenseless and defended scenarios, employing a dual approach to evaluate the effectiveness of existing defenses and the resilience of the models. The benchmark encompasses 14 categories of prompt extraction attacks, with additional compounded attacks that closely mimic the strategies of potential attackers, alongside a diverse collection of defense templates. This array is, to our knowledge, the most extensive compilation of prompt theft attacks and defense mechanisms to date. Our findings highlight universal susceptibility to prompt theft in the absence of defenses, with OpenAI models demonstrating notable resilience when protected. This paper aims to establish a more systematic benchmark for assessing LLM robustness against prompt extraction attacks, offering insights into their causes and potential countermeasures.",
}
|
With the proliferation of LLM-integrated applications such as GPT-s, millions are deployed, offering valuable services through proprietary instruction prompts. These systems, however, are prone to prompt extraction attacks through meticulously designed queries. To help mitigate this problem, we introduce the Raccoon benchmark which comprehensively evaluates a model{'}s susceptibility to prompt extraction attacks. Our novel evaluation method assesses models under both defenseless and defended scenarios, employing a dual approach to evaluate the effectiveness of existing defenses and the resilience of the models. The benchmark encompasses 14 categories of prompt extraction attacks, with additional compounded attacks that closely mimic the strategies of potential attackers, alongside a diverse collection of defense templates. This array is, to our knowledge, the most extensive compilation of prompt theft attacks and defense mechanisms to date. Our findings highlight universal susceptibility to prompt theft in the absence of defenses, with OpenAI models demonstrating notable resilience when protected. This paper aims to establish a more systematic benchmark for assessing LLM robustness against prompt extraction attacks, offering insights into their causes and potential countermeasures.
|
[
"Wang, Junlin",
"Yang, Tianyi",
"Xie, Roy",
"Dhingra, Bhuwan"
] |
Raccoon: Prompt Extraction Benchmark of {LLM}-Integrated Applications
|
findings-acl.791
|
Poster
|
2406.06737v1
|
https://aclanthology.org/2024.findings-acl.792.bib
|
@inproceedings{mo-etal-2024-history,
title = "History-Aware Conversational Dense Retrieval",
author = "Mo, Fengran and
Qu, Chen and
Mao, Kelong and
Zhu, Tianyu and
Su, Zhan and
Huang, Kaiyu and
Nie, Jian-Yun",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.792",
pages = "13366--13378",
abstract = "Conversational search facilitates complex information retrieval by enabling multi-turn interactions between users and the system. Supporting such interactions requires a comprehensive understanding of the conversational inputs to formulate a good search query based on historical information. In particular, the search query should include the relevant information from the previous conversation turns.However, current approaches for conversational dense retrieval primarily rely on fine-tuning a pre-trained ad-hoc retriever using the whole conversational search session, which can be lengthy and noisy. Moreover, existing approaches are limited by the amount of manual supervision signals in the existing datasets.To address the aforementioned issues, we propose a **H**istory-**A**ware **Conv**ersational **D**ense **R**etrieval (HAConvDR) system, which incorporates two ideas: context-denoised query reformulation and automatic mining of supervision signals based on the actual impact of historical turns.Experiments on two public conversational search datasets demonstrate the improved history modeling capability of HAConvDR, in particular for long conversations with topic shifts.",
}
|
Conversational search facilitates complex information retrieval by enabling multi-turn interactions between users and the system. Supporting such interactions requires a comprehensive understanding of the conversational inputs to formulate a good search query based on historical information. In particular, the search query should include the relevant information from the previous conversation turns.However, current approaches for conversational dense retrieval primarily rely on fine-tuning a pre-trained ad-hoc retriever using the whole conversational search session, which can be lengthy and noisy. Moreover, existing approaches are limited by the amount of manual supervision signals in the existing datasets.To address the aforementioned issues, we propose a **H**istory-**A**ware **Conv**ersational **D**ense **R**etrieval (HAConvDR) system, which incorporates two ideas: context-denoised query reformulation and automatic mining of supervision signals based on the actual impact of historical turns.Experiments on two public conversational search datasets demonstrate the improved history modeling capability of HAConvDR, in particular for long conversations with topic shifts.
|
[
"Mo, Fengran",
"Qu, Chen",
"Mao, Kelong",
"Zhu, Tianyu",
"Su, Zhan",
"Huang, Kaiyu",
"Nie, Jian-Yun"
] |
History-Aware Conversational Dense Retrieval
|
findings-acl.792
|
Poster
|
2309.06748v1
|
https://aclanthology.org/2024.findings-acl.793.bib
|
@inproceedings{zhang-etal-2024-light,
title = "Light Up the Shadows: Enhance Long-Tailed Entity Grounding with Concept-Guided Vision-Language Models",
author = "Zhang, Yikai and
He, Qianyu and
Wang, Xintao and
Yuan, Siyu and
Liang, Jiaqing and
Xiao, Yanghua",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.793",
pages = "13379--13389",
abstract = "Multi-Modal Knowledge Graphs (MMKGs) have proven valuable for various downstream tasks. However, scaling them up is challenging because building large-scale MMKGs often introduces mismatched images (i.e., noise). Most entities in KGs belong to the long tail, meaning there are few images of them available online. This scarcity makes it difficult to determine whether a found image matches the entity. To address this, we draw on the Triangle of Reference Theory and suggest enhancing vision-language models with concept guidance. Specifically, we introduce COG, a two-stage framework with COncept-Guided vision-language models. The framework comprises a Concept Integration module, which effectively identifies image-text pairs of long-tailed entities, and an Evidence Fusion module, which offers explainability and enables human verification. To demonstrate the effectiveness of COG, we create a dataset of 25k image-text pairs of long-tailed entities. Our comprehensive experiments show that COG not only improves the accuracy of recognizing long-tailed image-text pairs compared to baselines but also offers flexibility and explainability.",
}
|
Multi-Modal Knowledge Graphs (MMKGs) have proven valuable for various downstream tasks. However, scaling them up is challenging because building large-scale MMKGs often introduces mismatched images (i.e., noise). Most entities in KGs belong to the long tail, meaning there are few images of them available online. This scarcity makes it difficult to determine whether a found image matches the entity. To address this, we draw on the Triangle of Reference Theory and suggest enhancing vision-language models with concept guidance. Specifically, we introduce COG, a two-stage framework with COncept-Guided vision-language models. The framework comprises a Concept Integration module, which effectively identifies image-text pairs of long-tailed entities, and an Evidence Fusion module, which offers explainability and enables human verification. To demonstrate the effectiveness of COG, we create a dataset of 25k image-text pairs of long-tailed entities. Our comprehensive experiments show that COG not only improves the accuracy of recognizing long-tailed image-text pairs compared to baselines but also offers flexibility and explainability.
|
[
"Zhang, Yikai",
"He, Qianyu",
"Wang, Xintao",
"Yuan, Siyu",
"Liang, Jiaqing",
"Xiao, Yanghua"
] |
Light Up the Shadows: Enhance Long-Tailed Entity Grounding with Concept-Guided Vision-Language Models
|
findings-acl.793
|
Poster
|
2406.10902v1
|
https://aclanthology.org/2024.findings-acl.794.bib
|
@inproceedings{zhao-etal-2024-zerostance,
title = "{Z}ero{S}tance: Leveraging {C}hat{GPT} for Open-Domain Stance Detection via Dataset Generation",
author = "Zhao, Chenye and
Li, Yingjie and
Caragea, Cornelia and
Zhang, Yue",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.794",
pages = "13390--13405",
abstract = "Zero-shot stance detection that aims to detect the stance (typically against, favor, or neutral) towards unseen targets has attracted considerable attention. However, most previous studies only focus on targets from a single or limited text domains (e.g., financial domain), and thus zero-shot models cannot generalize well to unseen targets of diverse domains (e.g., political domain). In this paper, we consider a more realistic task, i.e., open-domain stance detection, which aims at training a model that is able to generalize well to unseen targets across multiple domains of interest. Particularly, we propose a novel dataset generation method ZeroStance, which leverages ChatGPT to construct a synthetic open-domain dataset CHATStance that covers a wide range of domains. We then train an open-domain model on our synthetic dataset after proper data filtering. Extensive results indicate that our model, when trained on this synthetic dataset, shows superior generalization to unseen targets of diverse domains over baselines on most benchmarks. Our method requires only a task description in the form of a prompt and is much more cost-effective and data-efficient than previous methods. We will release our code and data to facilitate future research.",
}
|
Zero-shot stance detection that aims to detect the stance (typically against, favor, or neutral) towards unseen targets has attracted considerable attention. However, most previous studies only focus on targets from a single or limited text domains (e.g., financial domain), and thus zero-shot models cannot generalize well to unseen targets of diverse domains (e.g., political domain). In this paper, we consider a more realistic task, i.e., open-domain stance detection, which aims at training a model that is able to generalize well to unseen targets across multiple domains of interest. Particularly, we propose a novel dataset generation method ZeroStance, which leverages ChatGPT to construct a synthetic open-domain dataset CHATStance that covers a wide range of domains. We then train an open-domain model on our synthetic dataset after proper data filtering. Extensive results indicate that our model, when trained on this synthetic dataset, shows superior generalization to unseen targets of diverse domains over baselines on most benchmarks. Our method requires only a task description in the form of a prompt and is much more cost-effective and data-efficient than previous methods. We will release our code and data to facilitate future research.
|
[
"Zhao, Chenye",
"Li, Yingjie",
"Caragea, Cornelia",
"Zhang, Yue"
] |
{Z}ero{S}tance: Leveraging {C}hat{GPT} for Open-Domain Stance Detection via Dataset Generation
|
findings-acl.794
|
Poster
|
2404.12171v1
|
https://aclanthology.org/2024.findings-acl.795.bib
|
@inproceedings{fazili-etal-2024-boosting,
title = "Boosting Zero-Shot Crosslingual Performance using {LLM}-Based Augmentations with Effective Data Selection",
author = "Fazili, Barah and
Agrawal, Ashish and
Jyothi, Preethi",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.795",
pages = "13406--13422",
abstract = "Large language models (LLMs) are very proficient text generators. We leverage this capability of LLMs to generate task-specific data via zero-shot prompting and promote cross-lingual transfer for low-resource target languages. Given task-specific data in a source language and a teacher model trained on this data, we propose using this teacher to label LLM generations and employ a set of simple data selection strategies that use the teacher{'}s label probabilities. Our data selection strategies help us identify a representative subset of diverse generations that help boost zero-shot accuracies while being efficient, in comparison to using all the LLM generations (without any subset selection). We also highlight other important design choices that affect cross-lingual performance such as the use of translations of source data and what labels are best to use for the LLM generations. We observe significant performance gains across sentiment analysis and natural language inference tasks (of up to a maximum of 7.13 absolute points and 1.5 absolute points on average) across a number of target languages (Hindi, Marathi, Urdu, Swahili) and domains.",
}
|
Large language models (LLMs) are very proficient text generators. We leverage this capability of LLMs to generate task-specific data via zero-shot prompting and promote cross-lingual transfer for low-resource target languages. Given task-specific data in a source language and a teacher model trained on this data, we propose using this teacher to label LLM generations and employ a set of simple data selection strategies that use the teacher{'}s label probabilities. Our data selection strategies help us identify a representative subset of diverse generations that help boost zero-shot accuracies while being efficient, in comparison to using all the LLM generations (without any subset selection). We also highlight other important design choices that affect cross-lingual performance such as the use of translations of source data and what labels are best to use for the LLM generations. We observe significant performance gains across sentiment analysis and natural language inference tasks (of up to a maximum of 7.13 absolute points and 1.5 absolute points on average) across a number of target languages (Hindi, Marathi, Urdu, Swahili) and domains.
|
[
"Fazili, Barah",
"Agrawal, Ashish",
"Jyothi, Preethi"
] |
Boosting Zero-Shot Crosslingual Performance using {LLM}-Based Augmentations with Effective Data Selection
|
findings-acl.795
|
Poster
|
2407.10582v1
|
https://aclanthology.org/2024.findings-acl.796.bib
|
@inproceedings{yang-etal-2024-reinforcement,
title = "Reinforcement Tuning for Detecting Stances and Debunking Rumors Jointly with Large Language Models",
author = "Yang, Ruichao and
Gao, Wei and
Ma, Jing and
Lin, Hongzhan and
Wang, Bo",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.796",
pages = "13423--13439",
abstract = "Learning multi-task models for jointly detecting stance and verifying rumors poses challenges due to the need for training data of stance at post level and rumor veracity at claim level, which are difficult to obtain. To address this issue, we leverage large language models (LLMs) as the foundation annotators for the joint stance detection (SD) and rumor verification (RV) tasks, dubbed as JSDRV. We introduce a novel reinforcement tuning framework to enhance the joint predictive capabilities of LLM-based SD and RV components. Specifically, we devise a policy for selecting LLM-annotated data at the two levels, employing a hybrid reward mechanism to choose high-quality labels for effective LLM fine-tuning on both tasks. Results demonstrate that JSDRV improves the capabilities of LLMs in the joint tasks, not only outperforming state-of-the-art methods but also generalizing to non-LLMs accommodated as task models.",
}
|
Learning multi-task models for jointly detecting stance and verifying rumors poses challenges due to the need for training data of stance at post level and rumor veracity at claim level, which are difficult to obtain. To address this issue, we leverage large language models (LLMs) as the foundation annotators for the joint stance detection (SD) and rumor verification (RV) tasks, dubbed as JSDRV. We introduce a novel reinforcement tuning framework to enhance the joint predictive capabilities of LLM-based SD and RV components. Specifically, we devise a policy for selecting LLM-annotated data at the two levels, employing a hybrid reward mechanism to choose high-quality labels for effective LLM fine-tuning on both tasks. Results demonstrate that JSDRV improves the capabilities of LLMs in the joint tasks, not only outperforming state-of-the-art methods but also generalizing to non-LLMs accommodated as task models.
|
[
"Yang, Ruichao",
"Gao, Wei",
"Ma, Jing",
"Lin, Hongzhan",
"Wang, Bo"
] |
Reinforcement Tuning for Detecting Stances and Debunking Rumors Jointly with Large Language Models
|
findings-acl.796
|
Poster
|
2406.02143v1
|
https://aclanthology.org/2024.findings-acl.797.bib
|
@inproceedings{fan-etal-2024-exploring,
title = "Exploring the Potential of Dense Information in Multimodal Alignment",
author = "Fan, Zhiyuan and
Chen, Zhihong and
Wang, Benyou",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.797",
pages = "13440--13451",
abstract = "Despite the success of data augmentation in improving CLIP model, existing methods that utilize LLM or SAM to enrich the information in captions still suffer from several limitations, including insufficient detail and excessive hallucinations, ultimately resulting in compromised alignment and masking the true potential of dense information. This can lead to erroneous conclusions about CLIP{'}s ability to handle rich data, impeding the development of more effective models. To address the limitations of existing methods, we introduce a novel pipeline that generates highly detailed, factually accurate captions for images, which facilitates in-depth analysis of the potential for dense information in multimodal alignment. Contrary to previous findings, our investigation revealed that lengthening captions boosts performance across diverse benchmarks, even surpassing the effectiveness of meticulously crafted hard negative samples. Building on these insights, DELIP is introduced, demonstrably enhancing both foundational multimodal alignment and compositional reasoning abilities. Finally, we explore strategies to expand the context window of the text encoder, unlocking the potential of richer data for CLIP and paving the way for advancements in leveraging dense information for multimodal alignment.",
}
|
Despite the success of data augmentation in improving CLIP model, existing methods that utilize LLM or SAM to enrich the information in captions still suffer from several limitations, including insufficient detail and excessive hallucinations, ultimately resulting in compromised alignment and masking the true potential of dense information. This can lead to erroneous conclusions about CLIP{'}s ability to handle rich data, impeding the development of more effective models. To address the limitations of existing methods, we introduce a novel pipeline that generates highly detailed, factually accurate captions for images, which facilitates in-depth analysis of the potential for dense information in multimodal alignment. Contrary to previous findings, our investigation revealed that lengthening captions boosts performance across diverse benchmarks, even surpassing the effectiveness of meticulously crafted hard negative samples. Building on these insights, DELIP is introduced, demonstrably enhancing both foundational multimodal alignment and compositional reasoning abilities. Finally, we explore strategies to expand the context window of the text encoder, unlocking the potential of richer data for CLIP and paving the way for advancements in leveraging dense information for multimodal alignment.
|
[
"Fan, Zhiyuan",
"Chen, Zhihong",
"Wang, Benyou"
] |
Exploring the Potential of Dense Information in Multimodal Alignment
|
findings-acl.797
|
Poster
|
2311.07594v2
|
https://aclanthology.org/2024.findings-acl.798.bib
|
@inproceedings{tang-etal-2024-referral,
title = "Referral Augmentation for Zero-Shot Information Retrieval",
author = "Tang, Michael and
Yao, Shunyu and
Yang, John and
Narasimhan, Karthik",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.798",
pages = "13452--13461",
abstract = "We propose Referral-Augmented Retrieval (RAR), a simple technique that concatenates document indices with referrals: text from other documents that cite or link to the given document. We find that RAR provides significant performance gains for tasks across paper retrieval, entity retrieval, and open-domain question-answering in both zero-shot and in-domain (e.g., fine-tuned) settings. We examine how RAR provides especially strong improvements on more structured tasks, and can greatly outperform generative text expansion techniques such as DocT5Query and Query2Doc, with a 37{\%} and 21{\%} absolute improvement on ACL paper retrieval, respectively. We also compare three ways to aggregate referrals for RAR. Overall, we believe RAR can help revive and re-contextualize the classic information retrieval idea of using anchor texts to improve the representations of documents in a wide variety of corpuses in the age of neural retrieval.",
}
|
We propose Referral-Augmented Retrieval (RAR), a simple technique that concatenates document indices with referrals: text from other documents that cite or link to the given document. We find that RAR provides significant performance gains for tasks across paper retrieval, entity retrieval, and open-domain question-answering in both zero-shot and in-domain (e.g., fine-tuned) settings. We examine how RAR provides especially strong improvements on more structured tasks, and can greatly outperform generative text expansion techniques such as DocT5Query and Query2Doc, with a 37{\%} and 21{\%} absolute improvement on ACL paper retrieval, respectively. We also compare three ways to aggregate referrals for RAR. Overall, we believe RAR can help revive and re-contextualize the classic information retrieval idea of using anchor texts to improve the representations of documents in a wide variety of corpuses in the age of neural retrieval.
|
[
"Tang, Michael",
"Yao, Shunyu",
"Yang, John",
"Narasimhan, Karthik"
] |
Referral Augmentation for Zero-Shot Information Retrieval
|
findings-acl.798
|
Poster
|
2305.15098v1
|
https://aclanthology.org/2024.findings-acl.799.bib
|
@inproceedings{wu-etal-2024-instructeval,
title = "{I}nstruct{E}val: Instruction-Tuned Text Evaluator from Human Preference",
author = "Wu, Wenhao and
Li, Wei and
Xiao, Xinyan and
Liu, Jiachen and
Li, Sujian",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.799",
pages = "13462--13474",
abstract = "This paper explores to construct a general text evaluator based on open-source Large Language Models (LLMs), a domain predominantly occupied by commercial counterparts such as GPT-4. Recognizing the limitations of open-source models like Llama in evaluative tasks, we introduce InstructEval, a general multi-aspect text evaluator developed through instruction tuning of open-source LLMs. To overcome the shortage of annotated resources for multi-aspect evaluations, InstructEval combines extensive open Human Preference Modeling (HPM) datasets with a small set of multi-aspect annotated data.This approach not only enhances effectiveness in overall evaluation tasks but also exhibits improved performance in multi-aspect evaluation tasks.As demonstrated by our extensive experiments, InstructEval achieves comparable or superior performance to commercial LLMs like ChatGPT or GPT-4 in terms of both overall and multi-aspect evaluation.",
}
|
This paper explores to construct a general text evaluator based on open-source Large Language Models (LLMs), a domain predominantly occupied by commercial counterparts such as GPT-4. Recognizing the limitations of open-source models like Llama in evaluative tasks, we introduce InstructEval, a general multi-aspect text evaluator developed through instruction tuning of open-source LLMs. To overcome the shortage of annotated resources for multi-aspect evaluations, InstructEval combines extensive open Human Preference Modeling (HPM) datasets with a small set of multi-aspect annotated data.This approach not only enhances effectiveness in overall evaluation tasks but also exhibits improved performance in multi-aspect evaluation tasks.As demonstrated by our extensive experiments, InstructEval achieves comparable or superior performance to commercial LLMs like ChatGPT or GPT-4 in terms of both overall and multi-aspect evaluation.
|
[
"Wu, Wenhao",
"Li, Wei",
"Xiao, Xinyan",
"Liu, Jiachen",
"Li, Sujian"
] |
{I}nstruct{E}val: Instruction-Tuned Text Evaluator from Human Preference
|
findings-acl.799
|
Poster
|
2306.09341v2
|
https://aclanthology.org/2024.findings-acl.800.bib
|
@inproceedings{cuong-etal-2024-curious,
title = "A Curious Case of Searching for the Correlation between Training Data and Adversarial Robustness of Transformer Textual Models",
author = "Cuong, Dang and
Le, Dung and
Le, Thai",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.800",
pages = "13475--13491",
abstract = "Existing works have shown that fine-tuned textual transformer models achieve state-of-the-art prediction performances but are also vulnerable to adversarial text perturbations. Traditional adversarial evaluation is often done \textit{only after} fine-tuning the models and ignoring the training data. In this paper, we want to prove that there is also a strong correlation between training data and model robustness. To this end, we extract 13 different features representing a wide range of input fine-tuning corpora properties and use them to predict the adversarial robustness of the fine-tuned models. Focusing mostly on encoder-only transformer models BERT and RoBERTa with additional results for BART, ELECTRA and GPT2, we provide diverse evidence to support our argument. First, empirical analyses show that (a) extracted features can be used with a lightweight classifier such as Random Forest to effectively predict the attack success rate and (b) features with the most influence on the model robustness have a clear correlation with the robustness. Second, our framework can be used as a fast and effective additional tool for robustness evaluation since it (a) saves 30x-193x runtime compared to the traditional technique, (b) is transferable across models, (c) can be used under adversarial training, and (d) robust to statistical randomness. Our code is publicly available at \url{https://github.com/CaptainCuong/RobustText_ACL2024}.",
}
|
Existing works have shown that fine-tuned textual transformer models achieve state-of-the-art prediction performances but are also vulnerable to adversarial text perturbations. Traditional adversarial evaluation is often done \textit{only after} fine-tuning the models and ignoring the training data. In this paper, we want to prove that there is also a strong correlation between training data and model robustness. To this end, we extract 13 different features representing a wide range of input fine-tuning corpora properties and use them to predict the adversarial robustness of the fine-tuned models. Focusing mostly on encoder-only transformer models BERT and RoBERTa with additional results for BART, ELECTRA and GPT2, we provide diverse evidence to support our argument. First, empirical analyses show that (a) extracted features can be used with a lightweight classifier such as Random Forest to effectively predict the attack success rate and (b) features with the most influence on the model robustness have a clear correlation with the robustness. Second, our framework can be used as a fast and effective additional tool for robustness evaluation since it (a) saves 30x-193x runtime compared to the traditional technique, (b) is transferable across models, (c) can be used under adversarial training, and (d) robust to statistical randomness. Our code is publicly available at \url{https://github.com/CaptainCuong/RobustText_ACL2024}.
