publication_date
stringlengths 20
20
| doc_url
stringlengths 33
34
| id
stringlengths 33
34
| authors
stringlengths 5
20.1k
| update_date
stringlengths 20
20
| category_all
stringlengths 5
138
| abstract
stringlengths 123
3.09k
| category_primary
stringclasses 78
values | title
stringlengths 10
211
|
|---|---|---|---|---|---|---|---|---|
2024-06-20T22:56:31Z | http://arxiv.org/pdf/2406.14774v1 | http://arxiv.org/abs/2406.14774v1 | Ivana Kajić, Olivia Wiles, Isabela Albuquerque, Matthias Bauer, Su Wang, Jordi Pont-Tuset, Aida Nematzadeh | 2024-06-20T22:56:31Z | cs.LG, cs.CL, cs.CV | Text-to-image generative models are capable of producing high-quality images
that often faithfully depict concepts described using natural language. In this
work, we comprehensively evaluate a range of text-to-image models on numerical
reasoning tasks of varying difficulty, and show that even the most advanced
models have only rudimentary numerical skills. Specifically, their ability to
correctly generate an exact number of objects in an image is limited to small
numbers, it is highly dependent on the context the number term appears in, and
it deteriorates quickly with each successive number. We also demonstrate that
models have poor understanding of linguistic quantifiers (such as "a few" or
"as many as"), the concept of zero, and struggle with more advanced concepts
such as partial quantities and fractional representations. We bundle prompts,
generated images and human annotations into GeckoNum, a novel benchmark for
evaluation of numerical reasoning. | cs.LG | Evaluating Numerical Reasoning in Text-to-Image Models |
2024-06-20T22:30:06Z | http://arxiv.org/pdf/2406.14765v1 | http://arxiv.org/abs/2406.14765v1 | Steven A. Lehr, Aylin Caliskan, Suneragiri Liyanage, Mahzarin R. Banaji | 2024-06-20T22:30:06Z | cs.AI, cs.CL, cs.CY, cs.IR, cs.LG, I.2.7; K.4.0; K.4.1; K.4.2 | How good a research scientist is ChatGPT? We systematically probed the
capabilities of GPT-3.5 and GPT-4 across four central components of the
scientific process: as a Research Librarian, Research Ethicist, Data Generator,
and Novel Data Predictor, using psychological science as a testing field. In
Study 1 (Research Librarian), unlike human researchers, GPT-3.5 and GPT-4
hallucinated, authoritatively generating fictional references 36.0% and 5.4% of
the time, respectively, although GPT-4 exhibited an evolving capacity to
acknowledge its fictions. In Study 2 (Research Ethicist), GPT-4 (though not
GPT-3.5) proved capable of detecting violations like p-hacking in fictional
research protocols, correcting 88.6% of blatantly presented issues, and 72.6%
of subtly presented issues. In Study 3 (Data Generator), both models
consistently replicated patterns of cultural bias previously discovered in
large language corpora, indicating that ChatGPT can simulate known results, an
antecedent to usefulness for both data generation and skills like hypothesis
generation. Contrastingly, in Study 4 (Novel Data Predictor), neither model was
successful at predicting new results absent in their training data, and neither
appeared to leverage substantially new information when predicting more versus
less novel outcomes. Together, these results suggest that GPT is a flawed but
rapidly improving librarian, a decent research ethicist already, capable of
data generation in simple domains with known characteristics but poor at
predicting novel patterns of empirical data to aid future experimentation. | cs.AI | ChatGPT as Research Scientist: Probing GPT's Capabilities as a Research Librarian, Research Ethicist, Data Generator and Data Predictor |
2024-06-20T22:28:11Z | http://arxiv.org/pdf/2406.14764v1 | http://arxiv.org/abs/2406.14764v1 | William Fleshman, Benjamin Van Durme | 2024-06-20T22:28:11Z | cs.IR, cs.AI, cs.CL, cs.LG | Large language models (LLMs) fine-tuned for text-retrieval have demonstrated
state-of-the-art results across several information retrieval (IR) benchmarks.
However, supervised training for improving these models requires numerous
labeled examples, which are generally unavailable or expensive to acquire. In
this work, we explore the effectiveness of extending reverse engineered
adaptation to the context of information retrieval (RE-AdaptIR). We use
RE-AdaptIR to improve LLM-based IR models using only unlabeled data. We
demonstrate improved performance both in training domains as well as zero-shot
in domains where the models have seen no queries. We analyze performance
changes in various fine-tuning scenarios and offer findings of immediate use to
practitioners. | cs.IR | RE-AdaptIR: Improving Information Retrieval through Reverse Engineered Adaptation |
2024-06-20T22:22:41Z | http://arxiv.org/pdf/2406.14763v1 | http://arxiv.org/abs/2406.14763v1 | Lingxi Zhang, Jing Zhang, Yanling Wang, Cuiping Li, Hong Chen | 2024-06-20T22:22:41Z | cs.CL, cs.AI | Large-scale knowledge bases (KBs) like Freebase and Wikidata house millions
of structured knowledge. Knowledge Base Question Answering (KBQA) provides a
user-friendly way to access these valuable KBs via asking natural language
questions. In order to improve the generalization capabilities of KBQA models,
extensive research has embraced a retrieve-then-reason framework to retrieve
relevant evidence for logical expression generation. These multi-stage efforts
prioritize acquiring external sources but overlook the incorporation of new
knowledge into their model parameters. In effect, even advanced language models
and retrievers have knowledge boundaries, thereby limiting the generalization
capabilities of previous KBQA models. Therefore, this paper develops KBLLaMA,
which follows a learn-then-reason framework to inject new KB knowledge into a
large language model for flexible end-to-end KBQA. At the core of KBLLaMA, we
study (1) how to organize new knowledge about KBQA and (2) how to facilitate
the learning of the organized knowledge. Extensive experiments on various KBQA
generalization tasks showcase the state-of-the-art performance of KBLLaMA.