|
[
"Cuong, Dang",
"Le, Dung",
"Le, Thai"
] |
A Curious Case of Searching for the Correlation between Training Data and Adversarial Robustness of Transformer Textual Models
|
findings-acl.800
|
Poster
|
2402.11469v2
|
https://aclanthology.org/2024.findings-acl.801.bib
|
@inproceedings{wang-etal-2024-instructgraph,
title = "{I}nstruct{G}raph: Boosting Large Language Models via Graph-centric Instruction Tuning and Preference Alignment",
author = "Wang, Jianing and
Wu, Junda and
Hou, Yupeng and
Liu, Yao and
Gao, Ming and
McAuley, Julian",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.801",
pages = "13492--13510",
abstract = "Do current large language models (LLMs) better solve graph reasoning and generation tasks with parameter updates? In this paper, we propose \textbf{InstructGraph}, a framework that empowers LLMs with the abilities of graph reasoning and generation by instruction tuning and preference alignment. Specifically, we first propose a structured format verbalizer to unify all graph data into a universal code-like format, which can simply represent the graph without any external graph-specific encoders. Furthermore, a graph instruction tuning stage is introduced to guide LLMs in solving graph reasoning and generation tasks. Finally, we identify potential hallucination problems in graph tasks and sample negative instances for preference alignment, the target of which is to enhance the output{'}s reliability of the model. Extensive experiments across multiple graph-centric tasks exhibit that InstructGraph can achieve the best performance and outperform GPT-4 and LLaMA2 by more than 13{\%} and 38{\%}, respectively.",
}
|
Do current large language models (LLMs) better solve graph reasoning and generation tasks with parameter updates? In this paper, we propose \textbf{InstructGraph}, a framework that empowers LLMs with the abilities of graph reasoning and generation by instruction tuning and preference alignment. Specifically, we first propose a structured format verbalizer to unify all graph data into a universal code-like format, which can simply represent the graph without any external graph-specific encoders. Furthermore, a graph instruction tuning stage is introduced to guide LLMs in solving graph reasoning and generation tasks. Finally, we identify potential hallucination problems in graph tasks and sample negative instances for preference alignment, the target of which is to enhance the output{'}s reliability of the model. Extensive experiments across multiple graph-centric tasks exhibit that InstructGraph can achieve the best performance and outperform GPT-4 and LLaMA2 by more than 13{\%} and 38{\%}, respectively.
|
[
"Wang, Jianing",
"Wu, Junda",
"Hou, Yupeng",
"Liu, Yao",
"Gao, Ming",
"McAuley, Julian"
] |
{I}nstruct{G}raph: Boosting Large Language Models via Graph-centric Instruction Tuning and Preference Alignment
|
findings-acl.801
|
Poster
|
2405.00402v1
|
https://aclanthology.org/2024.findings-acl.802.bib
|
@inproceedings{kim-etal-2024-rada,
title = "{R}a{DA}: Retrieval-augmented Web Agent Planning with {LLM}s",
author = "Kim, Minsoo and
Bursztyn, Victor and
Koh, Eunyee and
Guo, Shunan and
Hwang, Seung-won",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.802",
pages = "13511--13525",
abstract = "Agents powered by large language models (LLMs) inherit important limitations, such as the restricted context length, dependency on human-engineered exemplars (e.g., for task decomposition), and insufficient generalization. To address these challenges, we propose RaDA, a novel planning method for Web agents that does not require manual exemplars, efficiently leverages the LLMs{'} context, and enhances generalization. RaDA disentangles planning into two stages: for a new given task, during Retrieval-augmented Task Decomposition (RaD), it decomposes tasks into high-level subtasks; next, during Retrieval-augmented Action Generation (RaA), it traverses the trajectory obtained with RaD to iteratively synthesize actions based on dynamically retrieved exemplars. We compare RaDA with strong baselines covering a broad space of design choices, using both GPT-3.5 and GPT-4 as backbones; and we find consistent improvements over previous SOTA in two challenging benchmarks, CompWoB and Mind2Web, covering settings with different complexities. We show the contributions of RaDA via ablation studies and qualitative analysis; and we discuss the structural benefits of our more compositional design.",
}
|
Agents powered by large language models (LLMs) inherit important limitations, such as the restricted context length, dependency on human-engineered exemplars (e.g., for task decomposition), and insufficient generalization. To address these challenges, we propose RaDA, a novel planning method for Web agents that does not require manual exemplars, efficiently leverages the LLMs{'} context, and enhances generalization. RaDA disentangles planning into two stages: for a new given task, during Retrieval-augmented Task Decomposition (RaD), it decomposes tasks into high-level subtasks; next, during Retrieval-augmented Action Generation (RaA), it traverses the trajectory obtained with RaD to iteratively synthesize actions based on dynamically retrieved exemplars. We compare RaDA with strong baselines covering a broad space of design choices, using both GPT-3.5 and GPT-4 as backbones; and we find consistent improvements over previous SOTA in two challenging benchmarks, CompWoB and Mind2Web, covering settings with different complexities. We show the contributions of RaDA via ablation studies and qualitative analysis; and we discuss the structural benefits of our more compositional design.
|
[
"Kim, Minsoo",
"Bursztyn, Victor",
"Koh, Eunyee",
"Guo, Shunan",
"Hwang, Seung-won"
] |
{R}a{DA}: Retrieval-augmented Web Agent Planning with {LLM}s
|
findings-acl.802
|
Poster
|
2405.04497v2
|
https://aclanthology.org/2024.findings-acl.803.bib
|
@inproceedings{huang-etal-2024-competition,
title = "Competition-Level Problems are Effective {LLM} Evaluators",
author = "Huang, Yiming and
Lin, Zhenghao and
Liu, Xiao and
Gong, Yeyun and
Lu, Shuai and
Lei, Fangyu and
Liang, Yaobo and
Shen, Yelong and
Lin, Chen and
Duan, Nan and
Chen, Weizhu",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.803",
pages = "13526--13544",
abstract = "Large language models (LLMs) have demonstrated impressive reasoning capabilities, yet there is ongoing debate about these abilities and the potential data contamination problem recently. This paper aims to evaluate the reasoning capacities of LLMs, specifically in solving recent competition-level programming problems in Codeforces, which are expert-crafted and unique, requiring deep understanding and robust reasoning skills. We first provide a comprehensive evaluation of GPT-4{'}s perceived zero-shot performance on this task, considering various aspects such as problems{'} release time, difficulties, and types of errors encountered. Surprisingly, the perceived performance of GPT-4 has experienced a cliff like decline in problems after September 2021 consistently across all the difficulties and types of problems, which shows the potential data contamination, as well as the challenges for any existing LLM to solve unseen complex reasoning problems. We further explore various approaches such as fine-tuning, Chain-of-Thought prompting and problem description simplification. Unfortunately, none of them is able to consistently mitigate the challenges. Through our work, we emphasize the importance of this excellent data source for assessing the genuine reasoning capabilities of LLMs, and foster the development of LLMs with stronger reasoning abilities and better generalization in the future.",
}
|
Large language models (LLMs) have demonstrated impressive reasoning capabilities, yet there is ongoing debate about these abilities and the potential data contamination problem recently. This paper aims to evaluate the reasoning capacities of LLMs, specifically in solving recent competition-level programming problems in Codeforces, which are expert-crafted and unique, requiring deep understanding and robust reasoning skills. We first provide a comprehensive evaluation of GPT-4{'}s perceived zero-shot performance on this task, considering various aspects such as problems{'} release time, difficulties, and types of errors encountered. Surprisingly, the perceived performance of GPT-4 has experienced a cliff like decline in problems after September 2021 consistently across all the difficulties and types of problems, which shows the potential data contamination, as well as the challenges for any existing LLM to solve unseen complex reasoning problems. We further explore various approaches such as fine-tuning, Chain-of-Thought prompting and problem description simplification. Unfortunately, none of them is able to consistently mitigate the challenges. Through our work, we emphasize the importance of this excellent data source for assessing the genuine reasoning capabilities of LLMs, and foster the development of LLMs with stronger reasoning abilities and better generalization in the future.
|
[
"Huang, Yiming",
"Lin, Zhenghao",
"Liu, Xiao",
"Gong, Yeyun",
"Lu, Shuai",
"Lei, Fangyu",
"Liang, Yaobo",
"Shen, Yelong",
"Lin, Chen",
"Duan, Nan",
"Chen, Weizhu"
] |
Competition-Level Problems are Effective {LLM} Evaluators
|
findings-acl.803
|
Poster
|
2402.17916v3
|
https://aclanthology.org/2024.findings-acl.804.bib
|
@inproceedings{yang-etal-2024-large-language,
title = "Large Language Models for Automated Open-domain Scientific Hypotheses Discovery",
author = "Yang, Zonglin and
Du, Xinya and
Li, Junxian and
Zheng, Jie and
Poria, Soujanya and
Cambria, Erik",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.804",
pages = "13545--13565",
abstract = "Hypothetical induction is recognized as the main reasoning type when scientists make observations about the world and try to propose hypotheses to explain those observations. Past research on hypothetical induction is under a constrained setting: (1) the observation annotations in the dataset are carefully manually handpicked sentences (resulting in a close-domain setting); and (2) the ground truth hypotheses are mostly commonsense knowledge, making the task less challenging. In this work, we tackle these problems by proposing the first dataset for social science academic hypotheses discovery, with the final goal to create systems that automatically generate valid, novel, and helpful scientific hypotheses, given only a pile of raw web corpus. Unlike previous settings, the new dataset requires (1) using open-domain data (raw web corpus) as observations; and (2) proposing hypotheses even new to humanity. A multi-module framework is developed for the task, including three different feedback mechanisms to boost performance, which exhibits superior performance in terms of both GPT-4 based and expert-based evaluation.To the best of our knowledge, this is the first work showing that LLMs are able to generate novel ({''}not existing in literature{''}) and valid ({''}reflecting reality{''}) scientific hypotheses.",
}
|
Hypothetical induction is recognized as the main reasoning type when scientists make observations about the world and try to propose hypotheses to explain those observations. Past research on hypothetical induction is under a constrained setting: (1) the observation annotations in the dataset are carefully manually handpicked sentences (resulting in a close-domain setting); and (2) the ground truth hypotheses are mostly commonsense knowledge, making the task less challenging. In this work, we tackle these problems by proposing the first dataset for social science academic hypotheses discovery, with the final goal to create systems that automatically generate valid, novel, and helpful scientific hypotheses, given only a pile of raw web corpus. Unlike previous settings, the new dataset requires (1) using open-domain data (raw web corpus) as observations; and (2) proposing hypotheses even new to humanity. A multi-module framework is developed for the task, including three different feedback mechanisms to boost performance, which exhibits superior performance in terms of both GPT-4 based and expert-based evaluation.To the best of our knowledge, this is the first work showing that LLMs are able to generate novel ({''}not existing in literature{''}) and valid ({''}reflecting reality{''}) scientific hypotheses.
|
[
"Yang, Zonglin",
"Du, Xinya",
"Li, Junxian",
"Zheng, Jie",
"Poria, Soujanya",
"Cambria, Erik"
] |
Large Language Models for Automated Open-domain Scientific Hypotheses Discovery
|
findings-acl.804
|
Poster
|
2309.02726v3
|
https://aclanthology.org/2024.findings-acl.805.bib
|
@inproceedings{li-lyu-2024-gradual,
title = "{GRADUAL}: Granularity-aware Dual Prototype Learning for Better Few-Shot Relation Extraction",
author = "Li, Zhiming and
Lyu, Yuchen",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.805",
pages = "13566--13577",
abstract = "Recent studies have shown that fusing text labels and context sentences is an effective method for learning prototype representations in few-shot relation extraction. However, the **inconsistency of prototype representations** across different few-shot tasks persists due to different context sentences for the same relation, even with the integration of text labels into prototype representations. Conversely, the text label for each relation is unique and consistent, 1)which prompts us to propose a **dual prototype learning method**. Unlike previous methods that only construct support-based prototypes, we additionally construct label-based prototypes. Furthermore, we introduce a graph-based prototype adjustment module to construct topological information between support-based and label-based prototypes, thereby generating a more effective similarity measure through a simple linear combination. In addition, relations of different granularities have different distribution widths in the same semantic space, the **imbalanced distribution in the semantic space** leads to a lack of comparability among relations. To create a more discriminative semantic space, 2)we propose a **granularity-aware prototype learning method** that unifies the distribution width of relations, making relations of different granularities have similar distribution widths. Experimental results on two public benchmark datasets show that our proposed methods achieve state-of-the-art performance in few-shot relation classification.",
}
|
Recent studies have shown that fusing text labels and context sentences is an effective method for learning prototype representations in few-shot relation extraction. However, the **inconsistency of prototype representations** across different few-shot tasks persists due to different context sentences for the same relation, even with the integration of text labels into prototype representations. Conversely, the text label for each relation is unique and consistent, 1)which prompts us to propose a **dual prototype learning method**. Unlike previous methods that only construct support-based prototypes, we additionally construct label-based prototypes. Furthermore, we introduce a graph-based prototype adjustment module to construct topological information between support-based and label-based prototypes, thereby generating a more effective similarity measure through a simple linear combination. In addition, relations of different granularities have different distribution widths in the same semantic space, the **imbalanced distribution in the semantic space** leads to a lack of comparability among relations. To create a more discriminative semantic space, 2)we propose a **granularity-aware prototype learning method** that unifies the distribution width of relations, making relations of different granularities have similar distribution widths. Experimental results on two public benchmark datasets show that our proposed methods achieve state-of-the-art performance in few-shot relation classification.
|
[
"Li, Zhiming",
"Lyu, Yuchen"
] |
{GRADUAL}: Granularity-aware Dual Prototype Learning for Better Few-Shot Relation Extraction
|
findings-acl.805
|
Poster
|
2212.02560v2
|
https://aclanthology.org/2024.findings-acl.806.bib
|
@inproceedings{wei-etal-2024-training,
title = "Training a Better {C}hinese Spelling Correction Model via Prior-knowledge Guided Teacher",
author = "Wei, Chi and
Huang, Shaobin and
Li, Rongsheng and
Yan, Naiyu and
Wang, Rui",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.806",
pages = "13578--13589",
abstract = "Recent advancements in Chinese Spelling Correction (CSC) predominantly leverage pre-trained language models (PLMs). However, a notable challenge with fine-tuned PLM-based CSC models is their tendency to over-correct, leading to poor generalization for error patterns outside the standard distribution. To address this, we developed a teacher network guided by prior knowledge for distillation learning of CSC models. Unlike traditional teacher networks, which depend on task-related pre-training, our method infuses task-related prior information into the teacher network, offering guidance beyond mere labels to the student network. This strategy significantly enhances the CSC model{'}s language modeling capabilities, crucial for minimizing over-correction. Importantly, our approach is model-independent and the teacher network does not require task-related pre-training, making it broadly applicable for enhancing various PLM-based CSC models with minimal additional computational resources. Extensive experiments on widely used benchmarks demonstrate that our method achieves new state-of-the-art results. Additionally, we explored the potential of generalizing our method to other non-autoregressive text-generation tasks.",
}
|
Recent advancements in Chinese Spelling Correction (CSC) predominantly leverage pre-trained language models (PLMs). However, a notable challenge with fine-tuned PLM-based CSC models is their tendency to over-correct, leading to poor generalization for error patterns outside the standard distribution. To address this, we developed a teacher network guided by prior knowledge for distillation learning of CSC models. Unlike traditional teacher networks, which depend on task-related pre-training, our method infuses task-related prior information into the teacher network, offering guidance beyond mere labels to the student network. This strategy significantly enhances the CSC model{'}s language modeling capabilities, crucial for minimizing over-correction. Importantly, our approach is model-independent and the teacher network does not require task-related pre-training, making it broadly applicable for enhancing various PLM-based CSC models with minimal additional computational resources. Extensive experiments on widely used benchmarks demonstrate that our method achieves new state-of-the-art results. Additionally, we explored the potential of generalizing our method to other non-autoregressive text-generation tasks.
|
[
"Wei, Chi",
"Huang, Shaobin",
"Li, Rongsheng",
"Yan, Naiyu",
"Wang, Rui"
] |
Training a Better {C}hinese Spelling Correction Model via Prior-knowledge Guided Teacher
|
findings-acl.806
|
Poster
|
2203.00991v1
|
https://aclanthology.org/2024.findings-acl.807.bib
|
@inproceedings{caffagni-etal-2024-revolution,
title = "The Revolution of Multimodal Large Language Models: A Survey",
author = "Caffagni, Davide and
Cocchi, Federico and
Barsellotti, Luca and
Moratelli, Nicholas and
Sarto, Sara and
Baraldi, Lorenzo and
Baraldi, Lorenzo and
Cornia, Marcella and
Cucchiara, Rita",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.807",
pages = "13590--13618",
abstract = "Connecting text and visual modalities plays an essential role in generative intelligence. For this reason, inspired by the success of large language models, significant research efforts are being devoted to the development of Multimodal Large Language Models (MLLMs). These models can seamlessly integrate visual and textual modalities, while providing a dialogue-based interface and instruction-following capabilities. In this paper, we provide a comprehensive review of recent visual-based MLLMs, analyzing their architectural choices, multimodal alignment strategies, and training techniques. We also conduct a detailed analysis of these models across a wide range of tasks, including visual grounding, image generation and editing, visual understanding, and domain-specific applications. Additionally, we compile and describe training datasets and evaluation benchmarks, conducting comparisons among existing models in terms of performance and computational requirements. Overall, this survey offers a comprehensive overview of the current state of the art, laying the groundwork for future MLLMs.",
}
|
Connecting text and visual modalities plays an essential role in generative intelligence. For this reason, inspired by the success of large language models, significant research efforts are being devoted to the development of Multimodal Large Language Models (MLLMs). These models can seamlessly integrate visual and textual modalities, while providing a dialogue-based interface and instruction-following capabilities. In this paper, we provide a comprehensive review of recent visual-based MLLMs, analyzing their architectural choices, multimodal alignment strategies, and training techniques. We also conduct a detailed analysis of these models across a wide range of tasks, including visual grounding, image generation and editing, visual understanding, and domain-specific applications. Additionally, we compile and describe training datasets and evaluation benchmarks, conducting comparisons among existing models in terms of performance and computational requirements. Overall, this survey offers a comprehensive overview of the current state of the art, laying the groundwork for future MLLMs.
|
[
"Caffagni, Davide",
"Cocchi, Federico",
"Barsellotti, Luca",
"Moratelli, Nicholas",
"Sarto, Sara",
"Baraldi, Lorenzo",
"Baraldi, Lorenzo",
"Cornia, Marcella",
"Cucchiara, Rita"
] |
The Revolution of Multimodal Large Language Models: A Survey
|
findings-acl.807
|
Poster
|
2402.12451v2
|
https://aclanthology.org/2024.findings-acl.808.bib
|
@inproceedings{wang-etal-2024-oop,
title = "{OOP}: Object-Oriented Programming Evaluation Benchmark for Large Language Models",
author = "Wang, Shuai and
Ding, Liang and
Shen, Li and
Luo, Yong and
Du, Bo and
Tao, Dacheng",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.808",
pages = "13619--13639",
abstract = "",
}
|
[
"Wang, Shuai",
"Ding, Liang",
"Shen, Li",
"Luo, Yong",
"Du, Bo",
"Tao, Dacheng"
] |
{OOP}: Object-Oriented Programming Evaluation Benchmark for Large Language Models
|
findings-acl.808
|
Poster
|
2208.08227v4
|
|
https://aclanthology.org/2024.findings-acl.809.bib
|
@inproceedings{song-etal-2024-code,
title = "Code Needs Comments: Enhancing Code {LLM}s with Comment Augmentation",
author = "Song, Demin and
Guo, Honglin and
Zhou, Yunhua and
Xing, Shuhao and
Wang, Yudong and
Song, Zifan and
Zhang, Wenwei and
Guo, Qipeng and
Yan, Hang and
Qiu, Xipeng and
Lin, Dahua",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.809",
pages = "13640--13656",
abstract = "The programming skill is one crucial ability for Large Language Models (LLMs), necessitating a deep understanding of programming languages (PLs) and their correlation with natural languages (NLs). We examine the impact of pre-training data on code-focused LLMs{'} performance by assessing the comment density as a measure of PL-NL alignment. Given the scarcity of code-comment aligned data in pre-training corpora, we introduce a novel data augmentation method that generates comments for existing code, coupled with a data filtering strategy that filters out code data poorly correlated with natural language. We conducted experiments on three code-focused LLMs and observed consistent improvements in performance on two widely-used programming skill benchmarks. Notably, the model trained on the augmented data outperformed both the model used for generating comments and the model further trained on the data without augmentation.",
}
|
The programming skill is one crucial ability for Large Language Models (LLMs), necessitating a deep understanding of programming languages (PLs) and their correlation with natural languages (NLs). We examine the impact of pre-training data on code-focused LLMs{'} performance by assessing the comment density as a measure of PL-NL alignment. Given the scarcity of code-comment aligned data in pre-training corpora, we introduce a novel data augmentation method that generates comments for existing code, coupled with a data filtering strategy that filters out code data poorly correlated with natural language. We conducted experiments on three code-focused LLMs and observed consistent improvements in performance on two widely-used programming skill benchmarks. Notably, the model trained on the augmented data outperformed both the model used for generating comments and the model further trained on the data without augmentation.
|
[
"Song, Demin",
"Guo, Honglin",
"Zhou, Yunhua",
"Xing, Shuhao",
"Wang, Yudong",
"Song, Zifan",
"Zhang, Wenwei",
"Guo, Qipeng",
"Yan, Hang",
"Qiu, Xipeng",
"Lin, Dahua"
] |
Code Needs Comments: Enhancing Code {LLM}s with Comment Augmentation
|
findings-acl.809
|
Poster
|
2402.13013v1
|
https://aclanthology.org/2024.findings-acl.810.bib
|
@inproceedings{you-etal-2024-efficient,
title = "Efficient Domain Adaptation for Non-Autoregressive Machine Translation",
author = "You, WangJie and
Guo, Pei and
Li, Juntao and
Chen, Kehai and
Zhang, Min",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.810",
pages = "13657--13670",
abstract = "Domain adaptation remains a challenge in the realm of Neural Machine Translation (NMT), even in the era of large language models (LLMs). Existing non-parametric approaches like nearest neighbor machine translation have made small Autoregressive Translation (AT) models achieve efficient domain generalization and adaptation without updating parameters, but leaving the Non-Autoregressive Translation (NAT) counterparts under-explored. To fill this blank, we introduce $Bi$-$k$NN, an innovative and efficient domain adaptation approach for NAT models that tailors a k-nearest-neighbor algorithm for NAT. Specifically, we introduce an effective datastore construction and correlated updating strategies to conform the parallel nature of NAT. Additionally, we train a meta-network that seamlessly integrates the NN distribution with the NMT distribution robustly during the iterative decoding process of NAT. Our experimental results across four benchmark datasets demonstrate that our $Bi$-$k$NN not only achieves significant improvements over the Base-NAT model (7.8 BLEU on average) but also exhibits enhanced efficiency.",
}
|
Domain adaptation remains a challenge in the realm of Neural Machine Translation (NMT), even in the era of large language models (LLMs). Existing non-parametric approaches like nearest neighbor machine translation have made small Autoregressive Translation (AT) models achieve efficient domain generalization and adaptation without updating parameters, but leaving the Non-Autoregressive Translation (NAT) counterparts under-explored. To fill this blank, we introduce $Bi$-$k$NN, an innovative and efficient domain adaptation approach for NAT models that tailors a k-nearest-neighbor algorithm for NAT. Specifically, we introduce an effective datastore construction and correlated updating strategies to conform the parallel nature of NAT. Additionally, we train a meta-network that seamlessly integrates the NN distribution with the NMT distribution robustly during the iterative decoding process of NAT. Our experimental results across four benchmark datasets demonstrate that our $Bi$-$k$NN not only achieves significant improvements over the Base-NAT model (7.8 BLEU on average) but also exhibits enhanced efficiency.
|
[
"You, WangJie",
"Guo, Pei",
"Li, Juntao",
"Chen, Kehai",
"Zhang, Min"
] |
Efficient Domain Adaptation for Non-Autoregressive Machine Translation
|
findings-acl.810
|
Poster
|
2110.09574v1
|
https://aclanthology.org/2024.findings-acl.811.bib
|
@inproceedings{guo-etal-2024-exploring,
title = "Exploring Reversal Mathematical Reasoning Ability for Large Language Models",
author = "Guo, Pei and
You, WangJie and
Li, Juntao and
Bowen, Yan and
Zhang, Min",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.811",
pages = "13671--13685",
abstract = "Large language models (LLMs) have presented remarkable capabilities in the wide range of natural language understanding and reasoning tasks. Despite their success, a few works indicate that LLMs suffer from the {``}reversal curse{''}, in which LLMs can{'}t employ the inverted structure {``}B is A{''} when they are trained based on {``}A is B{''}. To explore the effect of the {``}reversal curse{''} for LLMs on complex mathematical reasoning tasks, we present two reversal datasets upon GSM8K and MathQA and verify that LLMs also struggle to solve reversal mathematical problems. We analyze the potential reason and attribute it to the insufficient modeling of the relationship between reasoning steps caused by the left-to-right objective. Consequently, based on the characteristics of multi-step reasoning, we design a novel training method to improve the general and reversal reasoning abilities. Finally, we conduct experiments on four mathematical datasets, and the results demonstrate that our method significantly improves the general reasoning capacities and alleviates the reversal problem. Our datasets and codes are available at https: //github.com/AllForward/ReversalMath.",
}
|
Large language models (LLMs) have presented remarkable capabilities in the wide range of natural language understanding and reasoning tasks. Despite their success, a few works indicate that LLMs suffer from the {``}reversal curse{''}, in which LLMs can{'}t employ the inverted structure {``}B is A{''} when they are trained based on {``}A is B{''}. To explore the effect of the {``}reversal curse{''} for LLMs on complex mathematical reasoning tasks, we present two reversal datasets upon GSM8K and MathQA and verify that LLMs also struggle to solve reversal mathematical problems. We analyze the potential reason and attribute it to the insufficient modeling of the relationship between reasoning steps caused by the left-to-right objective. Consequently, based on the characteristics of multi-step reasoning, we design a novel training method to improve the general and reversal reasoning abilities. Finally, we conduct experiments on four mathematical datasets, and the results demonstrate that our method significantly improves the general reasoning capacities and alleviates the reversal problem. Our datasets and codes are available at https: //github.com/AllForward/ReversalMath.
|
[
"Guo, Pei",
"You, WangJie",
"Li, Juntao",
"Bowen, Yan",
"Zhang, Min"
] |
Exploring Reversal Mathematical Reasoning Ability for Large Language Models
|
findings-acl.811
|
Poster
|
2402.14404v2
|
https://aclanthology.org/2024.findings-acl.812.bib
|
@inproceedings{guo-etal-2024-unified,
title = "A Unified Joint Approach with Topological Context Learning and Rule Augmentation for Knowledge Graph Completion",
author = "Guo, Jingtao and
Zhang, Chunxia and
Li, Lingxi and
Xue, Xiaojun and
Niu, Zhendong",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.812",
pages = "13686--13696",
abstract = "Knowledge graph completion (KGC) task is to infer the missing knowledge in the knowledge graph based on known factual triples. However, present KGC approaches still face the following two challenges. Those methods perform simple linear update on relation representation, and only local neighborhood information is aggregated, which makes it difficult to capture logic semantic between relations and global topological context information. To tackle the above challenges, we propose a unified joint approach with Topological Context learning and Rule Augmentation (TCRA) for KGC. The TCRA framework consists of an entity topological context learning mechanism based on dual-branch hierarchical graph attention network, and a relation rule context learning mechanism based on Rule-Transformer and rule-to-relation aggregator. The former mechanism encodes the topological structure features of entities, aggregates the local neighborhood topological context information of entities on the three levels (entity, relation and triple), and build clusters of global head or tail entities related to the same relation. It can capture the local and global topological context information of entities related to the same relation. The latter mechanism introduces chain-like Horn rules as the context information of relations, and encodes the logical semantic of relations to enrich the relation representation. Experimental performances on three benchmark datasets FB15k-237, WN18RR and Kinship indicate the effectiveness and superiority of our proposed approach. The codes are publicly available.",
}
|
Knowledge graph completion (KGC) task is to infer the missing knowledge in the knowledge graph based on known factual triples. However, present KGC approaches still face the following two challenges. Those methods perform simple linear update on relation representation, and only local neighborhood information is aggregated, which makes it difficult to capture logic semantic between relations and global topological context information. To tackle the above challenges, we propose a unified joint approach with Topological Context learning and Rule Augmentation (TCRA) for KGC. The TCRA framework consists of an entity topological context learning mechanism based on dual-branch hierarchical graph attention network, and a relation rule context learning mechanism based on Rule-Transformer and rule-to-relation aggregator. The former mechanism encodes the topological structure features of entities, aggregates the local neighborhood topological context information of entities on the three levels (entity, relation and triple), and build clusters of global head or tail entities related to the same relation. It can capture the local and global topological context information of entities related to the same relation. The latter mechanism introduces chain-like Horn rules as the context information of relations, and encodes the logical semantic of relations to enrich the relation representation. Experimental performances on three benchmark datasets FB15k-237, WN18RR and Kinship indicate the effectiveness and superiority of our proposed approach. The codes are publicly available.
|
[
"Guo, Jingtao",
"Zhang, Chunxia",
"Li, Lingxi",
"Xue, Xiaojun",
"Niu, Zhendong"
] |
A Unified Joint Approach with Topological Context Learning and Rule Augmentation for Knowledge Graph Completion
|
findings-acl.812
|
Poster
|
1611.04125v1
|
https://aclanthology.org/2024.findings-acl.813.bib
|
@inproceedings{vu-etal-2024-freshllms,
title = "{F}resh{LLM}s: Refreshing Large Language Models with Search Engine Augmentation",
author = "Vu, Tu and
Iyyer, Mohit and
Wang, Xuezhi and
Constant, Noah and
Wei, Jerry and
Wei, Jason and
Tar, Chris and
Sung, Yun-Hsuan and
Zhou, Denny and
Le, Quoc and
Luong, Thang",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.813",
pages = "13697--13720",
abstract = "Since most large language models (LLMs) are trained once and never updated, they struggle to dynamically adapt to our ever-changing world. In this work, we present FreshQA, a dynamic QA benchmark that tests a model{'}s ability to answer questions that may require reasoning over up-to-date world knowledge. We develop a two-mode human evaluation procedure to measure both correctness and hallucination, which we use to benchmark both closed and open-source LLMs by collecting {\textgreater}50K human judgments. We observe that all LLMs struggle to answer questions that require fast-changing world knowledge as well as questions with false premises that need to be debunked. In response, we develop FreshPrompt, a few-shot prompting method that curates and organizes relevant information from a search engine into an LLM{'}s prompt. Our experiments show that FreshPrompt outperforms both competing search engine-augmented prompting methods such as Self-Ask (Press et al., 2022) as well as commercial systems such as Perplexity.AI. To facilitate future work, we additionally develop FreshEval, a reliable autorater for quick evaluation and comparison on FreshQA. Our latest results with FreshEval suggest that open-source LLMs such as Mixtral (Jiang et al., 2024), when combined with FreshPrompt, are competitive with closed-source and commercial systems on search-augmented QA.",
}
|
Since most large language models (LLMs) are trained once and never updated, they struggle to dynamically adapt to our ever-changing world. In this work, we present FreshQA, a dynamic QA benchmark that tests a model{'}s ability to answer questions that may require reasoning over up-to-date world knowledge. We develop a two-mode human evaluation procedure to measure both correctness and hallucination, which we use to benchmark both closed and open-source LLMs by collecting {\textgreater}50K human judgments. We observe that all LLMs struggle to answer questions that require fast-changing world knowledge as well as questions with false premises that need to be debunked. In response, we develop FreshPrompt, a few-shot prompting method that curates and organizes relevant information from a search engine into an LLM{'}s prompt. Our experiments show that FreshPrompt outperforms both competing search engine-augmented prompting methods such as Self-Ask (Press et al., 2022) as well as commercial systems such as Perplexity.AI. To facilitate future work, we additionally develop FreshEval, a reliable autorater for quick evaluation and comparison on FreshQA. Our latest results with FreshEval suggest that open-source LLMs such as Mixtral (Jiang et al., 2024), when combined with FreshPrompt, are competitive with closed-source and commercial systems on search-augmented QA.
|
[
"Vu, Tu",
"Iyyer, Mohit",
"Wang, Xuezhi",
"Constant, Noah",
"Wei, Jerry",
"Wei, Jason",
"Tar, Chris",
"Sung, Yun-Hsuan",
"Zhou, Denny",
"Le, Quoc",
"Luong, Thang"
] |
{F}resh{LLM}s: Refreshing Large Language Models with Search Engine Augmentation
|
findings-acl.813
|
Poster
|
2110.08512v1
|
https://aclanthology.org/2024.findings-acl.814.bib
|
@inproceedings{zhong-etal-2024-rose,
title = "{ROSE} Doesn{'}t Do That: Boosting the Safety of Instruction-Tuned Large Language Models with Reverse Prompt Contrastive Decoding",
author = "Zhong, Qihuang and
Ding, Liang and
Liu, Juhua and
Du, Bo and
Tao, Dacheng",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.814",
pages = "13721--13736",
abstract = "With the development of instruction-tuned large language models (LLMs), improving the safety of LLMs has become more critical. However, the current approaches for aligning the LLMs output with expected safety usually require substantial training efforts, e.g., high-quality safety data and expensive computational resources, which are costly and inefficient. To this end, we present reverse prompt contrastive decoding (ROSE), a simple-yet-effective method to directly boost the safety of existing instruction-tuned LLMs without any additional training. The principle of ROSE is to improve the probability of desired safe output via suppressing the undesired output induced by the carefully-designed reverse prompts. Experiments on 6 safety and 2 general-purpose tasks show that, our ROSE not only brings consistent and significant safety improvements (up to +13.8{\%} safety score) upon 5 types of instruction-tuned LLMs, but also benefits the general-purpose ability of LLMs. In-depth analyses explore the underlying mechanism of ROSE, and reveal when and where to use it.",
}
|
With the development of instruction-tuned large language models (LLMs), improving the safety of LLMs has become more critical. However, the current approaches for aligning the LLMs output with expected safety usually require substantial training efforts, e.g., high-quality safety data and expensive computational resources, which are costly and inefficient. To this end, we present reverse prompt contrastive decoding (ROSE), a simple-yet-effective method to directly boost the safety of existing instruction-tuned LLMs without any additional training. The principle of ROSE is to improve the probability of desired safe output via suppressing the undesired output induced by the carefully-designed reverse prompts. Experiments on 6 safety and 2 general-purpose tasks show that, our ROSE not only brings consistent and significant safety improvements (up to +13.8{\%} safety score) upon 5 types of instruction-tuned LLMs, but also benefits the general-purpose ability of LLMs. In-depth analyses explore the underlying mechanism of ROSE, and reveal when and where to use it.
|
[
"Zhong, Qihuang",
"Ding, Liang",
"Liu, Juhua",
"Du, Bo",
"Tao, Dacheng"
] |
{ROSE} Doesn{'}t Do That: Boosting the Safety of Instruction-Tuned Large Language Models with Reverse Prompt Contrastive Decoding
|
findings-acl.814
|
Poster
|
2402.11889v2
|
https://aclanthology.org/2024.findings-acl.815.bib
|
@inproceedings{li-etal-2024-cr,
title = "{CR}-{LLM}: A Dataset and Optimization for Concept Reasoning of Large Language Models",
author = "Li, Nianqi and
Liu, Jingping and
Jiang, Sihang and
Jiang, Haiyun and
Xiao, Yanghua and
Liang, Jiaqing and
Liang, Zujie and
Wei, Feng and
Chen, Jinglei and
Hao, Zhenghong and
Han, Bing",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.815",
pages = "13737--13747",
abstract = "Concept reasoning is an important capability for models to understand the world. However, the existing datasets, such as concept extraction and concept generation, suffer from modeledge leakage and context leakage. To address these limitations, we construct a dataset of concept reasoning for large language models (CR-LLM) with modeledge leakage prevention and context leakage prevention, which consists of 2,167 samples and covers different concept types. In addition, we propose a hybrid reasoning method, consisting of inductive reasoning, deductive reasoning and a controller. This method allows large language models to adaptively select the optimal reasoning method for each input sample. Finally, we conduct extensive experiments on CR-LLM using different models and methods. The results show that existing large language models and reasoning methods perform sub-optimally in the concept reasoning task. In contrast, our proposed method significantly improves the capabilities, achieving a 7{\%} increase in accuracy compared to CoT and demonstrating better granularity. We release CR-LLM and code at https://github.com/Nianqi-Li/Concept-Reasoning-for-LLMs.",
}
|
Concept reasoning is an important capability for models to understand the world. However, the existing datasets, such as concept extraction and concept generation, suffer from modeledge leakage and context leakage. To address these limitations, we construct a dataset of concept reasoning for large language models (CR-LLM) with modeledge leakage prevention and context leakage prevention, which consists of 2,167 samples and covers different concept types. In addition, we propose a hybrid reasoning method, consisting of inductive reasoning, deductive reasoning and a controller. This method allows large language models to adaptively select the optimal reasoning method for each input sample. Finally, we conduct extensive experiments on CR-LLM using different models and methods. The results show that existing large language models and reasoning methods perform sub-optimally in the concept reasoning task. In contrast, our proposed method significantly improves the capabilities, achieving a 7{\%} increase in accuracy compared to CoT and demonstrating better granularity. We release CR-LLM and code at https://github.com/Nianqi-Li/Concept-Reasoning-for-LLMs.
|
[
"Li, Nianqi",
"Liu, Jingping",
"Jiang, Sihang",
"Jiang, Haiyun",
"Xiao, Yanghua",
"Liang, Jiaqing",
"Liang, Zujie",
"Wei, Feng",
"Chen, Jinglei",
"Hao, Zhenghong",
"Han, Bing"
] |
{CR}-{LLM}: A Dataset and Optimization for Concept Reasoning of Large Language Models
|
findings-acl.815
|
Poster
|
2402.14660v2
|
https://aclanthology.org/2024.findings-acl.816.bib
|
@inproceedings{min-etal-2024-data,
title = "{DATA}-{CUBE}: Data Curriculum for Instruction-based Sentence Representation Learning",
author = "Min, Yingqian and
Zhou, Kun and
Gao, Dawei and
Zhao, Xin and
Hu, He and
Li, Yaliang",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.816",
pages = "13748--13761",
abstract = "Recently, multi-task instruction tuning has been utilized to improve sentence representation learning (SRL). It enables SRL models to generate task-specific representations with the guidance of task instruction, thus exhibiting strong generalization ability on unseen tasks. However, these methods mostly neglect the potential interference problems across different tasks and instances, which may affect the training of the model.To address this issue, we propose a data curriculum method, namely **Data-CUBE**, that arranges the order of all the multi-task data for training, to minimize the interference risks from two aspects.At the task level, we aim to find the optimal task order to minimize the total cross-task interference risk and formulate this problem as the traveling salesman problem, which is further solved by a specially designed simulated annealing algorithm. At the instance level, we propose a measurement method to quantify the difficulty of all instances per task, and then arrange instances in an easy-to-difficult order for training.Experimental results show that our approach can boost the performance of state-of-the-art methods. Our code and data will be publicly released.",
}
|
Recently, multi-task instruction tuning has been utilized to improve sentence representation learning (SRL). It enables SRL models to generate task-specific representations with the guidance of task instruction, thus exhibiting strong generalization ability on unseen tasks. However, these methods mostly neglect the potential interference problems across different tasks and instances, which may affect the training of the model.To address this issue, we propose a data curriculum method, namely **Data-CUBE**, that arranges the order of all the multi-task data for training, to minimize the interference risks from two aspects.At the task level, we aim to find the optimal task order to minimize the total cross-task interference risk and formulate this problem as the traveling salesman problem, which is further solved by a specially designed simulated annealing algorithm. At the instance level, we propose a measurement method to quantify the difficulty of all instances per task, and then arrange instances in an easy-to-difficult order for training.Experimental results show that our approach can boost the performance of state-of-the-art methods. Our code and data will be publicly released.
|
[
"Min, Yingqian",
"Zhou, Kun",
"Gao, Dawei",
"Zhao, Xin",
"Hu, He",
"Li, Yaliang"
] |
{DATA}-{CUBE}: Data Curriculum for Instruction-based Sentence Representation Learning
|
findings-acl.816
|
Poster
|
2006.02014v1
|
https://aclanthology.org/2024.findings-acl.817.bib
|
@inproceedings{lin-etal-2024-combating,
title = "Combating Label Sparsity in Short Text Topic Modeling via Nearest Neighbor Augmentation",
author = "Lin, Yang and
Ma, Xinyu and
Gao, Xin and
Li, Ruiqing and
Wang, Yasha and
Chu, Xu",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.817",
pages = "13762--13774",
abstract = "Extracting semantic topics from short texts presents a significant challenge in the field of data mining. While efforts have been made to mitigate data sparsity issue, the limited length of short documents also results in the absence of semantically relevant words, causing biased evidence lower bound and incomplete labels for likelihood maximization. We refer to this issue as the label sparsity problem. To combat this problem, we propose kNNTM, a neural short text topic model that incorporates a $k$-Nearest-Neighbor-based label completion algorithm by augmenting the reconstruction label with $k$-nearest documents to complement these relevant but unobserved words. Furthermore, seeking a precise reflection of distances between documents, we propose a fused multi-view distances metric that takes both local word similarities and global topic semantics into consideration. Extensive experiments on multiple public short-text datasets show that kNNTM model outperforms the state-of-the-art baseline models and can derive both high-quality topics and document representations.",
}
|
Extracting semantic topics from short texts presents a significant challenge in the field of data mining. While efforts have been made to mitigate data sparsity issue, the limited length of short documents also results in the absence of semantically relevant words, causing biased evidence lower bound and incomplete labels for likelihood maximization. We refer to this issue as the label sparsity problem. To combat this problem, we propose kNNTM, a neural short text topic model that incorporates a $k$-Nearest-Neighbor-based label completion algorithm by augmenting the reconstruction label with $k$-nearest documents to complement these relevant but unobserved words. Furthermore, seeking a precise reflection of distances between documents, we propose a fused multi-view distances metric that takes both local word similarities and global topic semantics into consideration. Extensive experiments on multiple public short-text datasets show that kNNTM model outperforms the state-of-the-art baseline models and can derive both high-quality topics and document representations.
|
[
"Lin, Yang",
"Ma, Xinyu",
"Gao, Xin",
"Li, Ruiqing",
"Wang, Yasha",
"Chu, Xu"
] |
Combating Label Sparsity in Short Text Topic Modeling via Nearest Neighbor Augmentation
|
findings-acl.817
|
Poster
|
2211.12878v1
|
https://aclanthology.org/2024.findings-acl.818.bib
|
@inproceedings{yan-etal-2024-refutebench,
title = "{R}efute{B}ench: Evaluating Refuting Instruction-Following for Large Language Models",
author = "Yan, Jianhao and
Luo, Yun and
Zhang, Yue",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.818",
pages = "13775--13791",
abstract = "The application scope of large language models (LLMs) is increasingly expanding. In practical use, users might provide feedback based on the model{'}s output, hoping for a responsive model that can complete responses according to their feedback. Whether the model can appropriately respond to users{'} refuting feedback and consistently follow through with execution has not been thoroughly analyzed. In light of this, this paper proposes a comprehensive benchmark, \textbf{RefuteBench}, covering tasks such as question answering, machine translation, and email writing. The evaluation aims to assess whether models can positively accept feedback in form of refuting instructions and whether they can consistently adhere to user demands throughout the conversation. We conduct evaluations on numerous LLMs and find that LLMs are stubborn, i.e. exhibit inclination to their internal knowledge, often failing to comply with user feedback. Additionally, as the length of the conversation increases, models gradually forget the user{'}s stated feedback and roll back to their own responses. We further propose a \textit{recall-and-repeat} prompts as a simple and effective way to enhance the model{'}s responsiveness to feedback.",
}
|
The application scope of large language models (LLMs) is increasingly expanding. In practical use, users might provide feedback based on the model{'}s output, hoping for a responsive model that can complete responses according to their feedback. Whether the model can appropriately respond to users{'} refuting feedback and consistently follow through with execution has not been thoroughly analyzed. In light of this, this paper proposes a comprehensive benchmark, \textbf{RefuteBench}, covering tasks such as question answering, machine translation, and email writing. The evaluation aims to assess whether models can positively accept feedback in form of refuting instructions and whether they can consistently adhere to user demands throughout the conversation. We conduct evaluations on numerous LLMs and find that LLMs are stubborn, i.e. exhibit inclination to their internal knowledge, often failing to comply with user feedback. Additionally, as the length of the conversation increases, models gradually forget the user{'}s stated feedback and roll back to their own responses. We further propose a \textit{recall-and-repeat} prompts as a simple and effective way to enhance the model{'}s responsiveness to feedback.
|
[
"Yan, Jianhao",
"Luo, Yun",
"Zhang, Yue"
] |
{R}efute{B}ench: Evaluating Refuting Instruction-Following for Large Language Models
|
findings-acl.818
|
Poster
|
2310.08394v2
|
https://aclanthology.org/2024.findings-acl.819.bib
|
@inproceedings{xiao-cao-2024-complex,
title = "Complex Logical Query Answering by Calibrating Knowledge Graph Completion Models",
author = "Xiao, Changyi and
Cao, Yixin",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.819",
pages = "13792--13803",
abstract = "Complex logical query answering (CLQA) is a challenging task that involves finding answer entities for complex logical queries over incomplete knowledge graphs (KGs). Previous research has explored the use of pre-trained knowledge graph completion (KGC) models, which can predict the missing facts in KGs, to answer complex logical queries. However, KGC models are typically evaluated using ranking evaluation metrics, which may result in values of predictions of KGC models that are not well-calibrated. In this paper, we propose a method for calibrating KGC models, namely CKGC, which enables KGC models to adapt to answering complex logical queries. Notably, CKGC is lightweight and effective. The adaptation function is simple, allowing the model to quickly converge during the adaptation process. The core concept of CKGC is to map the values of predictions of KGC models to the range [0, 1], ensuring that values associated with true facts are close to 1, while values linked to false facts are close to 0. Through experiments on three benchmark datasets, we demonstrate that our proposed calibration method can significantly boost model performance in the CLQA task. Moreover, our approach can enhance the performance of CLQA while preserving the ranking evaluation metrics of KGC models. The code is available at https://github.com/changyi7231/CKGC.",
}
|
Complex logical query answering (CLQA) is a challenging task that involves finding answer entities for complex logical queries over incomplete knowledge graphs (KGs). Previous research has explored the use of pre-trained knowledge graph completion (KGC) models, which can predict the missing facts in KGs, to answer complex logical queries. However, KGC models are typically evaluated using ranking evaluation metrics, which may result in values of predictions of KGC models that are not well-calibrated. In this paper, we propose a method for calibrating KGC models, namely CKGC, which enables KGC models to adapt to answering complex logical queries. Notably, CKGC is lightweight and effective. The adaptation function is simple, allowing the model to quickly converge during the adaptation process. The core concept of CKGC is to map the values of predictions of KGC models to the range [0, 1], ensuring that values associated with true facts are close to 1, while values linked to false facts are close to 0. Through experiments on three benchmark datasets, we demonstrate that our proposed calibration method can significantly boost model performance in the CLQA task. Moreover, our approach can enhance the performance of CLQA while preserving the ranking evaluation metrics of KGC models. The code is available at https://github.com/changyi7231/CKGC.
|
[
"Xiao, Changyi",
"Cao, Yixin"
] |
Complex Logical Query Answering by Calibrating Knowledge Graph Completion Models
|
findings-acl.819
|
Poster
|
2205.10128v2
|
https://aclanthology.org/2024.findings-acl.820.bib
|
@inproceedings{lin-etal-2024-argument,
title = "Argument-Based Sentiment Analysis on Forward-Looking Statements",
author = "Lin, Chin-Yi and
Chen, Chung-Chi and
Huang, Hen-Hsen and
Chen, Hsin-Hsi",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.820",
pages = "13804--13815",
abstract = "This paper introduces a novel approach to analyzing the forward-looking statements in equity research reports by integrating argument mining with sentiment analysis. Recognizing the limitations of traditional models in capturing the nuances of future-oriented analysis, we propose a refined categorization of argument units into claims, premises, and scenarios, coupled with a unique sentiment analysis framework. Furthermore, we incorporate a temporal dimension to categorize the anticipated impact duration of market events. To facilitate this study, we present the Equity Argument Mining and Sentiment Analysis (Equity-AMSA) dataset. Our research investigates the extent to which detailed domain-specific annotations can be provided, the necessity of fine-grained human annotations in the era of large language models, and whether our proposed framework can improve performance in downstream tasks over traditional methods. Experimental results reveal the significance of manual annotations, especially for scenario identification and sentiment analysis. The study concludes that our annotation scheme and dataset contribute to a deeper understanding of forward-looking statements in equity research reports.",
}
|
This paper introduces a novel approach to analyzing the forward-looking statements in equity research reports by integrating argument mining with sentiment analysis. Recognizing the limitations of traditional models in capturing the nuances of future-oriented analysis, we propose a refined categorization of argument units into claims, premises, and scenarios, coupled with a unique sentiment analysis framework. Furthermore, we incorporate a temporal dimension to categorize the anticipated impact duration of market events. To facilitate this study, we present the Equity Argument Mining and Sentiment Analysis (Equity-AMSA) dataset. Our research investigates the extent to which detailed domain-specific annotations can be provided, the necessity of fine-grained human annotations in the era of large language models, and whether our proposed framework can improve performance in downstream tasks over traditional methods. Experimental results reveal the significance of manual annotations, especially for scenario identification and sentiment analysis. The study concludes that our annotation scheme and dataset contribute to a deeper understanding of forward-looking statements in equity research reports.