Especially on the general benchmark GrailQA and domain-specific benchmark
Bio-chemical, KBLLaMA respectively derives a performance gain of up to 3.8% and
9.8% compared to the baselines. | cs.CL | A Learn-Then-Reason Model Towards Generalization in Knowledge Base Question Answering |
2024-06-20T22:10:52Z | http://arxiv.org/pdf/2406.14760v1 | http://arxiv.org/abs/2406.14760v1 | Lexin Zhou, Youmna Farag, Andreas Vlachos | 2024-06-20T22:10:52Z | cs.CL, cs.AI, cs.LG | Research on dialogue constructiveness assessment focuses on (i) analysing
conversational factors that influence individuals to take specific actions, win
debates, change their perspectives or broaden their open-mindedness and (ii)
predicting constructive outcomes following dialogues for such use cases. These
objectives can be achieved by training either interpretable feature-based
models (which often involve costly human annotations) or neural models such as
pre-trained language models (which have empirically shown higher task accuracy
but lack interpretability). We propose a novel LLM feature-based framework that
combines the strengths of feature-based and neural approaches while mitigating
their downsides, in assessing dialogue constructiveness. The framework first
defines a set of dataset-independent and interpretable linguistic features,
which can be extracted by both prompting an LLM and simple heuristics. Such
features are then used to train LLM feature-based models. We apply this
framework to three datasets of dialogue constructiveness and find that our LLM
feature-based models significantly outperform standard feature-based models and
neural models, and tend to learn more robust prediction rules instead of
relying on superficial shortcuts (as seen with neural models). Further, we
demonstrate that interpreting these LLM feature-based models can yield valuable
insights into what makes a dialogue constructive. | cs.CL | An LLM Feature-based Framework for Dialogue Constructiveness Assessment |
2024-06-20T21:40:02Z | http://arxiv.org/pdf/2406.15516v1 | http://arxiv.org/abs/2406.15516v1 | Ali Aliyev | 2024-06-20T21:40:02Z | cs.CL, cs.SD, eess.AS | This paper describes our solution for the Diarization of Speaker and Language
in Conversational Environments Challenge (Displace 2023). We used a combination
of VAD for finding segfments with speech, Resnet architecture based CNN for
feature extraction from these segments, and spectral clustering for features
clustering. Even though it was not trained with using Hindi, the described
algorithm achieves the following metrics: DER 27. 1% and DER 27. 4%, on the
development and phase-1 evaluation parts of the dataset, respectively. | cs.CL | System Description for the Displace Speaker Diarization Challenge 2023 |
2024-06-20T21:39:04Z | http://arxiv.org/pdf/2406.14747v1 | http://arxiv.org/abs/2406.14747v1 | Varsha Suresh, Salah Aït-Mokhtar, Caroline Brun, Ioan Calapodescu | 2024-06-20T21:39:04Z | cs.CL, cs.AI | Self-supervised learning models have revolutionized the field of speech
processing. However, the process of fine-tuning these models on downstream
tasks requires substantial computational resources, particularly when dealing
with multiple speech-processing tasks. In this paper, we explore the potential
of adapter-based fine-tuning in developing a unified model capable of
effectively handling multiple spoken language processing tasks. The tasks we
investigate are Automatic Speech Recognition, Phoneme Recognition, Intent
Classification, Slot Filling, and Spoken Emotion Recognition. We validate our
approach through a series of experiments on the SUPERB benchmark, and our
results indicate that adapter-based fine-tuning enables a single
encoder-decoder model to perform multiple speech processing tasks with an
average improvement of 18.4% across the five target tasks while staying
efficient in terms of parameter updates. | cs.CL | An Adapter-Based Unified Model for Multiple Spoken Language Processing Tasks |
2024-06-20T21:27:57Z | http://arxiv.org/pdf/2406.14745v2 | http://arxiv.org/abs/2406.14745v2 | Sefika Efeoglu, Adrian Paschke | 2024-06-24T06:57:05Z | cs.CL, cs.AI | Information Extraction (IE) is crucial for converting unstructured data into
structured formats like Knowledge Graphs (KGs). A key task within IE is
Relation Extraction (RE), which identifies relationships between entities in
text. Various RE methods exist, including supervised, unsupervised, weakly
supervised, and rule-based approaches. Recent studies leveraging pre-trained
language models (PLMs) have shown significant success in this area. In the
current era dominated by Large Language Models (LLMs), fine-tuning these models
can overcome limitations associated with zero-shot LLM prompting-based RE
methods, especially regarding domain adaptation challenges and identifying
implicit relations between entities in sentences. These implicit relations,
which cannot be easily extracted from a sentence's dependency tree, require
logical inference for accurate identification. This work explores the
performance of fine-tuned LLMs and their integration into the Retrieval
Augmented-based (RAG) RE approach to address the challenges of identifying
implicit relations at the sentence level, particularly when LLMs act as
generators within the RAG framework. Empirical evaluations on the TACRED,
TACRED-Revisited (TACREV), Re-TACRED, and SemEVAL datasets show significant
performance improvements with fine-tuned LLMs, including Llama2-7B, Mistral-7B,
and T5 (Large). Notably, our approach achieves substantial gains on SemEVAL,
where implicit relations are common, surpassing previous results on this
dataset. Additionally, our method outperforms previous works on TACRED, TACREV,
and Re-TACRED, demonstrating exceptional performance across diverse evaluation
scenarios. | cs.CL | Relation Extraction with Fine-Tuned Large Language Models in Retrieval Augmented Generation Frameworks |
2024-06-20T21:07:55Z | http://arxiv.org/pdf/2406.14739v1 | http://arxiv.org/abs/2406.14739v1 | Yunmo Chen, Tongfei Chen, Harsh Jhamtani, Patrick Xia, Richard Shin, Jason Eisner, Benjamin Van Durme | 2024-06-20T21:07:55Z | cs.CL | We introduce iterative retrieval, a novel framework that empowers retrievers
to make iterative decisions through policy optimization. Finding an optimal
portfolio of retrieved items is a combinatorial optimization problem, generally
considered NP-hard. This approach provides a learned approximation to such a
solution, meeting specific task requirements under a given family of large
language models (LLMs). We propose a training procedure based on reinforcement
learning, incorporating feedback from LLMs. We instantiate an iterative
retriever for composing in-context learning (ICL) exemplars and apply it to
various semantic parsing tasks that demand synthesized programs as outputs. By
adding only 4M additional parameters for state encoding, we convert an
off-the-shelf dense retriever into a stateful iterative retriever,
outperforming previous methods in selecting ICL exemplars on semantic parsing
datasets such as CalFlow, TreeDST, and MTOP. Additionally, the trained
iterative retriever generalizes across different inference LLMs beyond the one
used during training. | cs.CL | Learning to Retrieve Iteratively for In-Context Learning |
2024-06-20T21:02:30Z | http://arxiv.org/pdf/2406.14737v1 | http://arxiv.org/abs/2406.14737v1 | Zhiqiang Pi, Annapurna Vadaparty, Benjamin K. Bergen, Cameron R. Jones | 2024-06-20T21:02:30Z | cs.CL | Recent empirical results have sparked a debate about whether or not Large
Language Models (LLMs) are capable of Theory of Mind (ToM). While some have
found LLMs to be successful on ToM evaluations such as the False Belief task
(Kosinski, 2023), others have argued that LLMs solve these tasks by exploiting
spurious correlations -- not representing beliefs -- since they fail on trivial
alterations to these tasks (Ullman, 2023). In this paper, we introduce SCALPEL:
a technique to generate targeted modifications for False Belief tasks to test
different specific hypotheses about why LLMs fail. We find that modifications
which make explicit common inferences -- such as that looking at a transparent
object implies recognizing its contents -- preserve LLMs' performance. This
suggests that LLMs' failures on modified ToM tasks could result from a lack of
more general commonsense reasoning, rather than a failure to represent mental
states. We argue that SCALPEL could be helpful for explaining LLM successes and
failures in other cases. | cs.CL | Dissecting the Ullman Variations with a SCALPEL: Why do LLMs fail at Trivial Alterations to the False Belief Task? |
2024-06-20T20:55:38Z | http://arxiv.org/pdf/2406.14732v1 | http://arxiv.org/abs/2406.14732v1 | Jayetri Bardhan, Bushi Xiao, Daisy Zhe Wang | 2024-06-20T20:55:38Z | cs.CL, cs.IR | Question answering (QA) over tables and text has gained much popularity over
the years. Multi-hop table-text QA requires multiple hops between the table and
text, making it a challenging QA task. Although several works have attempted to
solve the table-text QA task, most involve training the models and requiring
labeled data. In this paper, we have proposed a model - TTQA-RS: A break-down
prompting approach for Multi-hop Table-Text Question Answering with Reasoning
and Summarization. Our model uses augmented knowledge including table-text
summary with decomposed sub-question with answer for a reasoning-based
table-text QA. Using open-source language models our model outperformed all
existing prompting methods for table-text QA tasks on existing table-text QA
datasets like HybridQA and OTT-QA's development set. Our results are comparable
with the training-based state-of-the-art models, demonstrating the potential of
prompt-based approaches using open-source LLMs. Additionally, by using GPT-4
with LLaMA3-70B, our model achieved state-of-the-art performance for
prompting-based methods on multi-hop table-text QA. | cs.CL | TTQA-RS- A break-down prompting approach for Multi-hop Table-Text Question Answering with Reasoning and Summarization |
2024-06-20T20:32:53Z | http://arxiv.org/pdf/2406.14721v1 | http://arxiv.org/abs/2406.14721v1 | Yue Huang, Chenrui Fan, Yuan Li, Siyuan Wu, Tianyi Zhou, Xiangliang Zhang, Lichao Sun | 2024-06-20T20:32:53Z | cs.CL | Large Language Models (LLMs) have garnered significant attention due to their
remarkable ability to process information across various languages. Despite
their capabilities, they exhibit inconsistencies in handling identical queries
in different languages, presenting challenges for further advancement. This
paper introduces a method to enhance the multilingual performance of LLMs by
aggregating knowledge from diverse languages. This approach incorporates a
low-resource knowledge detector specific to a language, a language selection
process, and mechanisms for answer replacement and integration. Our experiments
demonstrate notable performance improvements, particularly in reducing language
performance disparity. An ablation study confirms that each component of our
method significantly contributes to these enhancements. This research
highlights the inherent potential of LLMs to harmonize multilingual
capabilities and offers valuable insights for further exploration. | cs.CL | 1+1>2: Can Large Language Models Serve as Cross-Lingual Knowledge Aggregators? |
2024-06-20T20:09:37Z | http://arxiv.org/pdf/2406.14711v2 | http://arxiv.org/abs/2406.14711v2 | Alfonso Amayuelas, Xianjun Yang, Antonis Antoniades, Wenyue Hua, Liangming Pan, William Wang | 2024-06-26T16:05:20Z | cs.CL, cs.AI, cs.MA | Large Language Models (LLMs) have shown exceptional results on current
benchmarks when working individually. The advancement in their capabilities,
along with a reduction in parameter size and inference times, has facilitated
the use of these models as agents, enabling interactions among multiple models
to execute complex tasks. Such collaborations offer several advantages,
including the use of specialized models (e.g. coding), improved confidence
through multiple computations, and enhanced divergent thinking, leading to more
diverse outputs. Thus, the collaborative use of language models is expected to
grow significantly in the coming years. In this work, we evaluate the behavior
of a network of models collaborating through debate under the influence of an
adversary. We introduce pertinent metrics to assess the adversary's
effectiveness, focusing on system accuracy and model agreement. Our findings
highlight the importance of a model's persuasive ability in influencing others.