|
[
"Lin, Chin-Yi",
"Chen, Chung-Chi",
"Huang, Hen-Hsen",
"Chen, Hsin-Hsi"
] |
Argument-Based Sentiment Analysis on Forward-Looking Statements
|
findings-acl.820
|
Poster
|
2310.17924v1
|
https://aclanthology.org/2024.findings-acl.821.bib
|
@inproceedings{zhang-etal-2024-paying,
title = "Paying More Attention to Source Context: Mitigating Unfaithful Translations from Large Language Model",
author = "Zhang, Hongbin and
Chen, Kehai and
Bai, Xuefeng and
Xiang, Yang and
Zhang, Min",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.821",
pages = "13816--13836",
abstract = "Large language models (LLMs) have showcased their remarkable capabilities to handle various downstream tasks, including multilingual machine translation ability. Despite their impressive performance, decoder-only LLMs lack an explicit alignment between source and target contexts, leading to translation that may not faithfully represent the original content. To address this, we propose three learning strategies to encourage LLMs to pay more attention to the source context during translation: 1) adjusting attention weights on the source context by adaptive attention re-weighting; 2) suppressing the irrelevant target prefix using contrastive decoding; 3) avoiding excessive reliance on the target prefix through target-constrained tuning. To verify the effectiveness of our model, we curate a new dataset specifically focusing on unfaithful translations generated by LLMs. Experimental results on both human-collected and general test sets verify the effectiveness of our model across multiple language pairs. Further human evaluation demonstrates the efficacy of our method in reducing hallucinatory translation and improving the fidelity of translations.",
}
|
Large language models (LLMs) have showcased their remarkable capabilities to handle various downstream tasks, including multilingual machine translation ability. Despite their impressive performance, decoder-only LLMs lack an explicit alignment between source and target contexts, leading to translation that may not faithfully represent the original content. To address this, we propose three learning strategies to encourage LLMs to pay more attention to the source context during translation: 1) adjusting attention weights on the source context by adaptive attention re-weighting; 2) suppressing the irrelevant target prefix using contrastive decoding; 3) avoiding excessive reliance on the target prefix through target-constrained tuning. To verify the effectiveness of our model, we curate a new dataset specifically focusing on unfaithful translations generated by LLMs. Experimental results on both human-collected and general test sets verify the effectiveness of our model across multiple language pairs. Further human evaluation demonstrates the efficacy of our method in reducing hallucinatory translation and improving the fidelity of translations.
|
[
"Zhang, Hongbin",
"Chen, Kehai",
"Bai, Xuefeng",
"Xiang, Yang",
"Zhang, Min"
] |
Paying More Attention to Source Context: Mitigating Unfaithful Translations from Large Language Model
|
findings-acl.821
|
Poster
|
2406.07036v1
|
https://aclanthology.org/2024.findings-acl.822.bib
|
@inproceedings{bhushan-etal-2024-unveiling,
title = "Unveiling the Power of Integration: Block Diagram Summarization through Local-Global Fusion",
author = "Bhushan, Shreyanshu and
Jung, Eun-Soo and
Lee, Minho",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.822",
pages = "13837--13856",
abstract = "Block Diagrams play an essential role in visualizing the relationships between components or systems. Generating summaries of block diagrams is important for document understanding or question answering (QA) tasks by providing concise overviews of complex systems. However, it{'}s a challenging task as it requires compressing complex relationships into informative descriptions. In this paper, we present {``}BlockNet{''}, a fusion framework that summarizes block diagrams by integrating local and global information, catering to both English and Korean languages. Additionally, we introduce a new multilingual method to produce block diagram data, resulting in a high-quality dataset called {``}BD-EnKo{''}. In BlockNet, we develop {``}BlockSplit{''}, an Optical Character Recognition (OCR) based algorithm employing the divide-and-conquer principle for local information extraction. We train an OCR-free transformer architecture for global information extraction using BD-EnKo and public data. To assess the effectiveness of our model, we conduct thorough experiments on different datasets. The assessment shows that BlockNet surpasses all previous methods and models, including GPT-4V, for block diagram summarization.",
}
|
Block Diagrams play an essential role in visualizing the relationships between components or systems. Generating summaries of block diagrams is important for document understanding or question answering (QA) tasks by providing concise overviews of complex systems. However, it{'}s a challenging task as it requires compressing complex relationships into informative descriptions. In this paper, we present {``}BlockNet{''}, a fusion framework that summarizes block diagrams by integrating local and global information, catering to both English and Korean languages. Additionally, we introduce a new multilingual method to produce block diagram data, resulting in a high-quality dataset called {``}BD-EnKo{''}. In BlockNet, we develop {``}BlockSplit{''}, an Optical Character Recognition (OCR) based algorithm employing the divide-and-conquer principle for local information extraction. We train an OCR-free transformer architecture for global information extraction using BD-EnKo and public data. To assess the effectiveness of our model, we conduct thorough experiments on different datasets. The assessment shows that BlockNet surpasses all previous methods and models, including GPT-4V, for block diagram summarization.
|
[
"Bhushan, Shreyanshu",
"Jung, Eun-Soo",
"Lee, Minho"
] |
Unveiling the Power of Integration: Block Diagram Summarization through Local-Global Fusion
|
findings-acl.822
|
Poster
|
2312.01017v1
|
https://aclanthology.org/2024.findings-acl.823.bib
|
@inproceedings{li-etal-2024-multisql,
title = "{M}ulti{SQL}: A Schema-Integrated Context-Dependent {T}ext2{SQL} Dataset with Diverse {SQL} Operations",
author = "Li, Chunhui and
Wang, Yifan and
Wu, Zhen and
Yu, Zhen and
Zhao, Fei and
Huang, Shujian and
Dai, Xinyu",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.823",
pages = "13857--13867",
abstract = "Text2SQL is a task that translates natural language into SQL statements. Context-dependent Text2SQL offers a more natural database interaction by simulating dialogues between users and databases, with CoSQL and SparC as representative datasets. Yet, these datasets struggle to accurately replicate real-world situations. To address this, we introduce MultiSQL, which extends them in three key aspects: (1) Diverse SQL Operations. We incorporate diverse SQL types such as Create, Update, and Insert to broaden the scope of SQL operations. (2) Schema-Integrated Context. We integrated query context with database schema dependencies to better depict database complexity. (3) Extended Dialogues. We expand dialogue length to better simulate long conversations and complex interactions. This multi-type, schema-integrated, context-dependent Text2SQL dataset comprises nearly 800 dialogue groups and over 9,000 interaction turns across 166 complex databases, offering a better benchmark for interactive user-database dialogue.Addressing MultiSQL{'}s challenges, we refined evaluation metrics to better capture diverse SQL types and schema dependencies. We designed a prompt framework that leverages historical data and self-refinement to accurately capture the dependency between text queries and database structures. Experiments with GPT-3.5, GPT-4, and LLaMA2-7B show both the effectiveness of our strategies and the challenges of MultiSQL. The datasets is available at https://github.com/grandchicken/MultiSQL.",
}
|
Text2SQL is a task that translates natural language into SQL statements. Context-dependent Text2SQL offers a more natural database interaction by simulating dialogues between users and databases, with CoSQL and SparC as representative datasets. Yet, these datasets struggle to accurately replicate real-world situations. To address this, we introduce MultiSQL, which extends them in three key aspects: (1) Diverse SQL Operations. We incorporate diverse SQL types such as Create, Update, and Insert to broaden the scope of SQL operations. (2) Schema-Integrated Context. We integrated query context with database schema dependencies to better depict database complexity. (3) Extended Dialogues. We expand dialogue length to better simulate long conversations and complex interactions. This multi-type, schema-integrated, context-dependent Text2SQL dataset comprises nearly 800 dialogue groups and over 9,000 interaction turns across 166 complex databases, offering a better benchmark for interactive user-database dialogue.Addressing MultiSQL{'}s challenges, we refined evaluation metrics to better capture diverse SQL types and schema dependencies. We designed a prompt framework that leverages historical data and self-refinement to accurately capture the dependency between text queries and database structures. Experiments with GPT-3.5, GPT-4, and LLaMA2-7B show both the effectiveness of our strategies and the challenges of MultiSQL. The datasets is available at https://github.com/grandchicken/MultiSQL.
|
[
"Li, Chunhui",
"Wang, Yifan",
"Wu, Zhen",
"Yu, Zhen",
"Zhao, Fei",
"Huang, Shujian",
"Dai, Xinyu"
] |
{M}ulti{SQL}: A Schema-Integrated Context-Dependent {T}ext2{SQL} Dataset with Diverse {SQL} Operations
|
findings-acl.823
|
Poster
|
2305.11061v1
|
https://aclanthology.org/2024.findings-acl.824.bib
|
@inproceedings{li-etal-2024-towards-demonstration,
title = "Towards Demonstration-Aware Large Language Models for Machine Translation",
author = "Li, Chen and
Zhang, Meishan and
Liu, Xuebo and
Li, Zhaocong and
Wong, Derek and
Zhang, Min",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.824",
pages = "13868--13881",
abstract = "Tuning-based large language models for machine translation (aka large translation model, LTM) have demonstrated significant performance in the field of machine translation. Despite their success, these models often face difficulties in leveraging demonstrations to further improve their performance. To tackle this challenge, we introduce a novel approach that integrates demonstration-aware training and inference strategies within the framework of tuning-based LTMs, hereby referred to as demonstration-aware LTMs. During training, we enrich the model{'}s learning process by incorporating both sentence- and document-level demonstrations derived from its original training dataset. During inference, the model synergizes its own contextual translations with retrieved high-quality demonstrations, leading to more precise and contextually appropriate outputs. Empirical results reveal that our demonstration-aware LTM not only mitigates the negative impacts traditionally associated with demonstrations but also secures substantial improvements in translation accuracy, particularly in domain-specific and document-level translation tasks. Source code and scripts are freely available at https://github.com/ChenLi0620/Demo-Aware-LLM-MT.",
}
|
Tuning-based large language models for machine translation (aka large translation model, LTM) have demonstrated significant performance in the field of machine translation. Despite their success, these models often face difficulties in leveraging demonstrations to further improve their performance. To tackle this challenge, we introduce a novel approach that integrates demonstration-aware training and inference strategies within the framework of tuning-based LTMs, hereby referred to as demonstration-aware LTMs. During training, we enrich the model{'}s learning process by incorporating both sentence- and document-level demonstrations derived from its original training dataset. During inference, the model synergizes its own contextual translations with retrieved high-quality demonstrations, leading to more precise and contextually appropriate outputs. Empirical results reveal that our demonstration-aware LTM not only mitigates the negative impacts traditionally associated with demonstrations but also secures substantial improvements in translation accuracy, particularly in domain-specific and document-level translation tasks. Source code and scripts are freely available at https://github.com/ChenLi0620/Demo-Aware-LLM-MT.
|
[
"Li, Chen",
"Zhang, Meishan",
"Liu, Xuebo",
"Li, Zhaocong",
"Wong, Derek",
"Zhang, Min"
] |
Towards Demonstration-Aware Large Language Models for Machine Translation
|
findings-acl.824
|
Poster
|
2104.10441v1
|
https://aclanthology.org/2024.findings-acl.825.bib
|
@inproceedings{lee-etal-2024-dada,
title = "{DADA}: Distribution-Aware Domain Adaptation of {PLM}s for Information Retrieval",
author = "Lee, Dohyeon and
Kim, Jongyoon and
Hwang, Seung-won and
Park, Joonsuk",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.825",
pages = "13882--13893",
abstract = "Pre-trained language models (PLMs) exhibit promise in retrieval tasks but struggle with out-of-domain data due to distribution shifts.Addressing this, generative domain adaptation (DA), known as GPL, tackles distribution shifts by generating pseudo queries and labels to train models for predicting query-document relationships in new domains.However, it overlooks the domain distribution, causing the model to struggle with aligning the distribution in the target domain.We, therefore, propose a Distribution-Aware Domain Adaptation (DADA) to guide the model to consider the domain distribution knowledge at the level of both a single document and the corpus, which is referred to as observation-level feedback and domain-level feedback, respectively.Our method effectively adapts the model to the target domain and expands document representation to unseen gold query terms using domain and observation feedback, as demonstrated by empirical results on the BEIR benchmark.",
}
|
Pre-trained language models (PLMs) exhibit promise in retrieval tasks but struggle with out-of-domain data due to distribution shifts.Addressing this, generative domain adaptation (DA), known as GPL, tackles distribution shifts by generating pseudo queries and labels to train models for predicting query-document relationships in new domains.However, it overlooks the domain distribution, causing the model to struggle with aligning the distribution in the target domain.We, therefore, propose a Distribution-Aware Domain Adaptation (DADA) to guide the model to consider the domain distribution knowledge at the level of both a single document and the corpus, which is referred to as observation-level feedback and domain-level feedback, respectively.Our method effectively adapts the model to the target domain and expands document representation to unseen gold query terms using domain and observation feedback, as demonstrated by empirical results on the BEIR benchmark.
|
[
"Lee, Dohyeon",
"Kim, Jongyoon",
"Hwang, Seung-won",
"Park, Joonsuk"
] |
{DADA}: Distribution-Aware Domain Adaptation of {PLM}s for Information Retrieval
|
findings-acl.825
|
Poster
|
2301.10410v5
|
https://aclanthology.org/2024.findings-acl.826.bib
|
@inproceedings{tyen-etal-2024-llms,
title = "{LLM}s cannot find reasoning errors, but can correct them given the error location",
author = "Tyen, Gladys and
Mansoor, Hassan and
Carbune, Victor and
Chen, Peter and
Mak, Tony",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.826",
pages = "13894--13908",
abstract = "While self-correction has shown promise in improving LLM outputs in terms of style and quality (e.g. Chen et al., 2023b; Madaan et al.,2023), recent attempts to self-correct logical or reasoning errors often cause correct answers to become incorrect, resulting in worse performances overall (Huang et al., 2023). In this paper, we show that poor self-correction performance stems from LLMs{'} inability tofind logical mistakes, rather than their ability to correct a known mistake. Firstly, we benchmark several state-of-the-art LLMs ontheir mistake-finding ability and demonstrate that they generally struggle with the task, even in highly objective, unambiguous cases. Secondly, we test the correction abilities of LLMs {--} separately from mistake finding {--} using a backtracking setup that feeds ground truth mistake location information to the model. We show that this boosts downstream task performance across our 5 reasoning tasks, indicating that LLMs{'} correction abilities are robust. Finally, we show that it is possible to obtain mistake location information without ground truth labels or in-domain training data. We train a small classifier with out-of-domain data, which exhibits stronger mistake-finding performance than prompting a large model. We release our dataset of LLM-generated logical mistakes, BIG-Bench Mistake, to enable further research into locating LLM reasoning mistakes.",
}
|
While self-correction has shown promise in improving LLM outputs in terms of style and quality (e.g. Chen et al., 2023b; Madaan et al.,2023), recent attempts to self-correct logical or reasoning errors often cause correct answers to become incorrect, resulting in worse performances overall (Huang et al., 2023). In this paper, we show that poor self-correction performance stems from LLMs{'} inability tofind logical mistakes, rather than their ability to correct a known mistake. Firstly, we benchmark several state-of-the-art LLMs ontheir mistake-finding ability and demonstrate that they generally struggle with the task, even in highly objective, unambiguous cases. Secondly, we test the correction abilities of LLMs {--} separately from mistake finding {--} using a backtracking setup that feeds ground truth mistake location information to the model. We show that this boosts downstream task performance across our 5 reasoning tasks, indicating that LLMs{'} correction abilities are robust. Finally, we show that it is possible to obtain mistake location information without ground truth labels or in-domain training data. We train a small classifier with out-of-domain data, which exhibits stronger mistake-finding performance than prompting a large model. We release our dataset of LLM-generated logical mistakes, BIG-Bench Mistake, to enable further research into locating LLM reasoning mistakes.
|
[
"Tyen, Gladys",
"Mansoor, Hassan",
"Carbune, Victor",
"Chen, Peter",
"Mak, Tony"
] |
{LLM}s cannot find reasoning errors, but can correct them given the error location
|
findings-acl.826
|
Poster
|
2311.08516v3
|
https://aclanthology.org/2024.findings-acl.827.bib
|
@inproceedings{ranaldi-etal-2024-investigating,
title = "Investigating the Impact of Data Contamination of Large Language Models in Text-to-{SQL} translation",
author = "Ranaldi, Federico and
Ruzzetti, Elena Sofia and
Onorati, Dario and
Ranaldi, Leonardo and
Giannone, Cristina and
Favalli, Andrea and
Romagnoli, Raniero and
Zanzotto, Fabio Massimo",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.827",
pages = "13909--13920",
abstract = "Understanding textual description to generate code seems to be an achieved capability of instruction-following Large Language Models (LLMs) in zero-shot scenario. However, there is a severe possibility that this translation ability may be influenced by having seen target textual descriptions and the related code. This effect is known as Data Contamination.In this study, we investigate the impact of Data Contamination on the performance of GPT-3.5 in the Text-to-SQL code-generating tasks. Hence, we introduce a novel method to detect Data Contamination in GPTs and examine GPT-3.5{'}s Text-to-SQL performances using the known Spider Dataset and our new unfamiliar dataset Termite. Furthermore, we analyze GPT-3.5{'}s efficacy on databases with modified information via an adversarial table disconnection (ATD) approach, complicating Text-to-SQL tasks by removing structural pieces of information from the database. Our results indicate a significant performance drop in GPT-3.5 on the unfamiliar Termite dataset, even with ATD modifications, highlighting the effect of Data Contamination on LLMs in Text-to-SQL translation tasks.",
}
|
Understanding textual description to generate code seems to be an achieved capability of instruction-following Large Language Models (LLMs) in zero-shot scenario. However, there is a severe possibility that this translation ability may be influenced by having seen target textual descriptions and the related code. This effect is known as Data Contamination.In this study, we investigate the impact of Data Contamination on the performance of GPT-3.5 in the Text-to-SQL code-generating tasks. Hence, we introduce a novel method to detect Data Contamination in GPTs and examine GPT-3.5{'}s Text-to-SQL performances using the known Spider Dataset and our new unfamiliar dataset Termite. Furthermore, we analyze GPT-3.5{'}s efficacy on databases with modified information via an adversarial table disconnection (ATD) approach, complicating Text-to-SQL tasks by removing structural pieces of information from the database. Our results indicate a significant performance drop in GPT-3.5 on the unfamiliar Termite dataset, even with ATD modifications, highlighting the effect of Data Contamination on LLMs in Text-to-SQL translation tasks.
|
[
"Ranaldi, Federico",
"Ruzzetti, Elena Sofia",
"Onorati, Dario",
"Ranaldi, Leonardo",
"Giannone, Cristina",
"Favalli, Andrea",
"Romagnoli, Raniero",
"Zanzotto, Fabio Massimo"
] |
Investigating the Impact of Data Contamination of Large Language Models in Text-to-{SQL} translation
|
findings-acl.827
|
Poster
|
2402.08100v1
|
https://aclanthology.org/2024.findings-acl.828.bib
|
@inproceedings{akhtar-etal-2024-chartcheck,
title = "{C}hart{C}heck: Explainable Fact-Checking over Real-World Chart Images",
author = "Akhtar, Mubashara and
Subedi, Nikesh and
Gupta, Vivek and
Tahmasebi, Sahar and
Cocarascu, Oana and
Simperl, Elena",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.828",
pages = "13921--13937",
abstract = "Whilst fact verification has attracted substantial interest in the natural language processing community, verifying misinforming statements against data visualizations such as charts has so far been overlooked. Charts are commonly used in the real-world to summarize and com municate key information, but they can also be easily misused to spread misinformation and promote certain agendas. In this paper, we introduce ChartCheck, a novel, large-scale dataset for explainable fact-checking against real-world charts, consisting of 1.7k charts and 10.5k human-written claims and explanations. We systematically evaluate ChartCheck using vision-language and chart-to-table models, and propose a baseline to the community. Finally, we study chart reasoning types and visual attributes that pose a challenge to these models.",
}
|
Whilst fact verification has attracted substantial interest in the natural language processing community, verifying misinforming statements against data visualizations such as charts has so far been overlooked. Charts are commonly used in the real-world to summarize and com municate key information, but they can also be easily misused to spread misinformation and promote certain agendas. In this paper, we introduce ChartCheck, a novel, large-scale dataset for explainable fact-checking against real-world charts, consisting of 1.7k charts and 10.5k human-written claims and explanations. We systematically evaluate ChartCheck using vision-language and chart-to-table models, and propose a baseline to the community. Finally, we study chart reasoning types and visual attributes that pose a challenge to these models.
|
[
"Akhtar, Mubashara",
"Subedi, Nikesh",
"Gupta, Vivek",
"Tahmasebi, Sahar",
"Cocarascu, Oana",
"Simperl, Elena"
] |
{C}hart{C}heck: Explainable Fact-Checking over Real-World Chart Images
|
findings-acl.828
|
Poster
|
2311.10774v2
|
https://aclanthology.org/2024.findings-acl.829.bib
|
@inproceedings{hoveyda-etal-2024-real,
title = "Real World Conversational Entity Linking Requires More Than Zero-Shots",
author = "Hoveyda, Mohanna and
Vries, Arjen and
Hasibi, Faegheh and
Rijke, Maarten",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.829",
pages = "13938--13946",
abstract = "Entity linking (EL) in conversations faces notable challenges in practical applications, primarily due to scarcity of entity-annotated conversational datasets and sparse knowledge bases (KB) containing domain-specific, long-tail entities. We designed targeted evaluation scenarios to measure the efficacy of EL models under resource constraints. Our evaluation employs two KBs: Fandom, exemplifying real-world EL complexities, and the widely used Wikipedia. First, we assess EL models{'} ability to generalize to a new unfamiliar KB using Fandom and a novel zero-shot conversational entity linking dataset that we curated based on Reddit discussions on Fandom entities. We then evaluate the adaptability of EL models to conversational settings without prior training. Our results indicate that current zero-shot EL models falter when introduced to new, domain-specific KBs without prior training, significantly dropping in performance.Our findings reveal that previous evaluation approaches fall short of capturing real-world complexities for zero-shot EL, highlighting the necessity for new approaches to design and assess conversational EL models to adapt to limited resources. The evaluation frame-work and dataset proposed are tailored to facilitate this research.",
}
|
Entity linking (EL) in conversations faces notable challenges in practical applications, primarily due to scarcity of entity-annotated conversational datasets and sparse knowledge bases (KB) containing domain-specific, long-tail entities. We designed targeted evaluation scenarios to measure the efficacy of EL models under resource constraints. Our evaluation employs two KBs: Fandom, exemplifying real-world EL complexities, and the widely used Wikipedia. First, we assess EL models{'} ability to generalize to a new unfamiliar KB using Fandom and a novel zero-shot conversational entity linking dataset that we curated based on Reddit discussions on Fandom entities. We then evaluate the adaptability of EL models to conversational settings without prior training. Our results indicate that current zero-shot EL models falter when introduced to new, domain-specific KBs without prior training, significantly dropping in performance.Our findings reveal that previous evaluation approaches fall short of capturing real-world complexities for zero-shot EL, highlighting the necessity for new approaches to design and assess conversational EL models to adapt to limited resources. The evaluation frame-work and dataset proposed are tailored to facilitate this research.