Additionally, we explore inference-time methods to generate more compelling
arguments and evaluate the potential of prompt-based mitigation as a defensive
strategy. | cs.CL | MultiAgent Collaboration Attack: Investigating Adversarial Attacks in Large Language Model Collaborations via Debate |
2024-06-20T20:03:37Z | http://arxiv.org/pdf/2406.14709v1 | http://arxiv.org/abs/2406.14709v1 | Rongxin Zhu, Jey Han Lau, Jianzhong Qi | 2024-06-20T20:03:37Z | cs.CL, F.2.2; I.2.7 | Factual consistency is an important quality in dialogue summarization. Large
language model (LLM)-based automatic text summarization models generate more
factually consistent summaries compared to those by smaller pretrained language
models, but they face deployment challenges in real-world applications due to
privacy or resource constraints. In this paper, we investigate the use of
symbolic knowledge distillation to improve the factual consistency of smaller
pretrained models for dialogue summarization. We employ zero-shot learning to
extract symbolic knowledge from LLMs, generating both factually consistent
(positive) and inconsistent (negative) summaries. We then apply two contrastive
learning objectives on these summaries to enhance smaller summarization models.
Experiments with BART, PEGASUS, and Flan-T5 indicate that our approach
surpasses strong baselines that rely on complex data augmentation strategies.
Our approach achieves better factual consistency while maintaining coherence,
fluency, and relevance, as confirmed by various automatic evaluation metrics.
We also provide access to the data and code to facilitate future research. | cs.CL | Factual Dialogue Summarization via Learning from Large Language Models |
2024-06-20T19:50:56Z | http://arxiv.org/pdf/2406.14703v1 | http://arxiv.org/abs/2406.14703v1 | Seungbeen Lee, Seungwon Lim, Seungju Han, Giyeong Oh, Hyungjoo Chae, Jiwan Chung, Minju Kim, Beong-woo Kwak, Yeonsoo Lee, Dongha Lee, Jinyoung Yeo, Youngjae Yu | 2024-06-20T19:50:56Z | cs.CL, cs.AI | The idea of personality in descriptive psychology, traditionally defined
through observable behavior, has now been extended to Large Language Models
(LLMs) to better understand their behavior. This raises a question: do LLMs
exhibit distinct and consistent personality traits, similar to humans? Existing
self-assessment personality tests, while applicable, lack the necessary
validity and reliability for precise personality measurements. To address this,
we introduce TRAIT, a new tool consisting of 8K multi-choice questions designed
to assess the personality of LLMs with validity and reliability. TRAIT is built
on the psychometrically validated human questionnaire, Big Five Inventory (BFI)
and Short Dark Triad (SD-3), enhanced with the ATOMIC10X knowledge graph for
testing personality in a variety of real scenarios. TRAIT overcomes the
reliability and validity issues when measuring personality of LLM with
self-assessment, showing the highest scores across three metrics: refusal rate,
prompt sensitivity, and option order sensitivity. It reveals notable insights
into personality of LLM: 1) LLMs exhibit distinct and consistent personality,
which is highly influenced by their training data (i.e., data used for
alignment tuning), and 2) current prompting techniques have limited
effectiveness in eliciting certain traits, such as high psychopathy or low
conscientiousness, suggesting the need for further research in this direction. | cs.CL | Do LLMs Have Distinct and Consistent Personality? TRAIT: Personality Testset designed for LLMs with Psychometrics |
2024-06-20T19:50:49Z | http://arxiv.org/pdf/2406.14701v1 | http://arxiv.org/abs/2406.14701v1 | Murali Karthick Baskar, Andrew Rosenberg, Bhuvana Ramabhadran, Neeraj Gaur, Zhong Meng | 2024-06-20T19:50:49Z | cs.AI, cs.CL, cs.SD, eess.AS | In this paper, we focus on addressing the constraints faced when applying
LLMs to ASR. Recent works utilize prefixLM-type models, which directly apply
speech as a prefix to LLMs for ASR. We have found that optimizing speech
prefixes leads to better ASR performance and propose applying RNNT loss to
perform speech prefix-tuning. This is a simple approach and does not increase
the model complexity or alter the inference pipeline. We also propose
language-based soft prompting to further improve with frozen LLMs. Empirical
analysis on realtime testset from 10 Indic languages demonstrate that our
proposed speech prefix-tuning yields improvements with both frozen and
fine-tuned LLMs. Our recognition results on an average of 10 Indics show that
the proposed prefix-tuning with RNNT loss results in a 12\% relative
improvement in WER over the baseline with a fine-tuned LLM. Our proposed
approches with the frozen LLM leads to a 31\% relative improvement over basic
soft-prompting prefixLM. | cs.AI | Speech Prefix-Tuning with RNNT Loss for Improving LLM Predictions |
2024-06-20T19:35:17Z | http://arxiv.org/pdf/2406.14695v1 | http://arxiv.org/abs/2406.14695v1 | Parker Seegmiller, Joseph Gatto, Sarah Masud Preum | 2024-06-20T19:35:17Z | cs.CL, cs.AI | Recent evaluations of cross-domain text classification models aim to measure
the ability of a model to obtain domain-invariant performance in a target
domain given labeled samples in a source domain. The primary strategy for this
evaluation relies on assumed differences between source domain samples and
target domain samples in benchmark datasets. This evaluation strategy fails to
account for the similarity between source and target domains, and may mask when
models fail to transfer learning to specific target samples which are highly
dissimilar from the source domain. We introduce Depth $F_1$, a novel
cross-domain text classification performance metric. Designed to be
complementary to existing classification metrics such as $F_1$, Depth $F_1$
measures how well a model performs on target samples which are dissimilar from
the source domain. We motivate this metric using standard cross-domain text
classification datasets and benchmark several recent cross-domain text
classification models, with the goal of enabling in-depth evaluation of the
semantic generalizability of cross-domain text classification models. | cs.CL | Depth $F_1$: Improving Evaluation of Cross-Domain Text Classification by Measuring Semantic Generalizability |