|
[
"Hoveyda, Mohanna",
"Vries, Arjen",
"Hasibi, Faegheh",
"Rijke, Maarten"
] |
Real World Conversational Entity Linking Requires More Than Zero-Shots
|
findings-acl.829
|
Poster
|
2101.11204v2
|
https://aclanthology.org/2024.findings-acl.830.bib
|
@inproceedings{zhang-etal-2024-cpsycoun,
title = "{CP}sy{C}oun: A Report-based Multi-turn Dialogue Reconstruction and Evaluation Framework for {C}hinese Psychological Counseling",
author = "Zhang, Chenhao and
Li, Renhao and
Tan, Minghuan and
Yang, Min and
Zhu, Jingwei and
Yang, Di and
Zhao, Jiahao and
Ye, Guancheng and
Li, Chengming and
Hu, Xiping",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.830",
pages = "13947--13966",
abstract = "Using large language models (LLMs) to assist psychological counseling is a significant but challenging task at present. Attempts have been made on improving empathetic conversations or acting as effective assistants in the treatment with LLMs. However, the existing datasets lack consulting knowledge, resulting in LLMs lacking professional consulting competence. Moreover, how to automatically evaluate multi-turn dialogues within the counseling process remains an understudied area. To bridge the gap, we propose CPsyCoun, a report-based multi-turn dialogue reconstruction and evaluation framework for Chinese psychological counseling. To fully exploit psychological counseling reports, a two-phase approach is devised to construct high-quality dialogues while a comprehensive evaluation benchmark is developed for the effective automatic evaluation of multi-turn psychological consultations. Competitive experimental results demonstrate the effectiveness of our proposed framework in psychological counseling. We open-source the datasets and model for future research.",
}
|
Using large language models (LLMs) to assist psychological counseling is a significant but challenging task at present. Attempts have been made on improving empathetic conversations or acting as effective assistants in the treatment with LLMs. However, the existing datasets lack consulting knowledge, resulting in LLMs lacking professional consulting competence. Moreover, how to automatically evaluate multi-turn dialogues within the counseling process remains an understudied area. To bridge the gap, we propose CPsyCoun, a report-based multi-turn dialogue reconstruction and evaluation framework for Chinese psychological counseling. To fully exploit psychological counseling reports, a two-phase approach is devised to construct high-quality dialogues while a comprehensive evaluation benchmark is developed for the effective automatic evaluation of multi-turn psychological consultations. Competitive experimental results demonstrate the effectiveness of our proposed framework in psychological counseling. We open-source the datasets and model for future research.
|
[
"Zhang, Chenhao",
"Li, Renhao",
"Tan, Minghuan",
"Yang, Min",
"Zhu, Jingwei",
"Yang, Di",
"Zhao, Jiahao",
"Ye, Guancheng",
"Li, Chengming",
"Hu, Xiping"
] |
{CP}sy{C}oun: A Report-based Multi-turn Dialogue Reconstruction and Evaluation Framework for {C}hinese Psychological Counseling
|
findings-acl.830
|
Poster
|
2405.16433v3
|
https://aclanthology.org/2024.findings-acl.831.bib
|
@inproceedings{yadav-etal-2024-tox,
title = "Tox-{BART}: Leveraging Toxicity Attributes for Explanation Generation of Implicit Hate Speech",
author = "Yadav, Neemesh and
Masud, Sarah and
Goyal, Vikram and
Akhtar, Md Shad and
Chakraborty, Tanmoy",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.831",
pages = "13967--13983",
abstract = "Employing language models to generate explanations for an incoming implicit hate post is an active area of research. The explanation is intended to make explicit the underlying stereotype and aid content moderators. The training often combines top-k relevant knowledge graph (KG) tuples to provide world knowledge and improve performance on standard metrics. Interestingly, our study presents conflicting evidence for the role of the quality of KG tuples in generating implicit explanations. Consequently, simpler models incorporating external toxicity signals outperform KG-infused models. Compared to the KG-based setup, we observe a comparable performance for SBIC (LatentHatred) datasets with a performance variation of +0.44 (+0.49), +1.83 (-1.56), and -4.59 (+0.77) in BLEU, ROUGE-L, and BERTScore. Further human evaluation and error analysis reveal that our proposed setup produces more precise explanations than zero-shot GPT-3.5, highlighting the intricate nature of the task.",
}
|
Employing language models to generate explanations for an incoming implicit hate post is an active area of research. The explanation is intended to make explicit the underlying stereotype and aid content moderators. The training often combines top-k relevant knowledge graph (KG) tuples to provide world knowledge and improve performance on standard metrics. Interestingly, our study presents conflicting evidence for the role of the quality of KG tuples in generating implicit explanations. Consequently, simpler models incorporating external toxicity signals outperform KG-infused models. Compared to the KG-based setup, we observe a comparable performance for SBIC (LatentHatred) datasets with a performance variation of +0.44 (+0.49), +1.83 (-1.56), and -4.59 (+0.77) in BLEU, ROUGE-L, and BERTScore. Further human evaluation and error analysis reveal that our proposed setup produces more precise explanations than zero-shot GPT-3.5, highlighting the intricate nature of the task.
|
[
"Yadav, Neemesh",
"Masud, Sarah",
"Goyal, Vikram",
"Akhtar, Md Shad",
"Chakraborty, Tanmoy"
] |
Tox-{BART}: Leveraging Toxicity Attributes for Explanation Generation of Implicit Hate Speech
|
findings-acl.831
|
Poster
|
2406.03953v1
|
https://aclanthology.org/2024.findings-acl.832.bib
|
@inproceedings{enouen-etal-2024-textgenshap,
title = "{T}ext{G}en{SHAP}: Scalable Post-Hoc Explanations in Text Generation with Long Documents",
author = "Enouen, James and
Nakhost, Hootan and
Ebrahimi, Sayna and
Arik, Sercan and
Liu, Yan and
Pfister, Tomas",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.832",
pages = "13984--14011",
abstract = "Large language models (LLMs) have attracted great interest in many real-world applications; however, their {``}black-box{''} nature necessitates scalable and faithful explanations. Shapley values have matured as an explainability method for deep learning, but extending them to LLMs is difficult due to long input contexts and autoregressive output generation. We introduce , an efficient post-hoc explanation method incorporating LLM-specific techniques, which leads to significant runtime improvements: token-level explanations in minutes not hours, and document-level explanations within seconds. We demonstrate how such explanations can improve end-to-end performance of retrieval augmented generation by localizing important words within long documents and reranking passages collected by retrieval systems. On various open-domain question answering benchmarks, we show TextGenSHAP improves the retrieval recall and prediction accuracy significantly.",
}
|
Large language models (LLMs) have attracted great interest in many real-world applications; however, their {``}black-box{''} nature necessitates scalable and faithful explanations. Shapley values have matured as an explainability method for deep learning, but extending them to LLMs is difficult due to long input contexts and autoregressive output generation. We introduce , an efficient post-hoc explanation method incorporating LLM-specific techniques, which leads to significant runtime improvements: token-level explanations in minutes not hours, and document-level explanations within seconds. We demonstrate how such explanations can improve end-to-end performance of retrieval augmented generation by localizing important words within long documents and reranking passages collected by retrieval systems. On various open-domain question answering benchmarks, we show TextGenSHAP improves the retrieval recall and prediction accuracy significantly.
|
[
"Enouen, James",
"Nakhost, Hootan",
"Ebrahimi, Sayna",
"Arik, Sercan",
"Liu, Yan",
"Pfister, Tomas"
] |
{T}ext{G}en{SHAP}: Scalable Post-Hoc Explanations in Text Generation with Long Documents
|
findings-acl.832
|
Poster
|
2312.01279v1
|
https://aclanthology.org/2024.findings-acl.833.bib
|
@inproceedings{shao-etal-2024-balanced,
title = "Balanced Data Sampling for Language Model Training with Clustering",
author = "Shao, Yunfan and
Li, Linyang and
Fei, Zhaoye and
Yan, Hang and
Lin, Dahua and
Qiu, Xipeng",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.833",
pages = "14012--14023",
abstract = "Data plays a fundamental role in the training of Large Language Models (LLMs). While attention has been paid to the collection and composition of datasets, determining the data sampling strategy in training remains an open question. Most LLMs are trained with a simple strategy, random sampling. However, this sampling strategy ignores the unbalanced nature of training data distribution, which can be sub-optimal. In this paper, we propose ClusterClip Sampling to balance the text distribution of training data for better model training. Specifically, ClusterClip Sampling utilizes data clustering to reflect the data distribution of the training set and balances the common samples and rare samples during training based on the cluster results. A repetition clip operation is introduced to mitigate the overfitting issue led by samples from certain clusters. Extensive experiments validate the effectiveness of ClusterClip Sampling, which outperforms random sampling and other cluster-based sampling variants under various training datasets and large language models.",
}
|
Data plays a fundamental role in the training of Large Language Models (LLMs). While attention has been paid to the collection and composition of datasets, determining the data sampling strategy in training remains an open question. Most LLMs are trained with a simple strategy, random sampling. However, this sampling strategy ignores the unbalanced nature of training data distribution, which can be sub-optimal. In this paper, we propose ClusterClip Sampling to balance the text distribution of training data for better model training. Specifically, ClusterClip Sampling utilizes data clustering to reflect the data distribution of the training set and balances the common samples and rare samples during training based on the cluster results. A repetition clip operation is introduced to mitigate the overfitting issue led by samples from certain clusters. Extensive experiments validate the effectiveness of ClusterClip Sampling, which outperforms random sampling and other cluster-based sampling variants under various training datasets and large language models.
|
[
"Shao, Yunfan",
"Li, Linyang",
"Fei, Zhaoye",
"Yan, Hang",
"Lin, Dahua",
"Qiu, Xipeng"
] |
Balanced Data Sampling for Language Model Training with Clustering
|
findings-acl.833
|
Poster
|
2402.14526v2
|
https://aclanthology.org/2024.findings-acl.834.bib
|
@inproceedings{wang-etal-2024-length,
title = "Length Generalization of Causal Transformers without Position Encoding",
author = "Wang, Jie and
Ji, Tao and
Wu, Yuanbin and
Yan, Hang and
Gui, Tao and
Zhang, Qi and
Huang, Xuanjing and
Wang, Xiaoling",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.834",
pages = "14024--14040",
abstract = "Generalizing to longer sentences is important for recent Transformer-based language models. Besides algorithms manipulating explicit position features, the success of Transformers without position encodings (NoPE) provides a new way to overcome the challenge. In this paper, we study the length generalization property of NoPE. We find that although NoPE can extend to longer sequences than the commonly used explicit position encodings, it still has a limited context length. We identify a connection between the failure of NoPE{'}s generalization and the distraction of attention distributions. We propose a parameter-efficient tuning for searching attention heads{'} best temperature hyper-parameters, which substantially expands NoPE{'}s context size. Experiments on long sequence language modeling, the synthetic passkey retrieval task and real-world long context tasks show that NoPE can achieve competitive performances with state-of-the-art length generalization algorithms. The source code is publicly accessible",
}
|
Generalizing to longer sentences is important for recent Transformer-based language models. Besides algorithms manipulating explicit position features, the success of Transformers without position encodings (NoPE) provides a new way to overcome the challenge. In this paper, we study the length generalization property of NoPE. We find that although NoPE can extend to longer sequences than the commonly used explicit position encodings, it still has a limited context length. We identify a connection between the failure of NoPE{'}s generalization and the distraction of attention distributions. We propose a parameter-efficient tuning for searching attention heads{'} best temperature hyper-parameters, which substantially expands NoPE{'}s context size. Experiments on long sequence language modeling, the synthetic passkey retrieval task and real-world long context tasks show that NoPE can achieve competitive performances with state-of-the-art length generalization algorithms. The source code is publicly accessible
|
[
"Wang, Jie",
"Ji, Tao",
"Wu, Yuanbin",
"Yan, Hang",
"Gui, Tao",
"Zhang, Qi",
"Huang, Xuanjing",
"Wang, Xiaoling"
] |
Length Generalization of Causal Transformers without Position Encoding
|
findings-acl.834
|
Poster
|
2404.12224v2
|
https://aclanthology.org/2024.findings-acl.835.bib
|
@inproceedings{guo-etal-2024-unsupervised,
title = "Unsupervised Sign Language Translation and Generation",
author = "Guo, Zhengsheng and
He, Zhiwei and
Jiao, Wenxiang and
Wang, Xing and
Wang, Rui and
Chen, Kehai and
Tu, Zhaopeng and
Xu, Yong and
Zhang, Min",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.835",
pages = "14041--14055",
abstract = "Motivated by the success of unsupervised neural machine translation (UNMT), we introduce an unsupervised sign language translation and generation network (USLNet), which learns from abundant single-modality (text and video) data without parallel sign language data. USLNet comprises two main components: single-modality reconstruction modules (text and video) that rebuild the input from its noisy version in the same modality and cross-modality back-translation modules (text-video-text and video-text-video) that reconstruct the input from its noisy version in the different modality using back-translation procedure. Unlike the single-modality back-translation procedure in text-based UNMT, USLNet faces the cross-modality discrepancy in feature representation, in which the length and the feature dimension mismatch between text and video sequences. We propose a sliding window method to address the issues of aligning variable-length text with video sequences. To our knowledge, USLNet is the first unsupervised sign language translation and generation model capable of generating both natural language text and sign language video in a unified manner. Experimental results on the BBC-Oxford Sign Language dataset and Open-Domain American Sign Language dataset reveal that USLNet achieves competitive results compared to supervised baseline models, indicating its effectiveness in sign language translation and generation.",
}
|
Motivated by the success of unsupervised neural machine translation (UNMT), we introduce an unsupervised sign language translation and generation network (USLNet), which learns from abundant single-modality (text and video) data without parallel sign language data. USLNet comprises two main components: single-modality reconstruction modules (text and video) that rebuild the input from its noisy version in the same modality and cross-modality back-translation modules (text-video-text and video-text-video) that reconstruct the input from its noisy version in the different modality using back-translation procedure. Unlike the single-modality back-translation procedure in text-based UNMT, USLNet faces the cross-modality discrepancy in feature representation, in which the length and the feature dimension mismatch between text and video sequences. We propose a sliding window method to address the issues of aligning variable-length text with video sequences. To our knowledge, USLNet is the first unsupervised sign language translation and generation model capable of generating both natural language text and sign language video in a unified manner. Experimental results on the BBC-Oxford Sign Language dataset and Open-Domain American Sign Language dataset reveal that USLNet achieves competitive results compared to supervised baseline models, indicating its effectiveness in sign language translation and generation.
|
[
"Guo, Zhengsheng",
"He, Zhiwei",
"Jiao, Wenxiang",
"Wang, Xing",
"Wang, Rui",
"Chen, Kehai",
"Tu, Zhaopeng",
"Xu, Yong",
"Zhang, Min"
] |
Unsupervised Sign Language Translation and Generation
|
findings-acl.835
|
Poster
|
2402.07726v1
|
https://aclanthology.org/2024.findings-acl.836.bib
|
@inproceedings{martinez-lorenzo-etal-2024-mitigating,
title = "Mitigating Data Scarcity in Semantic Parsing across Languages with the Multilingual Semantic Layer and its Dataset",
author = "Martinez Lorenzo, Abelardo Carlos and
Huguet Cabot, Pere-Llu{\'\i}s and
Ghonim, Karim and
Xu, Lu and
Choi, Hee-Soo and
Fern{\'a}ndez-Castro, Alberte and
Navigli, Roberto",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.836",
pages = "14056--14080",
abstract = "Data scarcity is a prevalent challenge in the era of Large Language Models (LLMs). The insatiable hunger of LLMs for large corpora becomes even more pronounced when dealing with non-English and low-resource languages. The issue is particularly exacerbated in Semantic Parsing (SP), i.e. the task of converting text into a formal representation. The complexity of semantic formalisms makes training human annotators and subsequent data annotation unfeasible on a large scale, especially across languages. To mitigate this, we first introduce the Multilingual Semantic Layer (MSL), a conceptual evolution of previous formalisms, which decouples from disambiguation and external inventories and simplifies the task. MSL provides the necessary tools to encode the meaning across languages, paving the way for developing a high-quality semantic parsing dataset across different languages in a semi-automatic strategy. Subsequently, we manually refine a portion of this dataset and fine-tune GPT-3.5 to propagate these refinements across the dataset. Then, we manually annotate 1,100 sentences in eleven languages, including low-resource ones. Finally, we assess our dataset{'}s quality, showcasing the performance gap reduction across languages in Semantic Parsing.",
}
|
Data scarcity is a prevalent challenge in the era of Large Language Models (LLMs). The insatiable hunger of LLMs for large corpora becomes even more pronounced when dealing with non-English and low-resource languages. The issue is particularly exacerbated in Semantic Parsing (SP), i.e. the task of converting text into a formal representation. The complexity of semantic formalisms makes training human annotators and subsequent data annotation unfeasible on a large scale, especially across languages. To mitigate this, we first introduce the Multilingual Semantic Layer (MSL), a conceptual evolution of previous formalisms, which decouples from disambiguation and external inventories and simplifies the task. MSL provides the necessary tools to encode the meaning across languages, paving the way for developing a high-quality semantic parsing dataset across different languages in a semi-automatic strategy. Subsequently, we manually refine a portion of this dataset and fine-tune GPT-3.5 to propagate these refinements across the dataset. Then, we manually annotate 1,100 sentences in eleven languages, including low-resource ones. Finally, we assess our dataset{'}s quality, showcasing the performance gap reduction across languages in Semantic Parsing.
|
[
"Martinez Lorenzo, Abelardo Carlos",
"Huguet Cabot, Pere-Llu{\\'\\i}s",
"Ghonim, Karim",
"Xu, Lu",
"Choi, Hee-Soo",
"Fern{\\'a}ndez-Castro, Alberte",
"Navigli, Roberto"
] |
Mitigating Data Scarcity in Semantic Parsing across Languages with the Multilingual Semantic Layer and its Dataset
|
findings-acl.836
|
Poster
|
2108.03509v2
|
https://aclanthology.org/2024.findings-acl.837.bib
|
@inproceedings{zhang-etal-2024-efficient,
title = "Efficient Sparse Attention needs Adaptive Token Release",
author = "Zhang, Chaoran and
Zou, Lixin and
Luo, Dan and
Luo, Xiangyang and
Li, Zihao and
Tang, Min and
Li, Chenliang",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.837",
pages = "14081--14094",
abstract = "In recent years, Large Language Models (LLMs) have demonstrated remarkable capabilities across a wide array of text-centric tasks. However, their {`}large{'} scale introduces significant computational and storage challenges, particularly in managing the key-value states of the transformer, which limits their wider applicability. Therefore, we propose to adaptively release resources from caches and rebuild the necessary key-value states. Particularly, we accomplish this by a lightweight controller module to approximate an ideal top-$K$ sparse attention. This module retains the tokens with the highest top-$K$ attention weights and simultaneously rebuilds the discarded but necessary tokens, which may become essential for future decoding. Comprehensive experiments in natural language generation and modeling reveal that our method is not only competitive with full attention in terms of performance but also achieves a significant throughput improvement of up to $\textbf{221.8}${\%}. The code for replication is available on the https://github.com/WHUIR/ADORE.",
}
|
In recent years, Large Language Models (LLMs) have demonstrated remarkable capabilities across a wide array of text-centric tasks. However, their {`}large{'} scale introduces significant computational and storage challenges, particularly in managing the key-value states of the transformer, which limits their wider applicability. Therefore, we propose to adaptively release resources from caches and rebuild the necessary key-value states. Particularly, we accomplish this by a lightweight controller module to approximate an ideal top-$K$ sparse attention. This module retains the tokens with the highest top-$K$ attention weights and simultaneously rebuilds the discarded but necessary tokens, which may become essential for future decoding. Comprehensive experiments in natural language generation and modeling reveal that our method is not only competitive with full attention in terms of performance but also achieves a significant throughput improvement of up to $\textbf{221.8}${\%}. The code for replication is available on the https://github.com/WHUIR/ADORE.
|
[
"Zhang, Chaoran",
"Zou, Lixin",
"Luo, Dan",
"Luo, Xiangyang",
"Li, Zihao",
"Tang, Min",
"Li, Chenliang"
] |
Efficient Sparse Attention needs Adaptive Token Release
|
findings-acl.837
|
Poster
|
2407.02328v1
|
https://aclanthology.org/2024.findings-acl.838.bib
|
@inproceedings{huang-etal-2024-learning,
title = "Learning Fine-Grained Grounded Citations for Attributed Large Language Models",
author = "Huang, Lei and
Feng, Xiaocheng and
Ma, Weitao and
Gu, Yuxuan and
Zhong, Weihong and
Feng, Xiachong and
Yu, Weijiang and
Peng, Weihua and
Tang, Duyu and
Tu, Dandan and
Qin, Bing",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.838",
pages = "14095--14113",
abstract = "Despite the impressive performance on information-seeking tasks, large language models (LLMs) still struggle with hallucinations. Attributed LLMs, which augment generated text with in-line citations, demonstrate potential in mitigating hallucinations and improving verifiability. However, current approaches suffer from suboptimal citation quality due to their reliance on in-context learning. Furthermore, the practice of merely citing document identifiers complicates the process for users to pinpoint specific supporting evidence. In this work, we introduce FRONT, a training framework that teaches LLMs to generate Fine-grained grounded citations. By initially grounding fine-grained supporting quotes, which then guide the generation process, these quotes not only provide supervision signals to improve citation quality but also serve as fine-grained attributions. Experiments on the ALCE benchmark demonstrate the efficacy of FRONT in generating superior grounded responses and highly supportive citations. With LLaMA-2-7B, the framework significantly outperforms all the baselines, achieving an average of 14.21{\%} improvement in citation quality across all datasets, even surpassing ChatGPT.",
}
|
Despite the impressive performance on information-seeking tasks, large language models (LLMs) still struggle with hallucinations. Attributed LLMs, which augment generated text with in-line citations, demonstrate potential in mitigating hallucinations and improving verifiability. However, current approaches suffer from suboptimal citation quality due to their reliance on in-context learning. Furthermore, the practice of merely citing document identifiers complicates the process for users to pinpoint specific supporting evidence. In this work, we introduce FRONT, a training framework that teaches LLMs to generate Fine-grained grounded citations. By initially grounding fine-grained supporting quotes, which then guide the generation process, these quotes not only provide supervision signals to improve citation quality but also serve as fine-grained attributions. Experiments on the ALCE benchmark demonstrate the efficacy of FRONT in generating superior grounded responses and highly supportive citations. With LLaMA-2-7B, the framework significantly outperforms all the baselines, achieving an average of 14.21{\%} improvement in citation quality across all datasets, even surpassing ChatGPT.
|
[
"Huang, Lei",
"Feng, Xiaocheng",
"Ma, Weitao",
"Gu, Yuxuan",
"Zhong, Weihong",
"Feng, Xiachong",
"Yu, Weijiang",
"Peng, Weihua",
"Tang, Duyu",
"Tu, D",
"an",
"Qin, Bing"
] |
Learning Fine-Grained Grounded Citations for Attributed Large Language Models
|
findings-acl.838
|
Poster
|
2408.04568v1
|
https://aclanthology.org/2024.findings-acl.839.bib
|
@inproceedings{orlando-etal-2024-relik,
title = "{R}e{L}i{K}: Retrieve and {L}in{K}, Fast and Accurate Entity Linking and Relation Extraction on an Academic Budget",
author = "Orlando, Riccardo and
Huguet Cabot, Pere-Llu{\'\i}s and
Barba, Edoardo and
Navigli, Roberto",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.839",
pages = "14114--14132",
abstract = "Entity Linking (EL) and Relation Extraction (RE) are fundamental tasks in Natural Language Processing, serving as critical components in a wide range of applications. In this paper, we propose ReLiK, a Retriever-Reader architecture for both EL and RE, where, given an input text, the Retriever module undertakes the identification of candidate entities or relations that could potentially appear within the text. Subsequently, the Reader module is tasked to discern the pertinent retrieved entities or relations and establish their alignment with the corresponding textual spans. Notably, we put forward an innovative input representation that incorporates the candidate entities or relations alongside the text, making it possible to link entities or extract relations in a single forward pass and to fully leverage pre-trained language models contextualization capabilities, in contrast with previous Retriever-Reader-based methods, which require a forward pass for each candidate. Our formulation of EL and RE achieves state-of-the-art performance in both in-domain and out-of-domain benchmarks while using academic budget training and with up to 40x inference speed compared to competitors. Finally, we show how our architecture can be used seamlessly for Information Extraction (cIE), i.e. EL + RE, and setting a new state of the art by employing a shared Reader that simultaneously extracts entities and relations.",
}
|
Entity Linking (EL) and Relation Extraction (RE) are fundamental tasks in Natural Language Processing, serving as critical components in a wide range of applications. In this paper, we propose ReLiK, a Retriever-Reader architecture for both EL and RE, where, given an input text, the Retriever module undertakes the identification of candidate entities or relations that could potentially appear within the text. Subsequently, the Reader module is tasked to discern the pertinent retrieved entities or relations and establish their alignment with the corresponding textual spans. Notably, we put forward an innovative input representation that incorporates the candidate entities or relations alongside the text, making it possible to link entities or extract relations in a single forward pass and to fully leverage pre-trained language models contextualization capabilities, in contrast with previous Retriever-Reader-based methods, which require a forward pass for each candidate. Our formulation of EL and RE achieves state-of-the-art performance in both in-domain and out-of-domain benchmarks while using academic budget training and with up to 40x inference speed compared to competitors. Finally, we show how our architecture can be used seamlessly for Information Extraction (cIE), i.e. EL + RE, and setting a new state of the art by employing a shared Reader that simultaneously extracts entities and relations.
|
[
"Orl",
"o, Riccardo",
"Huguet Cabot, Pere-Llu{\\'\\i}s",
"Barba, Edoardo",
"Navigli, Roberto"
] |
{R}e{L}i{K}: Retrieve and {L}in{K}, Fast and Accurate Entity Linking and Relation Extraction on an Academic Budget
|
findings-acl.839
|
Poster
|
2408.00103v1
|
https://aclanthology.org/2024.findings-acl.840.bib
|
@inproceedings{liang-etal-2024-synergizing,
title = "Synergizing Large Language Models and Pre-Trained Smaller Models for Conversational Intent Discovery",
author = "Liang, Jinggui and
Liao, Lizi and
Fei, Hao and
Jiang, Jing",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.840",
pages = "14133--14147",
abstract = "In Conversational Intent Discovery (CID), Small Language Models (SLMs) struggle with overfitting to familiar intents and fail to label newly discovered ones. This issue stems from their limited grasp of semantic nuances and their intrinsically discriminative framework. Therefore, we propose Synergizing Large Language Models (LLMs) with pre-trained SLMs for CID (SynCID). It harnesses the profound semantic comprehension of LLMs alongside the operational agility of SLMs. By utilizing LLMs to refine both utterances and existing intent labels, SynCID significantly enhances the semantic depth, subsequently realigning these enriched descriptors within the SLMs{'} feature space to correct cluster distortion and promote robust learning of representations. A key advantage is its capacity for the early identification of new intents, a critical aspect for deploying conversational agents successfully. Additionally, SynCID leverages the in-context learning strengths of LLMs to generate labels for new intents. Thorough evaluations across a wide array of datasets have demonstrated its superior performance over traditional CID methods.",
}
|
In Conversational Intent Discovery (CID), Small Language Models (SLMs) struggle with overfitting to familiar intents and fail to label newly discovered ones. This issue stems from their limited grasp of semantic nuances and their intrinsically discriminative framework. Therefore, we propose Synergizing Large Language Models (LLMs) with pre-trained SLMs for CID (SynCID). It harnesses the profound semantic comprehension of LLMs alongside the operational agility of SLMs. By utilizing LLMs to refine both utterances and existing intent labels, SynCID significantly enhances the semantic depth, subsequently realigning these enriched descriptors within the SLMs{'} feature space to correct cluster distortion and promote robust learning of representations. A key advantage is its capacity for the early identification of new intents, a critical aspect for deploying conversational agents successfully. Additionally, SynCID leverages the in-context learning strengths of LLMs to generate labels for new intents. Thorough evaluations across a wide array of datasets have demonstrated its superior performance over traditional CID methods.
|
[
"Liang, Jinggui",
"Liao, Lizi",
"Fei, Hao",
"Jiang, Jing"
] |
Synergizing Large Language Models and Pre-Trained Smaller Models for Conversational Intent Discovery
|
findings-acl.840
|
Poster
|
2305.19783v1
|
https://aclanthology.org/2024.findings-acl.841.bib
|
@inproceedings{scire-etal-2024-fenice,
title = "{FENICE}: Factuality Evaluation of summarization based on Natural language Inference and Claim Extraction",
author = "Scir{\`e}, Alessandro and
Ghonim, Karim and
Navigli, Roberto",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.841",
pages = "14148--14161",
abstract = "Recent advancements in text summarization, particularly with the advent of Large Language Models (LLMs), have shown remarkable performance. However, a notable challenge persists as a substantial number of automatically-generated summaries exhibit factual inconsistencies, such as hallucinations. In response to this issue, various approaches for the evaluation of consistency for summarization have emerged. Yet, these newly-introduced metrics face several limitations, including lack of interpretability, focus on short document summaries (e.g., news articles), and computational impracticality, especially for LLM-based metrics. To address these shortcomings, we propose Factuality Evaluation of summarization based on Natural language Inference and Claim Extraction (FENICE), a more interpretable and efficient factuality-oriented metric. FENICE leverages an NLI-based alignment between information in the source document and a set of atomic facts, referred to as claims, extracted from the summary. Our metric sets a new state of the art on AGGREFACT, the de-facto benchmark for factuality evaluation. Moreover, we extend our evaluation to a more challenging setting by conducting a human annotation process of long-form summarization. In the hope of fostering research in summarization factuality evaluation, we release the code of our metric and our factuality annotations of long-form summarization at anonymizedurl.",
}
|
Recent advancements in text summarization, particularly with the advent of Large Language Models (LLMs), have shown remarkable performance. However, a notable challenge persists as a substantial number of automatically-generated summaries exhibit factual inconsistencies, such as hallucinations. In response to this issue, various approaches for the evaluation of consistency for summarization have emerged. Yet, these newly-introduced metrics face several limitations, including lack of interpretability, focus on short document summaries (e.g., news articles), and computational impracticality, especially for LLM-based metrics. To address these shortcomings, we propose Factuality Evaluation of summarization based on Natural language Inference and Claim Extraction (FENICE), a more interpretable and efficient factuality-oriented metric. FENICE leverages an NLI-based alignment between information in the source document and a set of atomic facts, referred to as claims, extracted from the summary. Our metric sets a new state of the art on AGGREFACT, the de-facto benchmark for factuality evaluation. Moreover, we extend our evaluation to a more challenging setting by conducting a human annotation process of long-form summarization. In the hope of fostering research in summarization factuality evaluation, we release the code of our metric and our factuality annotations of long-form summarization at anonymizedurl.
|
[
"Scir{\\`e}, Aless",
"ro",
"Ghonim, Karim",
"Navigli, Roberto"
] |
{FENICE}: Factuality Evaluation of summarization based on Natural language Inference and Claim Extraction
|
findings-acl.841
|
Poster
|
2403.02270v2
|
https://aclanthology.org/2024.findings-acl.842.bib
|
@inproceedings{chen-etal-2024-self-para,
title = "Self-Para-Consistency: Improving Reasoning Tasks at Low Cost for Large Language Models",
author = "Chen, Wenqing and
Wang, Weicheng and
Chu, Zhixuan and
Ren, Kui and
Zheng, Zibin and
Lu, Zhichao",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.842",
pages = "14162--14167",
abstract = "Recently, the self-consistency decoding strategy has shown the ability to improve performance for complex reasoning tasks with large language models (LLMs). However, the costs may be high because the sampling process of the strategy generates some low-probability text, resulting in low-quality reasoning paths. As a consequence, it requires a relatively large sampling number to obtain good aggregation performance. In this paper, we propose an alternative strategy, \textit{self-para-consistency}. It first generates multiple paraphrases for each test question, then generates reasoning paths for the original and all the paraphrased questions based on greedy decoding, and finally selects the most consistent answer. Since all the candidate paths have relatively high probabilities, the sampling number could be much smaller than the self-consistency strategy. Extensive experiments on complex reasoning datasets demonstrate the effectiveness of our method in reducing the sampling number.",
}
|
Recently, the self-consistency decoding strategy has shown the ability to improve performance for complex reasoning tasks with large language models (LLMs). However, the costs may be high because the sampling process of the strategy generates some low-probability text, resulting in low-quality reasoning paths. As a consequence, it requires a relatively large sampling number to obtain good aggregation performance. In this paper, we propose an alternative strategy, \textit{self-para-consistency}. It first generates multiple paraphrases for each test question, then generates reasoning paths for the original and all the paraphrased questions based on greedy decoding, and finally selects the most consistent answer. Since all the candidate paths have relatively high probabilities, the sampling number could be much smaller than the self-consistency strategy. Extensive experiments on complex reasoning datasets demonstrate the effectiveness of our method in reducing the sampling number.
|
[
"Chen, Wenqing",
"Wang, Weicheng",
"Chu, Zhixuan",
"Ren, Kui",
"Zheng, Zibin",
"Lu, Zhichao"
] |
Self-Para-Consistency: Improving Reasoning Tasks at Low Cost for Large Language Models
|
findings-acl.842
|
Poster
|
2312.02433v1
|
https://aclanthology.org/2024.findings-acl.843.bib
|
@inproceedings{dukic-snajder-2024-looking,
title = "Looking Right is Sometimes Right: Investigating the Capabilities of Decoder-only {LLM}s for Sequence Labeling",
author = "Duki{\'c}, David and
Snajder, Jan",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.843",
pages = "14168--14181",
abstract = "Pre-trained language models based on masked language modeling (MLM) excel in natural language understanding (NLU) tasks. While fine-tuned MLM-based encoders consistently outperform causal language modeling decoders of comparable size, recent decoder-only large language models (LLMs) perform on par with smaller MLM-based encoders. Although their performance improves with scale, LLMs fall short of achieving state-of-the-art results in information extraction (IE) tasks, many of which are formulated as sequence labeling (SL). We hypothesize that LLMs{'} poor SL performance stems from causal masking, which prevents the model from attending to tokens on the right of the current token. Yet, how exactly and to what extent LLMs{'} performance on SL can be improved remains unclear. We explore techniques for improving the SL performance of open LLMs on IE tasks by applying layer-wise removal of the causal mask (CM) during LLM fine-tuning. This approach yields performance gains competitive with state-of-the-art SL models, matching or outperforming the results of CM removal from all blocks. Our findings hold for diverse SL tasks, demonstrating that open LLMs with layer-dependent CM removal outperform strong MLM-based encoders and even instruction-tuned LLMs.",
}
|
Pre-trained language models based on masked language modeling (MLM) excel in natural language understanding (NLU) tasks. While fine-tuned MLM-based encoders consistently outperform causal language modeling decoders of comparable size, recent decoder-only large language models (LLMs) perform on par with smaller MLM-based encoders. Although their performance improves with scale, LLMs fall short of achieving state-of-the-art results in information extraction (IE) tasks, many of which are formulated as sequence labeling (SL). We hypothesize that LLMs{'} poor SL performance stems from causal masking, which prevents the model from attending to tokens on the right of the current token. Yet, how exactly and to what extent LLMs{'} performance on SL can be improved remains unclear. We explore techniques for improving the SL performance of open LLMs on IE tasks by applying layer-wise removal of the causal mask (CM) during LLM fine-tuning. This approach yields performance gains competitive with state-of-the-art SL models, matching or outperforming the results of CM removal from all blocks. Our findings hold for diverse SL tasks, demonstrating that open LLMs with layer-dependent CM removal outperform strong MLM-based encoders and even instruction-tuned LLMs.
|
[
"Duki{\\'c}, David",
"Snajder, Jan"
] |
Looking Right is Sometimes Right: Investigating the Capabilities of Decoder-only {LLM}s for Sequence Labeling
|
findings-acl.843
|
Poster
|
2401.14556v3
|
https://aclanthology.org/2024.findings-acl.844.bib
|
@inproceedings{sakai-etal-2024-mcsqa,
title = "m{CSQA}: Multilingual Commonsense Reasoning Dataset with Unified Creation Strategy by Language Models and Humans",
author = "Sakai, Yusuke and
Kamigaito, Hidetaka and
Watanabe, Taro",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.844",
pages = "14182--14214",
abstract = "It is very challenging to curate a dataset for language-specific knowledge and common sense in order to evaluate natural language understanding capabilities of language models. Due to the limitation in the availability of annotators, most current multilingual datasets are created through translation, which cannot evaluate such language-specific aspects. Therefore, we propose Multilingual CommonsenseQA (mCSQA) based on the construction process of CSQA but leveraging language models for a more efficient construction, e.g., by asking LM to generate questions/answers, refine answers and verify QAs followed by reduced human efforts for verification. Constructed dataset is a benchmark for cross-lingual language-transfer capabilities of multilingual LMs, and experimental results showed high language-transfer capabilities for questions that LMs could easily solve, but lower transfer capabilities for questions requiring deep knowledge or commonsense. This highlights the necessity of language-specific datasets for evaluation and training. Finally, our method demonstrated that multilingual LMs could create QA including language-specific knowledge, significantly reducing the dataset creation cost compared to manual creation. The datasets are available at https://huggingface.co/datasets/yusuke1997/mCSQA.",
}
|
It is very challenging to curate a dataset for language-specific knowledge and common sense in order to evaluate natural language understanding capabilities of language models. Due to the limitation in the availability of annotators, most current multilingual datasets are created through translation, which cannot evaluate such language-specific aspects. Therefore, we propose Multilingual CommonsenseQA (mCSQA) based on the construction process of CSQA but leveraging language models for a more efficient construction, e.g., by asking LM to generate questions/answers, refine answers and verify QAs followed by reduced human efforts for verification. Constructed dataset is a benchmark for cross-lingual language-transfer capabilities of multilingual LMs, and experimental results showed high language-transfer capabilities for questions that LMs could easily solve, but lower transfer capabilities for questions requiring deep knowledge or commonsense. This highlights the necessity of language-specific datasets for evaluation and training. Finally, our method demonstrated that multilingual LMs could create QA including language-specific knowledge, significantly reducing the dataset creation cost compared to manual creation. The datasets are available at https://huggingface.co/datasets/yusuke1997/mCSQA.
|
[
"Sakai, Yusuke",
"Kamigaito, Hidetaka",
"Watanabe, Taro"
] |
m{CSQA}: Multilingual Commonsense Reasoning Dataset with Unified Creation Strategy by Language Models and Humans
|
findings-acl.844
|
Poster
|
2406.04215v1
|
https://aclanthology.org/2024.findings-acl.845.bib
|
@inproceedings{liu-etal-2024-dual,
title = "Dual-Stage Multi-Task Syntax-Oriented Pre-Training for Syntactically Controlled Paraphrase Generation",
author = "Liu, Hongxu and
Wang, Xiaojie and
Sun, Jiashen and
Zeng, Ke and
Guanglu, Wan",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.845",
pages = "14215--14231",
abstract = "Syntactically Controlled Paraphrase Generation (SCPG), which aims at generating sentences having syntactic structures resembling given exemplars, is attracting more research efforts in recent years. We took an empirical survey on previous SCPG datasets and methods and found three tacitly approved while seldom mentioned intrinsic shortcomings/trade-offs in terms of data obtaining, task formulation, and pre-training strategies. As a mitigation to these shortcomings, we proposed a novel Dual-Stage Multi-Task (DSMT) pre-training scheme, involving a series of structure-oriented and syntax-oriented tasks, which, in our opinion, gives sequential text models the ability of com-prehending intrinsically non-sequential structures like Linearized Constituency Trees (LCTs), understanding the underlying syntactics, and even generating them by parsing sentences. We performed further pre-training of the popular T5 model on these novel tasks and fine-tuned the trained model on every possible variant of SCPG task in literature, finding that our models significantly outperformed (up to 10+ BLEU-4) previous state-of-the-art methods. Finally, we carried out ablation studies which demonstrated the effectiveness of our DSMT methods and emphasized on the SCPG performance gains compared to vanilla T5 models, especially on hard samples or under few-shot settings.",
}
|
Syntactically Controlled Paraphrase Generation (SCPG), which aims at generating sentences having syntactic structures resembling given exemplars, is attracting more research efforts in recent years. We took an empirical survey on previous SCPG datasets and methods and found three tacitly approved while seldom mentioned intrinsic shortcomings/trade-offs in terms of data obtaining, task formulation, and pre-training strategies. As a mitigation to these shortcomings, we proposed a novel Dual-Stage Multi-Task (DSMT) pre-training scheme, involving a series of structure-oriented and syntax-oriented tasks, which, in our opinion, gives sequential text models the ability of com-prehending intrinsically non-sequential structures like Linearized Constituency Trees (LCTs), understanding the underlying syntactics, and even generating them by parsing sentences. We performed further pre-training of the popular T5 model on these novel tasks and fine-tuned the trained model on every possible variant of SCPG task in literature, finding that our models significantly outperformed (up to 10+ BLEU-4) previous state-of-the-art methods. Finally, we carried out ablation studies which demonstrated the effectiveness of our DSMT methods and emphasized on the SCPG performance gains compared to vanilla T5 models, especially on hard samples or under few-shot settings.
|
[
"Liu, Hongxu",
"Wang, Xiaojie",
"Sun, Jiashen",
"Zeng, Ke",
"Guanglu, Wan"
] |
Dual-Stage Multi-Task Syntax-Oriented Pre-Training for Syntactically Controlled Paraphrase Generation
|
findings-acl.845
|
Poster
|
2101.10579v1
|
https://aclanthology.org/2024.findings-acl.846.bib
|
@inproceedings{su-etal-2024-demonstration,
title = "Demonstration Augmentation for Zero-shot In-context Learning",
author = "Su, Yi and
Tai, Yunpeng and
Ji, Yixin and
Li, Juntao and
Bowen, Yan and
Zhang, Min",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.846",
pages = "14232--14244",
abstract = "Large Language Models (LLMs) have demonstrated an impressive capability known as In-context Learning (ICL), which enables them to acquire knowledge from textual demonstrations without the need for parameter updates.However, many studies have highlighted that the model{'}s performance is sensitive to the choice of demonstrations, presenting a significant challenge for practical applications where we lack prior knowledge of user queries.Consequently, we need to construct an extensive demonstration pool and incorporate external databases to assist the model, leading to considerable time and financial costs.In light of this, some recent research has shifted focus towards zero-shot ICL, aiming to reduce the model{'}s reliance on external information by leveraging their inherent generative capabilities. Despite the effectiveness of these approaches, the content generated by the model may be unreliable, and the generation process is time-consuming.To address these issues, we propose Demonstration Augmentation for In-context Learning (DAIL), which employs the model{'}s previously predicted historical samples as demonstrations for subsequent ones.DAIL brings no additional inference cost and does not rely on the model{'}s generative capabilities.Our experiments reveal that DAIL can significantly improve the model{'}s performance over direct zero-shot inference and can even outperform few-shot ICL without any external information.",
}
|
Large Language Models (LLMs) have demonstrated an impressive capability known as In-context Learning (ICL), which enables them to acquire knowledge from textual demonstrations without the need for parameter updates.However, many studies have highlighted that the model{'}s performance is sensitive to the choice of demonstrations, presenting a significant challenge for practical applications where we lack prior knowledge of user queries.Consequently, we need to construct an extensive demonstration pool and incorporate external databases to assist the model, leading to considerable time and financial costs.In light of this, some recent research has shifted focus towards zero-shot ICL, aiming to reduce the model{'}s reliance on external information by leveraging their inherent generative capabilities. Despite the effectiveness of these approaches, the content generated by the model may be unreliable, and the generation process is time-consuming.To address these issues, we propose Demonstration Augmentation for In-context Learning (DAIL), which employs the model{'}s previously predicted historical samples as demonstrations for subsequent ones.DAIL brings no additional inference cost and does not rely on the model{'}s generative capabilities.Our experiments reveal that DAIL can significantly improve the model{'}s performance over direct zero-shot inference and can even outperform few-shot ICL without any external information.
|
[
"Su, Yi",
"Tai, Yunpeng",
"Ji, Yixin",
"Li, Juntao",
"Bowen, Yan",
"Zhang, Min"
] |
Demonstration Augmentation for Zero-shot In-context Learning
|
findings-acl.846
|
Poster
|
2305.09731v1
|
https://aclanthology.org/2024.findings-acl.847.bib
|
@inproceedings{tsiamas-etal-2024-pushing,
title = "Pushing the Limits of Zero-shot End-to-End Speech Translation",
author = "Tsiamas, Ioannis and
G{\'a}llego, Gerard and
Fonollosa, Jos{\'e} and
Costa-juss{\`a}, Marta",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.847",
pages = "14245--14267",
abstract = "Data scarcity and the modality gap between the speech and text modalities are two major obstacles of end-to-end Speech Translation (ST) systems, thus hindering their performance. Prior work has attempted to mitigate these challenges by leveraging external MT data and optimizing distance metrics that bring closer the speech-text representations. However, achieving competitive results typically requires some ST data. For this reason, we introduce ZeroSwot, a method for zero-shot ST that bridges the modality gap without any paired ST data. Leveraging a novel CTC compression and Optimal Transport, we train a speech encoder using only ASR data, to align with the representation space of a massively multilingual MT model. The speech encoder seamlessly integrates with the MT model at inference, enabling direct translation from speech to text, across all languages supported by the MT model. Our experiments show that we can effectively close the modality gap without ST data, while our results on MuST-C and CoVoST demonstrate our method{'}s superiority over not only previous zero-shot models, but also supervised ones, achieving state-of-the-art results.",
}
|
Data scarcity and the modality gap between the speech and text modalities are two major obstacles of end-to-end Speech Translation (ST) systems, thus hindering their performance. Prior work has attempted to mitigate these challenges by leveraging external MT data and optimizing distance metrics that bring closer the speech-text representations. However, achieving competitive results typically requires some ST data. For this reason, we introduce ZeroSwot, a method for zero-shot ST that bridges the modality gap without any paired ST data. Leveraging a novel CTC compression and Optimal Transport, we train a speech encoder using only ASR data, to align with the representation space of a massively multilingual MT model. The speech encoder seamlessly integrates with the MT model at inference, enabling direct translation from speech to text, across all languages supported by the MT model. Our experiments show that we can effectively close the modality gap without ST data, while our results on MuST-C and CoVoST demonstrate our method{'}s superiority over not only previous zero-shot models, but also supervised ones, achieving state-of-the-art results.
|
[
"Tsiamas, Ioannis",
"G{\\'a}llego, Gerard",
"Fonollosa, Jos{\\'e}",
"Costa-juss{\\`a}, Marta"
] |
Pushing the Limits of Zero-shot End-to-End Speech Translation
|
findings-acl.847
|
Poster
|
2402.10422v2
|
https://aclanthology.org/2024.findings-acl.848.bib
|
@inproceedings{xu-etal-2024-numcot,
title = "{NUMC}o{T}: Numerals and Units of Measurement in Chain-of-Thought Reasoning using Large Language Models",
author = "Xu, Ancheng and
Tan, Minghuan and
Wang, Lei and
Yang, Min and
Xu, Ruifeng",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.848",
pages = "14268--14290",
abstract = "Numeral systems and units of measurement are two conjoined topics in activities of human beings and have mutual effects with the languages expressing them. Currently, the evaluation of Large Language Models (LLMs) often involves mathematical reasoning, yet little attention is given to how minor changes in numbers or units can drastically alter the complexity of problems and the performance of LLMs. In this paper, we scrutinize existing LLMs on processing of numerals and units of measurement by constructing datasets with perturbations. We first anatomize the reasoning of math word problems to different sub-procedures like numeral conversions from language to numbers and measurement conversions based on units. Then we further annotate math word problems from ancient Chinese arithmetic works which are challenging in numerals and units of measurement. Experiments on perturbed datasets demonstrate that LLMs still encounter difficulties in handling numeral and measurement conversions.",
}
|
Numeral systems and units of measurement are two conjoined topics in activities of human beings and have mutual effects with the languages expressing them. Currently, the evaluation of Large Language Models (LLMs) often involves mathematical reasoning, yet little attention is given to how minor changes in numbers or units can drastically alter the complexity of problems and the performance of LLMs. In this paper, we scrutinize existing LLMs on processing of numerals and units of measurement by constructing datasets with perturbations. We first anatomize the reasoning of math word problems to different sub-procedures like numeral conversions from language to numbers and measurement conversions based on units. Then we further annotate math word problems from ancient Chinese arithmetic works which are challenging in numerals and units of measurement. Experiments on perturbed datasets demonstrate that LLMs still encounter difficulties in handling numeral and measurement conversions.