2024-06-20T19:20:00Z | http://arxiv.org/pdf/2406.14686v1 | http://arxiv.org/abs/2406.14686v1 | Alkis Koudounas, Flavio Giobergia, Eliana Pastor, Elena Baralis | 2024-06-20T19:20:00Z | cs.CL, cs.LG, eess.AS | Speech models may be affected by performance imbalance in different
population subgroups, raising concerns about fair treatment across these
groups. Prior attempts to mitigate unfairness either focus on user-defined
subgroups, potentially overlooking other affected subgroups, or do not
explicitly improve the internal representation at the subgroup level. This
paper proposes the first adoption of contrastive learning to mitigate speech
model bias in underperforming subgroups. We employ a three-level learning
technique that guides the model in focusing on different scopes for the
contrastive loss, i.e., task, subgroup, and the errors within subgroups. The
experiments on two spoken language understanding datasets and two languages
demonstrate that our approach improves internal subgroup representations, thus
reducing model bias and enhancing performance. | cs.CL | A Contrastive Learning Approach to Mitigate Bias in Speech Models |
2024-06-20T19:11:35Z | http://arxiv.org/pdf/2406.14683v1 | http://arxiv.org/abs/2406.14683v1 | Jiarui Feng, Hao Liu, Lecheng Kong, Yixin Chen, Muhan Zhang | 2024-06-20T19:11:35Z | cs.LG, cs.CL | In this report, we present TAGLAS, an atlas of text-attributed graph (TAG)
datasets and benchmarks. TAGs are graphs with node and edge features
represented in text, which have recently gained wide applicability in training
graph-language or graph foundation models. In TAGLAS, we collect and integrate
more than 23 TAG datasets with domains ranging from citation graphs to molecule
graphs and tasks from node classification to graph question-answering. Unlike
previous graph datasets and benchmarks, all datasets in TAGLAS have a unified
node and edge text feature format, which allows a graph model to be
simultaneously trained and evaluated on multiple datasets from various domains.
Further, we provide a standardized, efficient, and simplified way to load all
datasets and tasks. We also provide useful utils like text-to-embedding
conversion, and graph-to-text conversion, which can facilitate different
evaluation scenarios. Finally, we also provide standard and easy-to-use
evaluation utils. The project is open-sourced at
https://github.com/JiaruiFeng/TAGLAS and is still under construction. Please
expect more datasets/features in the future. | cs.LG | TAGLAS: An atlas of text-attributed graph datasets in the era of large graph and language models |
2024-06-20T18:59:46Z | http://arxiv.org/pdf/2406.14680v1 | http://arxiv.org/abs/2406.14680v1 | Taraka Rama, Sowmya Vajjala | 2024-06-20T18:59:46Z | cs.CL | The Universal Dependencies (UD) project aims to create a cross-linguistically
consistent dependency annotation for multiple languages, to facilitate
multilingual NLP. It currently supports 114 languages. Dravidian languages are
spoken by over 200 million people across the word, and yet there are only two
languages from this family in UD. This paper examines some of the morphological
and syntactic features of Dravidian languages and explores how they can be
annotated in the UD framework. | cs.CL | Dravidian language family through Universal Dependencies lens |
2024-06-20T18:58:11Z | http://arxiv.org/pdf/2406.14678v1 | http://arxiv.org/abs/2406.14678v1 | Pamela D. Rivière, Anne L. Beatty-Martínez, Sean Trott | 2024-06-20T18:58:11Z | cs.CL | Lexical ambiguity -- where a single wordform takes on distinct,
context-dependent meanings -- serves as a useful tool to compare across
different large language models' (LLMs') ability to form distinct,
contextualized representations of the same stimulus. Few studies have
systematically compared LLMs' contextualized word embeddings for languages
beyond English. Here, we evaluate multiple bidirectional transformers' (BERTs')
semantic representations of Spanish ambiguous nouns in context. We develop a
novel dataset of minimal-pair sentences evoking the same or different sense for
a target ambiguous noun. In a pre-registered study, we collect contextualized
human relatedness judgments for each sentence pair. We find that various
BERT-based LLMs' contextualized semantic representations capture some variance
in human judgments but fall short of the human benchmark, and for Spanish --
unlike English -- model scale is uncorrelated with performance. We also
identify stereotyped trajectories of target noun disambiguation as a proportion
of traversal through a given LLM family's architecture, which we partially
replicate in English. We contribute (1) a dataset of controlled, Spanish
sentence stimuli with human relatedness norms, and (2) to our evolving
understanding of the impact that LLM specification (architectures, training
protocols) exerts on contextualized embeddings. | cs.CL | Bidirectional Transformer Representations of (Spanish) Ambiguous Words in Context: A New Lexical Resource and Empirical Analysis |
2024-06-20T18:50:44Z | http://arxiv.org/pdf/2406.14673v1 | http://arxiv.org/abs/2406.14673v1 | Taiming Lu, Muhan Gao, Kuai Yu, Adam Byerly, Daniel Khashabi | 2024-06-20T18:50:44Z | cs.CL | Large Language Models (LLMs) exhibit positional bias, struggling to utilize
information from the middle or end of long contexts. Our study explores LLMs'
long-context reasoning by probing their hidden representations. We find that
while LLMs encode the position of target information, they often fail to
leverage this in generating accurate responses. This reveals a disconnect
between information retrieval and utilization, a "know but don't tell"
phenomenon. We further analyze the relationship between extraction time and
final accuracy, offering insights into the underlying mechanics of transformer
models. | cs.CL | Insights into LLM Long-Context Failures: When Transformers Know but Don't Tell |
2024-06-20T18:47:43Z | http://arxiv.org/pdf/2406.14670v1 | http://arxiv.org/abs/2406.14670v1 | Atula Tejaswi, Nilesh Gupta, Eunsol Choi | 2024-06-20T18:47:43Z | cs.CL, cs.AI, cs.LG | Despite rapid progress in large language models (LLMs), their performance on
a vast majority of languages remain unsatisfactory. In this paper, we study
building language-specific LLMs by adapting monolingual and multilingual LLMs.
We conduct systematic experiments on how design choices (base model selection,
vocabulary extension, and continued fine-tuning) impact the adapted LLM, both
in terms of efficiency (how many tokens are needed to encode the same amount of
information) and end task performance. We find that (1) the initial performance
before the adaptation is not always indicative of the final performance. (2)
Efficiency can easily improved with simple vocabulary extension and continued
fine-tuning in most LLMs we study, and (3) The optimal adaptation method is
highly language-dependent, and the simplest approach works well across various
experimental settings. Adapting English-centric models can yield better results
than adapting multilingual models despite their worse initial performance on
low-resource languages. Together, our work lays foundations on efficiently
building language-specific LLMs by adapting existing LLMs. | cs.CL | Exploring Design Choices for Building Language-Specific LLMs |
2024-06-20T18:37:36Z | http://arxiv.org/pdf/2406.15513v1 | http://arxiv.org/abs/2406.15513v1 | Jiaming Ji, Donghai Hong, Borong Zhang, Boyuan Chen, Josef Dai, Boren Zheng, Tianyi Qiu, Boxun Li, Yaodong Yang | 2024-06-20T18:37:36Z | cs.AI, cs.CL | In this work, we introduce the PKU-SafeRLHF dataset, designed to promote
research on safety alignment in large language models (LLMs). As a sibling
project to SafeRLHF and BeaverTails, we separate annotations of helpfulness and
harmlessness for question-answering pairs, providing distinct perspectives on
these coupled attributes. Overall, we provide 44.6k refined prompts and 265k
question-answer pairs with safety meta-labels for 19 harm categories and three
severity levels ranging from minor to severe, with answers generated by
Llama-family models. Based on this, we collected 166.8k preference data,
including dual-preference (helpfulness and harmlessness decoupled) and
single-preference data (trade-off the helpfulness and harmlessness from
scratch), respectively. Using the large-scale annotation data, we further train
severity-sensitive moderation for the risk control of LLMs and safety-centric
RLHF algorithms for the safety alignment of LLMs. We believe this dataset will
be a valuable resource for the community, aiding in the safe deployment of
LLMs. | cs.AI | PKU-SafeRLHF: A Safety Alignment Preference Dataset for Llama Family Models |
2024-06-20T18:35:47Z | http://arxiv.org/pdf/2406.14666v1 | http://arxiv.org/abs/2406.14666v1 | Mobashir Sadat, Cornelia Caragea | 2024-06-20T18:35:47Z | cs.CL | Scientific Natural Language Inference (NLI) is the task of predicting the
semantic relation between a pair of sentences extracted from research articles.
The automatic annotation method based on distant supervision for the training
set of SciNLI (Sadat and Caragea, 2022b), the first and most popular dataset
for this task, results in label noise which inevitably degenerates the
performance of classifiers. In this paper, we propose a novel co-training
method that assigns weights based on the training dynamics of the classifiers
to the distantly supervised labels, reflective of the manner they are used in
the subsequent training epochs. That is, unlike the existing semi-supervised
learning (SSL) approaches, we consider the historical behavior of the
classifiers to evaluate the quality of the automatically annotated labels.
Furthermore, by assigning importance weights instead of filtering out examples
based on an arbitrary threshold on the predicted confidence, we maximize the
usage of automatically labeled data, while ensuring that the noisy labels have
a minimal impact on model training. The proposed method obtains an improvement
of 1.5% in Macro F1 over the distant supervision baseline, and substantial
improvements over several other strong SSL baselines. We make our code and data
available on Github. | cs.CL | Co-training for Low Resource Scientific Natural Language Inference |
2024-06-20T18:22:59Z | http://arxiv.org/pdf/2406.14657v1 | http://arxiv.org/abs/2406.14657v1 | Allen Roush, Yusuf Shabazz, Arvind Balaji, Peter Zhang, Stefano Mezza, Markus Zhang, Sanjay Basu, Sriram Vishwanath, Mehdi Fatemi, Ravid Schwartz-Ziv | 2024-06-20T18:22:59Z | cs.CL, cs.AI, cs.LG | We introduce OpenDebateEvidence, a comprehensive dataset for argument mining
and summarization sourced from the American Competitive Debate community. This
dataset includes over 3.5 million documents with rich metadata, making it one
of the most extensive collections of debate evidence. OpenDebateEvidence
captures the complexity of arguments in high school and college debates,
providing valuable resources for training and evaluation. Our extensive
experiments demonstrate the efficacy of fine-tuning state-of-the-art large
language models for argumentative abstractive summarization across various
methods, models, and datasets. By providing this comprehensive resource, we aim
to advance computational argumentation and support practical applications for
debaters, educators, and researchers. OpenDebateEvidence is publicly available
to support further research and innovation in computational argumentation.