|
[
"Xu, Ancheng",
"Tan, Minghuan",
"Wang, Lei",
"Yang, Min",
"Xu, Ruifeng"
] |
{NUMC}o{T}: Numerals and Units of Measurement in Chain-of-Thought Reasoning using Large Language Models
|
findings-acl.848
|
Poster
|
2406.02864v1
|
https://aclanthology.org/2024.findings-acl.849.bib
|
@inproceedings{mullick-etal-2024-persona,
title = "On The Persona-based Summarization of Domain-Specific Documents",
author = "Mullick, Ankan and
Bose, Sombit and
Saha, Rounak and
Bhowmick, Ayan and
Goyal, Pawan and
Ganguly, Niloy and
Dey, Prasenjit and
Kokku, Ravi",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.849",
pages = "14291--14307",
abstract = "In an ever-expanding world of domain-specific knowledge, the increasing complexity of consuming, and storing information necessitates the generation of summaries from large information repositories. However, every persona of a domain has different requirements of information and hence their summarization. For example, in the healthcare domain, a persona-based (such as Doctor, Nurse, Patient etc.) approach is imperative to deliver targeted medical information efficiently. Persona-based summarization of domain-specific information by humans is a high cognitive load task and is generally not preferred. The summaries generated by two different humans have high variability and do not scale in cost and subject matter expertise as domains and personas grow. Further, AI-generated summaries using generic Large Language Models (LLMs) may not necessarily offer satisfactory accuracy for different domains unless they have been specifically trained on domain-specific data and can also be very expensive to use in day-to-day operations. Our contribution in this paper is two-fold: 1) We present an approach to efficiently fine-tune a domain-specific small foundation LLM using a healthcare corpus and also show that we can effectively evaluate the summarization quality using AI-based critiquing. 2) We further show that AI-based critiquing has good concordance with Human-based critiquing of the summaries. Hence, such AI-based pipelines to generate domain-specific persona-based summaries can be easily scaled to other domains such as legal, enterprise documents, education etc. in a very efficient and cost-effective manner.",
}
|
In an ever-expanding world of domain-specific knowledge, the increasing complexity of consuming, and storing information necessitates the generation of summaries from large information repositories. However, every persona of a domain has different requirements of information and hence their summarization. For example, in the healthcare domain, a persona-based (such as Doctor, Nurse, Patient etc.) approach is imperative to deliver targeted medical information efficiently. Persona-based summarization of domain-specific information by humans is a high cognitive load task and is generally not preferred. The summaries generated by two different humans have high variability and do not scale in cost and subject matter expertise as domains and personas grow. Further, AI-generated summaries using generic Large Language Models (LLMs) may not necessarily offer satisfactory accuracy for different domains unless they have been specifically trained on domain-specific data and can also be very expensive to use in day-to-day operations. Our contribution in this paper is two-fold: 1) We present an approach to efficiently fine-tune a domain-specific small foundation LLM using a healthcare corpus and also show that we can effectively evaluate the summarization quality using AI-based critiquing. 2) We further show that AI-based critiquing has good concordance with Human-based critiquing of the summaries. Hence, such AI-based pipelines to generate domain-specific persona-based summaries can be easily scaled to other domains such as legal, enterprise documents, education etc. in a very efficient and cost-effective manner.
|
[
"Mullick, Ankan",
"Bose, Sombit",
"Saha, Rounak",
"Bhowmick, Ayan",
"Goyal, Pawan",
"Ganguly, Niloy",
"Dey, Prasenjit",
"Kokku, Ravi"
] |
On The Persona-based Summarization of Domain-Specific Documents
|
findings-acl.849
|
Poster
|
0308033v1
|
https://aclanthology.org/2024.findings-acl.850.bib
|
@inproceedings{kaur-etal-2024-evaluating,
title = "Evaluating Large Language Models for Health-related Queries with Presuppositions",
author = "Kaur, Navreet and
Choudhury, Monojit and
Pruthi, Danish",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.850",
pages = "14308--14331",
abstract = "As corporations rush to integrate large language models (LLMs) it is critical that they provide factually accurate information, that is robust to any presuppositions that a user may express. In this work, we introduce UPHILL, a dataset consisting of health-related queries with varying degrees of presuppositions. Using UPHILL, we evaluate the factual accuracy and consistency of InstructGPT, ChatGPT, GPT-4 and Bing Copilot models. We find that while model responses rarely contradict true health claims (posed as questions), all investigated models fail to challenge false claims. Alarmingly, responses from these models agree with 23-32{\%} of the existing false claims, and 49-55{\%} with novel fabricated claims. As we increase the extent of presupposition in input queries, responses from all models except Bing Copilot agree with the claim considerably more often, regardless of its veracity. Given the moderate factual accuracy, and the inability of models to challenge false assumptions, our work calls for a careful assessment of current LLMs for use in high-stakes scenarios.",
}
|
As corporations rush to integrate large language models (LLMs) it is critical that they provide factually accurate information, that is robust to any presuppositions that a user may express. In this work, we introduce UPHILL, a dataset consisting of health-related queries with varying degrees of presuppositions. Using UPHILL, we evaluate the factual accuracy and consistency of InstructGPT, ChatGPT, GPT-4 and Bing Copilot models. We find that while model responses rarely contradict true health claims (posed as questions), all investigated models fail to challenge false claims. Alarmingly, responses from these models agree with 23-32{\%} of the existing false claims, and 49-55{\%} with novel fabricated claims. As we increase the extent of presupposition in input queries, responses from all models except Bing Copilot agree with the claim considerably more often, regardless of its veracity. Given the moderate factual accuracy, and the inability of models to challenge false assumptions, our work calls for a careful assessment of current LLMs for use in high-stakes scenarios.
|
[
"Kaur, Navreet",
"Choudhury, Monojit",
"Pruthi, Danish"
] |
Evaluating Large Language Models for Health-related Queries with Presuppositions
|
findings-acl.850
|
Poster
|
2312.08800v2
|
https://aclanthology.org/2024.findings-acl.851.bib
|
@inproceedings{bejgu-etal-2024-word,
title = "Word Sense Linking: Disambiguating Outside the Sandbox",
author = "Bejgu, Andrei and
Barba, Edoardo and
Procopio, Luigi and
Fern{\'a}ndez-Castro, Alberte and
Navigli, Roberto",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.851",
pages = "14332--14347",
abstract = "Word Sense Disambiguation (WSD) is the task of associating a word in a given context with its most suitable meaning among a set of possible candidates. While the task has recently witnessed renewed interest, with systems achieving performances above the estimated inter-annotator agreement, at the time of writing it still struggles to find downstream applications. We argue that one of the reasons behind this is the difficulty of applying WSD to plain text. Indeed, in the standard formulation, models work under the assumptions that a) all the spans to disambiguate have already been identified, and b) all the possible candidate senses of each span are provided, both of which are requirements that are far from trivial. In this work, we present a new task called Word Sense Linking (WSL) where, given an input text and a reference sense inventory, systems have to both identify which spans to disambiguate and then link them to their most suitable meaning.We put forward a transformer-based architecture for the task and thoroughly evaluate both its performance and those of state-of-the-art WSD systems scaled to WSL, iteratively relaxing the assumptions of WSD. We hope that our work will foster easier integration of lexical semantics into downstream applications.",
}
|
Word Sense Disambiguation (WSD) is the task of associating a word in a given context with its most suitable meaning among a set of possible candidates. While the task has recently witnessed renewed interest, with systems achieving performances above the estimated inter-annotator agreement, at the time of writing it still struggles to find downstream applications. We argue that one of the reasons behind this is the difficulty of applying WSD to plain text. Indeed, in the standard formulation, models work under the assumptions that a) all the spans to disambiguate have already been identified, and b) all the possible candidate senses of each span are provided, both of which are requirements that are far from trivial. In this work, we present a new task called Word Sense Linking (WSL) where, given an input text and a reference sense inventory, systems have to both identify which spans to disambiguate and then link them to their most suitable meaning.We put forward a transformer-based architecture for the task and thoroughly evaluate both its performance and those of state-of-the-art WSD systems scaled to WSL, iteratively relaxing the assumptions of WSD. We hope that our work will foster easier integration of lexical semantics into downstream applications.
|
[
"Bejgu, Andrei",
"Barba, Edoardo",
"Procopio, Luigi",
"Fern{\\'a}ndez-Castro, Alberte",
"Navigli, Roberto"
] |
Word Sense Linking: Disambiguating Outside the Sandbox
|
findings-acl.851
|
Poster
|
1608.04303v1
|
https://aclanthology.org/2024.findings-acl.852.bib
|
@inproceedings{dankers-titov-2024-generalisation,
title = "Generalisation First, Memorisation Second? Memorisation Localisation for Natural Language Classification Tasks",
author = "Dankers, Verna and
Titov, Ivan",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.852",
pages = "14348--14366",
abstract = "Memorisation is a natural part of learning from real-world data: neural models pick up on atypical input-output combinations and store those training examples in their parameter space. That this happens is well-known, but how and where are questions that remain largely unanswered. Given a multi-layered neural model, where does memorisation occur in the millions of parameters?Related work reports conflicting findings: a dominant hypothesis based on image classification is that lower layers learn generalisable features and that deeper layers specialise and memorise. Work from NLP suggests this does not apply to language models, but has been mainly focused on memorisation of facts.We expand the scope of the localisation question to 12 natural language classification tasks and apply 4 memorisation localisation techniques.Our results indicate that memorisation is a gradual process rather than a localised one, establish that memorisation is task-dependent, and give nuance to the generalisation first, memorisation second hypothesis.",
}
|
Memorisation is a natural part of learning from real-world data: neural models pick up on atypical input-output combinations and store those training examples in their parameter space. That this happens is well-known, but how and where are questions that remain largely unanswered. Given a multi-layered neural model, where does memorisation occur in the millions of parameters?Related work reports conflicting findings: a dominant hypothesis based on image classification is that lower layers learn generalisable features and that deeper layers specialise and memorise. Work from NLP suggests this does not apply to language models, but has been mainly focused on memorisation of facts.We expand the scope of the localisation question to 12 natural language classification tasks and apply 4 memorisation localisation techniques.Our results indicate that memorisation is a gradual process rather than a localised one, establish that memorisation is task-dependent, and give nuance to the generalisation first, memorisation second hypothesis.
|
[
"Dankers, Verna",
"Titov, Ivan"
] |
Generalisation First, Memorisation Second? Memorisation Localisation for Natural Language Classification Tasks
|
findings-acl.852
|
Poster
|
2408.04965v1
|
https://aclanthology.org/2024.findings-acl.853.bib
|
@inproceedings{liu-etal-2024-towards-multi,
title = "Towards Multi-Relational Multi-Hop Reasoning over Dense Temporal Knowledge Graphs",
author = "Liu, Jian and
Liu, Zihe and
Lyu, Xueqiang and
Jin, Peng and
Xu, Jinan",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.853",
pages = "14367--14378",
abstract = "Temporal knowledge graph reasoning has emerged as a crucial task for answering time-dependent questions within a knowledge graph (KG).Despite tremendous progress, the present research is impeded by the sparsity of a temporal KG and an over-reliance on simple single-relational reasoning patterns. To overcome these challenges, we introduce MulQuestions, a new temporal KG reasoning benchmark featuring over 200k entities and 960k questions designed to facilitate complex, multi-relational and multi-hop reasoning. Additionally, we propose a new model adept at conducting pattern-aware and time-sensitive reasoning across temporal KGs. The model{'}s efficacy is confirmed through rigorous evaluations, showcasing its effectiveness in sparse data conditions and adeptness at handling questions with long reasoning chains. We have made our benchmark and model publicly accessible at [https://anonymous].",
}
|
Temporal knowledge graph reasoning has emerged as a crucial task for answering time-dependent questions within a knowledge graph (KG).Despite tremendous progress, the present research is impeded by the sparsity of a temporal KG and an over-reliance on simple single-relational reasoning patterns. To overcome these challenges, we introduce MulQuestions, a new temporal KG reasoning benchmark featuring over 200k entities and 960k questions designed to facilitate complex, multi-relational and multi-hop reasoning. Additionally, we propose a new model adept at conducting pattern-aware and time-sensitive reasoning across temporal KGs. The model{'}s efficacy is confirmed through rigorous evaluations, showcasing its effectiveness in sparse data conditions and adeptness at handling questions with long reasoning chains. We have made our benchmark and model publicly accessible at [https://anonymous].
|
[
"Liu, Jian",
"Liu, Zihe",
"Lyu, Xueqiang",
"Jin, Peng",
"Xu, Jinan"
] |
Towards Multi-Relational Multi-Hop Reasoning over Dense Temporal Knowledge Graphs
|
findings-acl.853
|
Poster
|
2103.10379v1
|
https://aclanthology.org/2024.findings-acl.854.bib
|
@inproceedings{su-etal-2024-unsupervised,
title = "Unsupervised Real-Time Hallucination Detection based on the Internal States of Large Language Models",
author = "Su, Weihang and
Wang, Changyue and
Ai, Qingyao and
Hu, Yiran and
Wu, Zhijing and
Zhou, Yujia and
Liu, Yiqun",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.854",
pages = "14379--14391",
abstract = "Hallucinations in large language models (LLMs) refer to the phenomenon of LLMs producing responses that are coherent yet factually inaccurate. This issue undermines the effectiveness of LLMs in practical applications, necessitating research into detecting and mitigating hallucinations of LLMs. Previous studies have mainly concentrated on post-processing techniques for hallucination detection, which tend to be computationally intensive and limited in effectiveness due to their separation from the LLM{'}s inference process. To overcome these limitations, we introduce MIND, an unsupervised training framework that leverages the internal states of LLMs for real-time hallucination detection without requiring manual annotations. Additionally, we present HELM, a new benchmark for evaluating hallucination detection across multiple LLMs, featuring diverse LLM outputs and the internal states of LLMs during their inference process. Our experiments demonstrate that MIND outperforms existing state-of-the-art methods in hallucination detection.",
}
|
Hallucinations in large language models (LLMs) refer to the phenomenon of LLMs producing responses that are coherent yet factually inaccurate. This issue undermines the effectiveness of LLMs in practical applications, necessitating research into detecting and mitigating hallucinations of LLMs. Previous studies have mainly concentrated on post-processing techniques for hallucination detection, which tend to be computationally intensive and limited in effectiveness due to their separation from the LLM{'}s inference process. To overcome these limitations, we introduce MIND, an unsupervised training framework that leverages the internal states of LLMs for real-time hallucination detection without requiring manual annotations. Additionally, we present HELM, a new benchmark for evaluating hallucination detection across multiple LLMs, featuring diverse LLM outputs and the internal states of LLMs during their inference process. Our experiments demonstrate that MIND outperforms existing state-of-the-art methods in hallucination detection.
|
[
"Su, Weihang",
"Wang, Changyue",
"Ai, Qingyao",
"Hu, Yiran",
"Wu, Zhijing",
"Zhou, Yujia",
"Liu, Yiqun"
] |
Unsupervised Real-Time Hallucination Detection based on the Internal States of Large Language Models
|
findings-acl.854
|
Poster
|
2403.06448v2
|
https://aclanthology.org/2024.findings-acl.855.bib
|
@inproceedings{hou-etal-2024-progressive,
title = "Progressive Tuning: Towards Generic Sentiment Abilities for Large Language Models",
author = "Hou, Guiyang and
Shen, Yongliang and
Lu, Weiming",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.855",
pages = "14392--14402",
abstract = "Understanding sentiment is arguably an advanced and important capability of AI agents in the physical world. In previous works, many efforts have been devoted to individual sentiment subtasks, without considering interrelated sentiment knowledge among these subtasks. Although some recent works model multiple sentiment subtasks in a unified manner, they merely simply combine these subtasks without deeply exploring the hierarchical relationships among subtasks. In this paper, we introduce GSA-7B, an open-source large language model specific to the sentiment domain. Specifically, we deeply explore the hierarchical relationships between sentiment subtasks, proposing progressive sentiment reasoning benchmark and progressive task instructions. Subsequently, we use Llama2-7B as the backbone model and propose parameter-efficient progressive tuning paradigm which is implemented by constructing chain of LoRA, resulting in the creation of GSA-7B. Experimental results show that GSA-7B as a unified model performs well across all datasets in the progressive sentiment reasoning benchmark. Additionally, under the few-shot setting, GSA-7B also exhibits good generalization ability for sentiment subtasks and datasets that were not encountered during its training phase.",
}
|
Understanding sentiment is arguably an advanced and important capability of AI agents in the physical world. In previous works, many efforts have been devoted to individual sentiment subtasks, without considering interrelated sentiment knowledge among these subtasks. Although some recent works model multiple sentiment subtasks in a unified manner, they merely simply combine these subtasks without deeply exploring the hierarchical relationships among subtasks. In this paper, we introduce GSA-7B, an open-source large language model specific to the sentiment domain. Specifically, we deeply explore the hierarchical relationships between sentiment subtasks, proposing progressive sentiment reasoning benchmark and progressive task instructions. Subsequently, we use Llama2-7B as the backbone model and propose parameter-efficient progressive tuning paradigm which is implemented by constructing chain of LoRA, resulting in the creation of GSA-7B. Experimental results show that GSA-7B as a unified model performs well across all datasets in the progressive sentiment reasoning benchmark. Additionally, under the few-shot setting, GSA-7B also exhibits good generalization ability for sentiment subtasks and datasets that were not encountered during its training phase.
|
[
"Hou, Guiyang",
"Shen, Yongliang",
"Lu, Weiming"
] |
Progressive Tuning: Towards Generic Sentiment Abilities for Large Language Models
|
findings-acl.855
|
Poster
|
2407.02340v1
|
https://aclanthology.org/2024.findings-acl.856.bib
|
@inproceedings{hoang-etal-2024-fooling,
title = "Fooling the Textual Fooler via Randomizing Latent Representations",
author = "Hoang, Duy and
Hung-Quang, Nguyen and
Manchanda, Saurav and
Peng, Minlong and
Wong, Kok-Seng and
Doan, Khoa",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.856",
pages = "14403--14421",
abstract = "Despite outstanding performance in a variety of Natural Language Processing (NLP) tasks, recent studies have revealed that NLP models are vulnerable to adversarial attacks that slightly perturb the input to cause the models to misbehave. Several attacks can even compromise the model without requiring access to the model architecture or model parameters (i.e., a blackbox setting), and thus are detrimental to existing NLP applications. To perform these attacks, the adversary queries the victim model many times to determine the most important parts in an input text and transform. In this work, we propose a lightweight and attack-agnostic defense whose main goal is to perplex the process of generating an adversarial example in these query-based black-box attacks; that is to fool the textual fooler. This defense, named AdvFooler, works by randomizing the latent representation of the input at inference time. Different from existing defenses, AdvFooler does not necessitate additional computational overhead during training nor does it rely on assumptions about the potential adversarial perturbation set while having a negligible impact on the model{'}s accuracy. Our theoretical and empirical analyses highlight the significance of robustness resulting from confusing the adversary via randomizing the latent space, as well as the impact of randomization on clean accuracy. Finally, we empirically demonstrate near state-of-the-art robustness of AdvFooler against representative adversarial attacks on two benchmark datasets.",
}
|
Despite outstanding performance in a variety of Natural Language Processing (NLP) tasks, recent studies have revealed that NLP models are vulnerable to adversarial attacks that slightly perturb the input to cause the models to misbehave. Several attacks can even compromise the model without requiring access to the model architecture or model parameters (i.e., a blackbox setting), and thus are detrimental to existing NLP applications. To perform these attacks, the adversary queries the victim model many times to determine the most important parts in an input text and transform. In this work, we propose a lightweight and attack-agnostic defense whose main goal is to perplex the process of generating an adversarial example in these query-based black-box attacks; that is to fool the textual fooler. This defense, named AdvFooler, works by randomizing the latent representation of the input at inference time. Different from existing defenses, AdvFooler does not necessitate additional computational overhead during training nor does it rely on assumptions about the potential adversarial perturbation set while having a negligible impact on the model{'}s accuracy. Our theoretical and empirical analyses highlight the significance of robustness resulting from confusing the adversary via randomizing the latent space, as well as the impact of randomization on clean accuracy. Finally, we empirically demonstrate near state-of-the-art robustness of AdvFooler against representative adversarial attacks on two benchmark datasets.
|
[
"Hoang, Duy",
"Hung-Quang, Nguyen",
"Manch",
"a, Saurav",
"Peng, Minlong",
"Wong, Kok-Seng",
"Doan, Khoa"
] |
Fooling the Textual Fooler via Randomizing Latent Representations
|
findings-acl.856
|
Poster
|
2310.01452v2
|
https://aclanthology.org/2024.findings-acl.857.bib
|
@inproceedings{kumar-etal-2024-part,
title = "Part-of-speech Tagging for Extremely Low-resource {I}ndian Languages",
author = "Kumar, Sanjeev and
Jyothi, Preethi and
Bhattacharyya, Pushpak",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.857",
pages = "14422--14431",
abstract = "Modern natural language processing (NLP) systems thrive when given access to large datasets. However, a large fraction of the world{'}s languages are not privy to such benefits due to sparse documentation and inadequate digital representation. This is especially true for Indian regional languages. As a first step towards expanding the reach of NLP technologies to extremely low-resource Indian languages, we present a new parallel part-of-speech (POS) evaluation dataset for Angika, Magahi, Bhojpuri and Hindi. Angika, Magahi, Bhojpuri, along with the more well-known Hindi, are all languages spoken in the Indian states of Bihar, Jharkhand and West Bengal. Ours is notably the first NLP resource, even for a shallow NLP task like POS-tagging, for Angika. We establish POS-tagging baselines using state-of-the-art multilingual pretrained language models (PLMs) finetuned on Hindi data, and show zero-shot evaluations on the other three languages. While all four languages use the same Devanagari script, pretrained tokenizers underperform in zero-shot on the three languages. We propose a simple look-back fix to address the tokenization challenge yielding F1-score improvements of up to 8{\%} on Angika and show how it comes very close to an oracle setting when the underlying Hindi word is known (and can be accurately tokenized).",
}
|
Modern natural language processing (NLP) systems thrive when given access to large datasets. However, a large fraction of the world{'}s languages are not privy to such benefits due to sparse documentation and inadequate digital representation. This is especially true for Indian regional languages. As a first step towards expanding the reach of NLP technologies to extremely low-resource Indian languages, we present a new parallel part-of-speech (POS) evaluation dataset for Angika, Magahi, Bhojpuri and Hindi. Angika, Magahi, Bhojpuri, along with the more well-known Hindi, are all languages spoken in the Indian states of Bihar, Jharkhand and West Bengal. Ours is notably the first NLP resource, even for a shallow NLP task like POS-tagging, for Angika. We establish POS-tagging baselines using state-of-the-art multilingual pretrained language models (PLMs) finetuned on Hindi data, and show zero-shot evaluations on the other three languages. While all four languages use the same Devanagari script, pretrained tokenizers underperform in zero-shot on the three languages. We propose a simple look-back fix to address the tokenization challenge yielding F1-score improvements of up to 8{\%} on Angika and show how it comes very close to an oracle setting when the underlying Hindi word is known (and can be accurately tokenized).
|
[
"Kumar, Sanjeev",
"Jyothi, Preethi",
"Bhattacharyya, Pushpak"
] |
Part-of-speech Tagging for Extremely Low-resource {I}ndian Languages
|
findings-acl.857
|
Poster
|
1702.00167v2
|
https://aclanthology.org/2024.findings-acl.858.bib
|
@inproceedings{lan-etal-2024-focus,
title = "{FOCUS}: Forging Originality through Contrastive Use in Self-Plagiarism for Language Models",
author = "Lan, Kaixin and
Fang, Tao and
Wong, Derek and
Xu, Yabo and
Chao, Lidia and
Zhao, Cecilia",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.858",
pages = "14432--14447",
abstract = "Pre-trained Language Models (PLMs) have shown impressive results in various Natural Language Generation (NLG) tasks, such as powering chatbots and generating stories. However, an ethical concern arises due to their potential to produce verbatim copies of paragraphs from their training data. This is problematic as PLMs are trained on corpora constructed by human authors. As such, there is a pressing need for research to promote the generation of original content by these models. In this study, we introduce a unique {``}self-plagiarism{''} contrastive decoding strategy, aimed at boosting the originality of text produced by PLMs. Our method entails modifying prompts in LLMs to develop an amateur model and a professional model. Specifically, the amateur model is urged to plagiarize using three plagiarism templates we have designed, while the professional model maintains its standard language model status. This strategy employs prompts to stimulate the model{'}s capacity to identify non-original candidate token combinations and subsequently impose penalties. The application of this strategy is integrated prior to the model{'}s final layer, ensuring smooth integration with most existing PLMs (T5, GPT, LLaMA) without necessitating further adjustments. Implementing our strategy, we noted a significant decline in non-original sequences comprised of more than three words in the academic AASC dataset and the story-based ROCStories dataset. Source code and scripts will be released after the paper{'}s acceptance and publication.",
}
|
Pre-trained Language Models (PLMs) have shown impressive results in various Natural Language Generation (NLG) tasks, such as powering chatbots and generating stories. However, an ethical concern arises due to their potential to produce verbatim copies of paragraphs from their training data. This is problematic as PLMs are trained on corpora constructed by human authors. As such, there is a pressing need for research to promote the generation of original content by these models. In this study, we introduce a unique {``}self-plagiarism{''} contrastive decoding strategy, aimed at boosting the originality of text produced by PLMs. Our method entails modifying prompts in LLMs to develop an amateur model and a professional model. Specifically, the amateur model is urged to plagiarize using three plagiarism templates we have designed, while the professional model maintains its standard language model status. This strategy employs prompts to stimulate the model{'}s capacity to identify non-original candidate token combinations and subsequently impose penalties. The application of this strategy is integrated prior to the model{'}s final layer, ensuring smooth integration with most existing PLMs (T5, GPT, LLaMA) without necessitating further adjustments. Implementing our strategy, we noted a significant decline in non-original sequences comprised of more than three words in the academic AASC dataset and the story-based ROCStories dataset. Source code and scripts will be released after the paper{'}s acceptance and publication.