Access it here: https://huggingface.co/datasets/Yusuf5/OpenCaselist | cs.CL | OpenDebateEvidence: A Massive-Scale Argument Mining and Summarization Dataset |
2024-06-20T18:17:58Z | http://arxiv.org/pdf/2406.14654v1 | http://arxiv.org/abs/2406.14654v1 | Kawshik Manikantan, Shubham Toshniwal, Makarand Tapaswi, Vineet Gandhi | 2024-06-20T18:17:58Z | cs.CL, cs.AI, cs.LG, I.2.7 | The limited generalization of coreference resolution (CR) models has been a
major bottleneck in the task's broad application. Prior work has identified
annotation differences, especially for mention detection, as one of the main
reasons for the generalization gap and proposed using additional annotated
target domain data. Rather than relying on this additional annotation, we
propose an alternative formulation of the CR task, Major Entity Identification
(MEI), where we: (a) assume the target entities to be specified in the input,
and (b) limit the task to only the frequent entities. Through extensive
experiments, we demonstrate that MEI models generalize well across domains on
multiple datasets with supervised models and LLM-based few-shot prompting.
Additionally, the MEI task fits the classification framework, which enables the
use of classification-based metrics that are more robust than the current CR
metrics. Finally, MEI is also of practical use as it allows a user to search
for all mentions of a particular entity or a group of entities of interest. | cs.CL | Major Entity Identification: A Generalizable Alternative to Coreference Resolution |
2024-06-20T18:09:40Z | http://arxiv.org/pdf/2406.19415v1 | http://arxiv.org/abs/2406.19415v1 | Kim Trong Vu, Michael Krumdick, Varshini Reddy, Franck Dernoncourt, Viet Dac Lai | 2024-06-20T18:09:40Z | cs.CL | FActScore has gained popularity as a metric to estimate the factuality of
long-form texts generated by Large Language Models (LLMs) in English. However,
there has not been any work in studying the behavior of FActScore in other
languages. This paper studies the limitations of each component in the
four-component pipeline of FActScore in the multilingual setting. We introduce
a new dataset for FActScore on texts generated by strong multilingual LLMs. Our
evaluation shows that LLMs exhibit distinct behaviors in both fact extraction
and fact scoring tasks. No LLM produces consistent and reliable FActScore
across languages with varying levels of resources. We also find that the
knowledge source plays an important role in the quality of the estimated
FActScore. Using Wikipedia as the knowledge source may hinder the true
FActScore of long-form text due to its limited coverage in medium- and
low-resource languages. We also incorporate three mitigations to our knowledge
source that ultimately improve FActScore estimation across all languages. | cs.CL | An Analysis of Multilingual FActScore |
2024-06-20T18:07:26Z | http://arxiv.org/pdf/2406.14644v1 | http://arxiv.org/abs/2406.14644v1 | Chunyuan Deng, Yilun Zhao, Yuzhao Heng, Yitong Li, Jiannan Cao, Xiangru Tang, Arman Cohan | 2024-06-20T18:07:26Z | cs.CL | Data contamination has garnered increased attention in the era of large
language models (LLMs) due to the reliance on extensive internet-derived
training corpora. The issue of training corpus overlap with evaluation
benchmarks--referred to as contamination--has been the focus of significant
recent research. This body of work aims to identify contamination, understand
its impacts, and explore mitigation strategies from diverse perspectives.
However, comprehensive studies that provide a clear pathway from foundational
concepts to advanced insights are lacking in this nascent field. Therefore, we
present a comprehensive survey in the field of data contamination, laying out
the key issues, methodologies, and findings to date, and highlighting areas in
need of further research and development. In particular, we begin by examining
the effects of data contamination across various stages and forms. We then
provide a detailed analysis of current contamination detection methods,
categorizing them to highlight their focus, assumptions, strengths, and
limitations. We also discuss mitigation strategies, offering a clear guide for
future research. This survey serves as a succinct overview of the most recent
advancements in data contamination research, providing a straightforward guide
for the benefit of future research endeavors. | cs.CL | Unveiling the Spectrum of Data Contamination in Language Models: A Survey from Detection to Remediation |
2024-06-20T18:07:19Z | http://arxiv.org/pdf/2406.14643v1 | http://arxiv.org/abs/2406.14643v1 | Minqian Liu, Zhiyang Xu, Zihao Lin, Trevor Ashby, Joy Rimchala, Jiaxin Zhang, Lifu Huang | 2024-06-20T18:07:19Z | cs.CV, cs.AI, cs.CL | Interleaved text-and-image generation has been an intriguing research
direction, where the models are required to generate both images and text
pieces in an arbitrary order. Despite the emerging advancements in interleaved
generation, the progress in its evaluation still significantly lags behind.