|
[
"Lan, Kaixin",
"Fang, Tao",
"Wong, Derek",
"Xu, Yabo",
"Chao, Lidia",
"Zhao, Cecilia"
] |
{FOCUS}: Forging Originality through Contrastive Use in Self-Plagiarism for Language Models
|
findings-acl.858
|
Poster
|
2109.07983v3
|
https://aclanthology.org/2024.findings-acl.859.bib
|
@inproceedings{zhang-etal-2024-amanda,
title = "Amanda: Adaptively Modality-Balanced Domain Adaptation for Multimodal Emotion Recognition",
author = "Zhang, Xinxin and
Sun, Jun and
Hong, Simin and
Li, Taihao",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.859",
pages = "14448--14458",
abstract = "This paper investigates unsupervised multimodal domain adaptation for multimodal emotion recognition, which is a solution for data scarcity yet remains under studied. Due to the varying distribution discrepancies of different modalities between source and target domains, the primary challenge lies in how to balance the domain alignment across modalities to guarantee they are all well aligned. To achieve this, we first develop our model based on the information bottleneck theory to learn optimal representation for each modality independently. Then, we align the domains via matching the label distributions and the representations. In order to balance the representation alignment, we propose to minimize a surrogate of the alignment losses, which is equivalent to adaptively adjusting the weights of the modalities throughout training, thus achieving balanced domain alignment across modalities. Overall, the proposed approach features \textbf{A}daptively \textbf{m}odality-bal\textbf{an}ced \textbf{d}omain \textbf{a}daptation, dubbed \textbf{Amanda}, for multimodal emotion recognition. Extensive empirical results on commonly used benchmark datasets demonstrate that Amanda significantly outperforms competing approaches. The code is available at \url{https://github.com/sunjunaimer/Amanda}.",
}
|
This paper investigates unsupervised multimodal domain adaptation for multimodal emotion recognition, which is a solution for data scarcity yet remains under studied. Due to the varying distribution discrepancies of different modalities between source and target domains, the primary challenge lies in how to balance the domain alignment across modalities to guarantee they are all well aligned. To achieve this, we first develop our model based on the information bottleneck theory to learn optimal representation for each modality independently. Then, we align the domains via matching the label distributions and the representations. In order to balance the representation alignment, we propose to minimize a surrogate of the alignment losses, which is equivalent to adaptively adjusting the weights of the modalities throughout training, thus achieving balanced domain alignment across modalities. Overall, the proposed approach features \textbf{A}daptively \textbf{m}odality-bal\textbf{an}ced \textbf{d}omain \textbf{a}daptation, dubbed \textbf{Amanda}, for multimodal emotion recognition. Extensive empirical results on commonly used benchmark datasets demonstrate that Amanda significantly outperforms competing approaches. The code is available at \url{https://github.com/sunjunaimer/Amanda}.
|
[
"Zhang, Xinxin",
"Sun, Jun",
"Hong, Simin",
"Li, Taihao"
] |
Amanda: Adaptively Modality-Balanced Domain Adaptation for Multimodal Emotion Recognition
|
findings-acl.859
|
Poster
|
1909.05645v2
|
https://aclanthology.org/2024.findings-acl.860.bib
|
@inproceedings{vladika-etal-2024-medreqal,
title = "{M}ed{REQAL}: Examining Medical Knowledge Recall of Large Language Models via Question Answering",
author = "Vladika, Juraj and
Schneider, Phillip and
Matthes, Florian",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.860",
pages = "14459--14469",
abstract = "In recent years, Large Language Models (LLMs) have demonstrated an impressive ability to encode knowledge during pre-training on large text corpora. They can leverage this knowledge for downstream tasks like question answering (QA), even in complex areas involving health topics. Considering their high potential for facilitating clinical work in the future, understanding the quality of encoded medical knowledge and its recall in LLMs is an important step forward. In this study, we examine the capability of LLMs to exhibit medical knowledge recall by constructing a novel dataset derived from systematic reviews {--} studies synthesizing evidence-based answers for specific medical questions. Through experiments on the new MedREQAL dataset, comprising question-answer pairs extracted from rigorous systematic reviews, we assess six LLMs, such as GPT and Mixtral, analyzing their classification and generation performance. Our experimental insights into LLM performance on the novel biomedical QA dataset reveal the still challenging nature of this task.",
}
|
In recent years, Large Language Models (LLMs) have demonstrated an impressive ability to encode knowledge during pre-training on large text corpora. They can leverage this knowledge for downstream tasks like question answering (QA), even in complex areas involving health topics. Considering their high potential for facilitating clinical work in the future, understanding the quality of encoded medical knowledge and its recall in LLMs is an important step forward. In this study, we examine the capability of LLMs to exhibit medical knowledge recall by constructing a novel dataset derived from systematic reviews {--} studies synthesizing evidence-based answers for specific medical questions. Through experiments on the new MedREQAL dataset, comprising question-answer pairs extracted from rigorous systematic reviews, we assess six LLMs, such as GPT and Mixtral, analyzing their classification and generation performance. Our experimental insights into LLM performance on the novel biomedical QA dataset reveal the still challenging nature of this task.
|
[
"Vladika, Juraj",
"Schneider, Phillip",
"Matthes, Florian"
] |
{M}ed{REQAL}: Examining Medical Knowledge Recall of Large Language Models via Question Answering
|
findings-acl.860
|
Poster
|
2406.05845v1
|
https://aclanthology.org/2024.findings-acl.861.bib
|
@inproceedings{munir-etal-2024-deepfake,
title = "Deepfake Defense: Constructing and Evaluating a Specialized {U}rdu Deepfake Audio Dataset",
author = "Munir, Sheza and
Sajjad, Wassay and
Raza, Mukeet and
Abbas, Emaan and
Azeemi, Abdul Hameed and
Qazi, Ihsan Ayyub and
Raza, Agha Ali",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.861",
pages = "14470--14480",
abstract = "Deepfakes, particularly in the auditory domain, have become a significant threat, necessitating the development of robust countermeasures. This paper addresses the escalating challenges posed by deepfake attacks on Automatic Speaker Verification (ASV) systems. We present a novel Urdu deepfake audio dataset for deepfake detection, focusing on two spoofing attacks {--} Tacotron and VITS TTS. The dataset construction involves careful consideration of phonemic cover and balance and comparison with existing corpora like PRUS and PronouncUR. Evaluation with AASIST-L model shows EERs of 0.495 and 0.524 for VITS TTS and Tacotron-generated audios, respectively, with variability across speakers. Further, this research implements a detailed human evaluation, incorporating a user study to gauge whether people are able to discern deepfake audios from real (bonafide) audios. The ROC curve analysis shows an area under the curve (AUC) of 0.63, indicating that individuals demonstrate a limited ability to detect deepfakes (approximately 1 in 3 fake audio samples are regarded as real). Our work contributes a valuable resource for training deepfake detection models in low-resource languages like Urdu, addressing the critical gap in existing datasets. The dataset is publicly available at: https://github.com/CSALT-LUMS/urdu-deepfake-dataset.",
}
|
Deepfakes, particularly in the auditory domain, have become a significant threat, necessitating the development of robust countermeasures. This paper addresses the escalating challenges posed by deepfake attacks on Automatic Speaker Verification (ASV) systems. We present a novel Urdu deepfake audio dataset for deepfake detection, focusing on two spoofing attacks {--} Tacotron and VITS TTS. The dataset construction involves careful consideration of phonemic cover and balance and comparison with existing corpora like PRUS and PronouncUR. Evaluation with AASIST-L model shows EERs of 0.495 and 0.524 for VITS TTS and Tacotron-generated audios, respectively, with variability across speakers. Further, this research implements a detailed human evaluation, incorporating a user study to gauge whether people are able to discern deepfake audios from real (bonafide) audios. The ROC curve analysis shows an area under the curve (AUC) of 0.63, indicating that individuals demonstrate a limited ability to detect deepfakes (approximately 1 in 3 fake audio samples are regarded as real). Our work contributes a valuable resource for training deepfake detection models in low-resource languages like Urdu, addressing the critical gap in existing datasets. The dataset is publicly available at: https://github.com/CSALT-LUMS/urdu-deepfake-dataset.
|
[
"Munir, Sheza",
"Sajjad, Wassay",
"Raza, Mukeet",
"Abbas, Emaan",
"Azeemi, Abdul Hameed",
"Qazi, Ihsan Ayyub",
"Raza, Agha Ali"
] |
Deepfake Defense: Constructing and Evaluating a Specialized {U}rdu Deepfake Audio Dataset
|
findings-acl.861
|
Poster
|
2108.05080v4
|
https://aclanthology.org/2024.findings-acl.862.bib
|
@inproceedings{zhang-perez-beltrachini-2024-leveraging,
title = "Leveraging Entailment Judgements in Cross-Lingual Summarisation",
author = "Zhang, Huajian and
Perez-Beltrachini, Laura",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.862",
pages = "14481--14497",
abstract = "Synthetically created Cross-Lingual Summarisation (CLS) datasets are prone to include document-summary pairs where the reference summary is unfaithful to the corresponding document as it contains content not supported by the document (i.e., hallucinated content). This low data quality misleads model learning and obscures evaluation results. Automatic ways to assess hallucinations and improve training have been proposed for monolingual summarisation, predominantly in English. For CLS, we propose to use off-the-shelf cross-lingual Natural Language Inference (X-NLI) to evaluate faithfulness of reference and model generated summaries. Then, we study training approaches that are aware of faithfulness issues in the training data and propose an approach that uses unlikelihood loss to teach a model about unfaithful summary sequences. Our results show that it is possible to train CLS models that yield more faithful and at the same time informative summaries.",
}
|
Synthetically created Cross-Lingual Summarisation (CLS) datasets are prone to include document-summary pairs where the reference summary is unfaithful to the corresponding document as it contains content not supported by the document (i.e., hallucinated content). This low data quality misleads model learning and obscures evaluation results. Automatic ways to assess hallucinations and improve training have been proposed for monolingual summarisation, predominantly in English. For CLS, we propose to use off-the-shelf cross-lingual Natural Language Inference (X-NLI) to evaluate faithfulness of reference and model generated summaries. Then, we study training approaches that are aware of faithfulness issues in the training data and propose an approach that uses unlikelihood loss to teach a model about unfaithful summary sequences. Our results show that it is possible to train CLS models that yield more faithful and at the same time informative summaries.
|
[
"Zhang, Huajian",
"Perez-Beltrachini, Laura"
] |
Leveraging Entailment Judgements in Cross-Lingual Summarisation
|
findings-acl.862
|
Poster
|
2408.00675v1
|
https://aclanthology.org/2024.findings-acl.863.bib
|
@inproceedings{zhang-etal-2024-recognizing,
title = "Recognizing Everything from All Modalities at Once: Grounded Multimodal Universal Information Extraction",
author = "Zhang, Meishan and
Fei, Hao and
Wang, Bin and
Wu, Shengqiong and
Cao, Yixin and
Li, Fei and
Zhang, Min",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.863",
pages = "14498--14511",
abstract = "In the field of information extraction (IE), tasks across a wide range of modalities and their combinations have been traditionally studied in isolation, leaving a gap in deeply recognizing and analyzing cross-modal information. To address this, this work for the first time introduces the concept of grounded Multimodal Universal Information Extraction (MUIE), providing a unified task framework to analyze any IE tasks over various modalities, along with their fine-grained groundings. To tackle MUIE, we tailor a multimodal large language model (MLLM), Reamo, capable of extracting and grounding information from all modalities, i.e., recognizing everything from all modalities at once. Reamo is updated via varied tuning strategies, equipping it with powerful capabilities for information recognition and fine-grained multimodal grounding. To address the absence of a suitable benchmark for grounded MUIE, we curate a high-quality, diverse, and challenging test set, which encompasses IE tasks across 9 common modality combinations with the corresponding multimodal groundings. The extensive comparison of Reamo with existing MLLMs integrated into pipeline approaches demonstrates its advantages across all evaluation dimensions, establishing a strong benchmark for the follow-up research. Our resources are publicly released at https://haofei.vip/MUIE.",
}
|
In the field of information extraction (IE), tasks across a wide range of modalities and their combinations have been traditionally studied in isolation, leaving a gap in deeply recognizing and analyzing cross-modal information. To address this, this work for the first time introduces the concept of grounded Multimodal Universal Information Extraction (MUIE), providing a unified task framework to analyze any IE tasks over various modalities, along with their fine-grained groundings. To tackle MUIE, we tailor a multimodal large language model (MLLM), Reamo, capable of extracting and grounding information from all modalities, i.e., recognizing everything from all modalities at once. Reamo is updated via varied tuning strategies, equipping it with powerful capabilities for information recognition and fine-grained multimodal grounding. To address the absence of a suitable benchmark for grounded MUIE, we curate a high-quality, diverse, and challenging test set, which encompasses IE tasks across 9 common modality combinations with the corresponding multimodal groundings. The extensive comparison of Reamo with existing MLLMs integrated into pipeline approaches demonstrates its advantages across all evaluation dimensions, establishing a strong benchmark for the follow-up research. Our resources are publicly released at https://haofei.vip/MUIE.
|
[
"Zhang, Meishan",
"Fei, Hao",
"Wang, Bin",
"Wu, Shengqiong",
"Cao, Yixin",
"Li, Fei",
"Zhang, Min"
] |
Recognizing Everything from All Modalities at Once: Grounded Multimodal Universal Information Extraction
|
findings-acl.863
|
Poster
|
2406.03701v2
|
https://aclanthology.org/2024.findings-acl.864.bib
|
@inproceedings{li-etal-2024-enhanced,
title = "Enhanced Visual Instruction Tuning with Synthesized Image-Dialogue Data",
author = "Li, Yanda and
Zhang, Chi and
Yu, Gang and
Yang, Wanqi and
Wang, Zhibin and
Fu, Bin and
Lin, Guosheng and
Shen, Chunhua and
Chen, Ling and
Wei, Yunchao",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.864",
pages = "14512--14531",
abstract = "The remarkable multimodal capabilities demonstrated by OpenAI{'}s GPT-4 have sparked significant interest in the development of multimodal Large Language Models (LLMs). A primary research objective of such models is to align visual and textual modalities effectively while comprehending human instructions.Current methodologies often rely on annotations derived from benchmark datasets to construct image-dialogue datasets for training purposes, akin to instruction tuning in LLMs. However, these datasets often exhibit domain bias, potentially constraining the generative capabilities of the models. In an effort to mitigate these limitations, we propose a novel data collection methodology that synchronously synthesizes images and dialogues for visual instruction tuning. This approach harnesses the power of generative models, marrying the abilities of ChatGPT and text-to-image generative models to yield a diverse and controllable dataset with varied image content. This not only provides greater flexibility compared to existing methodologies but also significantly enhances several model capabilities. Our research includes comprehensive experiments conducted on various datasets using the open-source LLAVA model as a testbed for our proposed pipeline. Our results underscore marked enhancements across more than ten commonly assessed capabilities.",
}
|
The remarkable multimodal capabilities demonstrated by OpenAI{'}s GPT-4 have sparked significant interest in the development of multimodal Large Language Models (LLMs). A primary research objective of such models is to align visual and textual modalities effectively while comprehending human instructions.Current methodologies often rely on annotations derived from benchmark datasets to construct image-dialogue datasets for training purposes, akin to instruction tuning in LLMs. However, these datasets often exhibit domain bias, potentially constraining the generative capabilities of the models. In an effort to mitigate these limitations, we propose a novel data collection methodology that synchronously synthesizes images and dialogues for visual instruction tuning. This approach harnesses the power of generative models, marrying the abilities of ChatGPT and text-to-image generative models to yield a diverse and controllable dataset with varied image content. This not only provides greater flexibility compared to existing methodologies but also significantly enhances several model capabilities. Our research includes comprehensive experiments conducted on various datasets using the open-source LLAVA model as a testbed for our proposed pipeline. Our results underscore marked enhancements across more than ten commonly assessed capabilities.
|
[
"Li, Y",
"a",
"Zhang, Chi",
"Yu, Gang",
"Yang, Wanqi",
"Wang, Zhibin",
"Fu, Bin",
"Lin, Guosheng",
"Shen, Chunhua",
"Chen, Ling",
"Wei, Yunchao"
] |
Enhanced Visual Instruction Tuning with Synthesized Image-Dialogue Data
|
findings-acl.864
|
Poster
|
2402.11690v1
|
https://aclanthology.org/2024.findings-acl.865.bib
|
@inproceedings{haga-etal-2024-modeling,
title = "Modeling Overregularization in Children with Small Language Models",
author = "Haga, Akari and
Sugawara, Saku and
Fukatsu, Akiyo and
Oba, Miyu and
Ouchi, Hiroki and
Watanabe, Taro and
Oseki, Yohei",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.865",
pages = "14532--14550",
abstract = "The imitation of the children{'}s language acquisition process has been explored to make language models (LMs) more efficient.In particular, errors caused by children{'}s regularization (so-called overregularization, e.g., using wroted for the past tense of write) have been widely studied to reveal the mechanisms of language acquisition. Existing research has analyzed regularization in language acquisition only by modeling word inflection directly, which is unnatural in light of human language acquisition. In this paper, we hypothesize that language models that imitate the errors children make during language acquisition have a learning process more similar to humans. To verify this hypothesis, we analyzed the learning curve and error preferences of verb inflections in small-scale LMs using acceptability judgments. We analyze the differences in results by model architecture, data, and tokenization. Our model shows child-like U-shaped learning curves clearly for certain verbs, but the preferences for types of overgeneralization did not fully match the observations in children.",
}
|
The imitation of the children{'}s language acquisition process has been explored to make language models (LMs) more efficient.In particular, errors caused by children{'}s regularization (so-called overregularization, e.g., using wroted for the past tense of write) have been widely studied to reveal the mechanisms of language acquisition. Existing research has analyzed regularization in language acquisition only by modeling word inflection directly, which is unnatural in light of human language acquisition. In this paper, we hypothesize that language models that imitate the errors children make during language acquisition have a learning process more similar to humans. To verify this hypothesis, we analyzed the learning curve and error preferences of verb inflections in small-scale LMs using acceptability judgments. We analyze the differences in results by model architecture, data, and tokenization. Our model shows child-like U-shaped learning curves clearly for certain verbs, but the preferences for types of overgeneralization did not fully match the observations in children.
|
[
"Haga, Akari",
"Sugawara, Saku",
"Fukatsu, Akiyo",
"Oba, Miyu",
"Ouchi, Hiroki",
"Watanabe, Taro",
"Oseki, Yohei"
] |
Modeling Overregularization in Children with Small Language Models
|
findings-acl.865
|
Poster
|
2308.00073v1
|
https://aclanthology.org/2024.findings-acl.866.bib
|
@inproceedings{liu-etal-2024-fantastic,
title = "Fantastic Semantics and Where to Find Them: Investigating Which Layers of Generative {LLM}s Reflect Lexical Semantics",
author = "Liu, Zhu and
Kong, Cunliang and
Liu, Ying and
Sun, Maosong",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.866",
pages = "14551--14558",
abstract = "Large language models have achieved remarkable success in general language understanding tasks. However, as a family of generative methods with the objective of next token prediction, the semantic evolution with the depth of these models are not fully explored, unlike their predecessors, such as BERT-like architectures. In this paper, we specifically investigate the bottom-up evolution of lexical semantics for a popular LLM, namely Llama2, by probing its hidden states at the end of each layer using a contextualized word identification task. Our experiments show that the representations in lower layers encode lexical semantics, while the higher layers, with weaker semantic induction, are responsible for prediction. This is in contrast to models with discriminative objectives, such as mask language modeling, where the higher layers obtain better lexical semantics. The conclusion is further supported by the monotonic increase in performance via the hidden states for the last meaningless symbols, such as punctuation, in the prompting strategy. Our codes are available at https://github.com/RyanLiut/LLM{\_}LexSem.",
}
|
Large language models have achieved remarkable success in general language understanding tasks. However, as a family of generative methods with the objective of next token prediction, the semantic evolution with the depth of these models are not fully explored, unlike their predecessors, such as BERT-like architectures. In this paper, we specifically investigate the bottom-up evolution of lexical semantics for a popular LLM, namely Llama2, by probing its hidden states at the end of each layer using a contextualized word identification task. Our experiments show that the representations in lower layers encode lexical semantics, while the higher layers, with weaker semantic induction, are responsible for prediction. This is in contrast to models with discriminative objectives, such as mask language modeling, where the higher layers obtain better lexical semantics. The conclusion is further supported by the monotonic increase in performance via the hidden states for the last meaningless symbols, such as punctuation, in the prompting strategy. Our codes are available at https://github.com/RyanLiut/LLM{\_}LexSem.
|
[
"Liu, Zhu",
"Kong, Cunliang",
"Liu, Ying",
"Sun, Maosong"
] |
Fantastic Semantics and Where to Find Them: Investigating Which Layers of Generative {LLM}s Reflect Lexical Semantics
|
findings-acl.866
|
Poster
|
2403.01509v2
|
https://aclanthology.org/2024.findings-acl.867.bib
|
@inproceedings{zhong-etal-2024-harnessing,
title = "Harnessing Large Language Models as Post-hoc Correctors",
author = "Zhong, Zhiqiang and
Zhou, Kuangyu and
Mottin, Davide",
editor = "Ku, Lun-Wei and
Martins, Andre and
Srikumar, Vivek",
booktitle = "Findings of the Association for Computational Linguistics ACL 2024",
month = aug,
year = "2024",
address = "Bangkok, Thailand and virtual meeting",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-acl.867",
pages = "14559--14574",
abstract = "As Machine Learning (ML) models grow in size and demand higher-quality training data, the expenses associated with re-training and fine-tuning these models are escalating rapidly. Inspired by recent impressive achievements of Large Language Models (LLMs) in different fields, this paper delves into the question: can LLMs efficiently improve an ML{'}s performance at a minimal cost? We show that, through our proposed training-free framework LLMCorr, an LLM can work as a post-hoc corrector to propose corrections for the predictions of an arbitrary ML model. In particular, we form a contextual knowledge database by incorporating the dataset{'}s label information and the ML model{'}s predictions on the validation dataset. Leveraging the in-context learning capability of LLMs, we ask the LLM to summarise the instances in which the ML model makes mistakes and the correlation between primary predictions and true labels. Following this, the LLM can transfer its acquired knowledge to suggest corrections for the ML model{'}s predictions. Our experimental results on text analysis and the challenging molecular predictions show that LLMCorr improves the performance of a number of models by up to 39{\%}.",
}
|
As Machine Learning (ML) models grow in size and demand higher-quality training data, the expenses associated with re-training and fine-tuning these models are escalating rapidly. Inspired by recent impressive achievements of Large Language Models (LLMs) in different fields, this paper delves into the question: can LLMs efficiently improve an ML{'}s performance at a minimal cost? We show that, through our proposed training-free framework LLMCorr, an LLM can work as a post-hoc corrector to propose corrections for the predictions of an arbitrary ML model. In particular, we form a contextual knowledge database by incorporating the dataset{'}s label information and the ML model{'}s predictions on the validation dataset. Leveraging the in-context learning capability of LLMs, we ask the LLM to summarise the instances in which the ML model makes mistakes and the correlation between primary predictions and true labels. Following this, the LLM can transfer its acquired knowledge to suggest corrections for the ML model{'}s predictions. Our experimental results on text analysis and the challenging molecular predictions show that LLMCorr improves the performance of a number of models by up to 39{\%}.
|
[
"Zhong, Zhiqiang",
"Zhou, Kuangyu",
"Mottin, Davide"
] |
Harnessing Large Language Models as Post-hoc Correctors
|
findings-acl.867
|
Poster
|
2402.13414v2
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.