Existing evaluation benchmarks do not support arbitrarily interleaved images
and text for both inputs and outputs, and they only cover a limited number of
domains and use cases. Also, current works predominantly use similarity-based
metrics which fall short in assessing the quality in open-ended scenarios. To
this end, we introduce InterleavedBench, the first benchmark carefully curated
for the evaluation of interleaved text-and-image generation. InterleavedBench
features a rich array of tasks to cover diverse real-world use cases. In
addition, we present InterleavedEval, a strong reference-free metric powered by
GPT-4o to deliver accurate and explainable evaluation. We carefully define five
essential evaluation aspects for InterleavedEval, including text quality,
perceptual quality, image coherence, text-image coherence, and helpfulness, to
ensure a comprehensive and fine-grained assessment. Through extensive
experiments and rigorous human evaluation, we show that our benchmark and
metric can effectively evaluate the existing models with a strong correlation
with human judgments surpassing previous reference-based metrics. We also
provide substantial findings and insights to foster future research in
interleaved generation and its evaluation. | cs.CV | Holistic Evaluation for Interleaved Text-and-Image Generation |
2024-06-20T18:00:17Z | http://arxiv.org/pdf/2406.14629v1 | http://arxiv.org/abs/2406.14629v1 | Xuefei Ning, Zifu Wang, Shiyao Li, Zinan Lin, Peiran Yao, Tianyu Fu, Matthew B. Blaschko, Guohao Dai, Huazhong Yang, Yu Wang | 2024-06-20T18:00:17Z | cs.CL, cs.AI | Teaching to improve student models (e.g., knowledge distillation) is an
extensively studied methodology in LLMs. However, for humans, teaching not only
improves students but also improves teachers. We ask: Can LLMs also learn by
teaching (LbT)? If yes, we can potentially unlock the possibility of
continuously advancing the models without solely relying on human-produced data
or stronger models. In this paper, we provide a preliminary exploration of this
ambitious agenda. We show that LbT ideas can be incorporated into existing LLM
training/prompting pipelines and provide noticeable improvements. Specifically,
we design three methods, each mimicking one of the three levels of LbT in
humans: observing students' feedback, learning from the feedback, and learning
iteratively, with the goals of improving answer accuracy without training and
improving models' inherent capability with fine-tuning. The findings are
encouraging. For example, similar to LbT in human, we see that: (1) LbT can
induce weak-to-strong generalization: strong models can improve themselves by
teaching other weak models; (2) Diversity in students might help: teaching
multiple students could be better than teaching one student or the teacher
itself. We hope that this early promise can inspire future research on LbT and
more broadly adopting the advanced techniques in education to improve LLMs. The
code is available at https://github.com/imagination-research/lbt. | cs.CL | Can LLMs Learn by Teaching? A Preliminary Study |
2024-06-20T17:59:58Z | http://arxiv.org/pdf/2406.14563v1 | http://arxiv.org/abs/2406.14563v1 | Hasan Abed Al Kader Hammoud, Umberto Michieli, Fabio Pizzati, Philip Torr, Adel Bibi, Bernard Ghanem, Mete Ozay | 2024-06-20T17:59:58Z | cs.CL, cs.AI, cs.LG | Merging Large Language Models (LLMs) is a cost-effective technique for
combining multiple expert LLMs into a single versatile model, retaining the
expertise of the original ones. However, current approaches often overlook the
importance of safety alignment during merging, leading to highly misaligned
models. This work investigates the effects of model merging on alignment. We
evaluate several popular model merging techniques, demonstrating that existing
methods do not only transfer domain expertise but also propagate misalignment.
We propose a simple two-step approach to address this problem: (i) generating
synthetic safety and domain-specific data, and (ii) incorporating these
generated data into the optimization process of existing data-aware model
merging techniques. This allows us to treat alignment as a skill that can be
maximized in the resulting merged LLM. Our experiments illustrate the
effectiveness of integrating alignment-related data during merging, resulting
in models that excel in both domain expertise and alignment. | cs.CL | Model Merging and Safety Alignment: One Bad Model Spoils the Bunch |
2024-06-20T17:59:45Z | http://arxiv.org/pdf/2406.14562v1 | http://arxiv.org/abs/2406.14562v1 | Sachit Menon, Richard Zemel, Carl Vondrick | 2024-06-20T17:59:45Z | cs.CL, cs.AI, cs.CV | When presented with questions involving visual thinking, humans naturally
switch reasoning modalities, often forming mental images or drawing visual
aids. Large language models have shown promising results in arithmetic and
symbolic reasoning by expressing intermediate reasoning in text as a chain of
thought, yet struggle to extend this capability to answer text queries that are
easily solved by visual reasoning, even with extensive multimodal pretraining.
We introduce a simple method, whiteboard-of-thought prompting, to unlock the
visual reasoning capabilities of multimodal large language models across
modalities. Whiteboard-of-thought prompting provides multimodal large language
models with a metaphorical `whiteboard' to draw out reasoning steps as images,
then returns these images back to the model for further processing. We find
this can be accomplished with no demonstrations or specialized modules, instead
leveraging models' existing ability to write code with libraries such as
Matplotlib and Turtle. This simple approach shows state-of-the-art results on
four difficult natural language tasks that involve visual and spatial
reasoning. We identify multiple settings where GPT-4o using chain-of-thought
fails dramatically, including more than one where it achieves $0\%$ accuracy,
while whiteboard-of-thought enables up to $92\%$ accuracy in these same
settings. We present a detailed exploration of where the technique succeeds as
well as its sources of error. | cs.CL | Whiteboard-of-Thought: Thinking Step-by-Step Across Modalities |
2024-06-20T17:59:42Z | http://arxiv.org/pdf/2406.14561v1 | http://arxiv.org/abs/2406.14561v1 | Tiago Pimentel, Clara Meister | 2024-06-20T17:59:42Z | cs.CL | Language models (LMs) estimate the probability distribution over sequences of
natural language; these distributions are crucial for computing perplexity and
surprisal in linguistics research. While we are usually concerned with
measuring these values for words, most LMs operate over subwords. Despite
seemingly straightforward, accurately computing probabilities over one unit
given probabilities over the other requires care. Indeed, we show here that
many recent linguistic studies have been incorrectly computing these values.
This paper derives the correct methods for computing word probabilities,
highlighting issues when relying on language models that use beginning-of-word
(bow)-marking tokenisers, e.g., the GPT family. Empirically, we show that
correcting the widespread bug in probability computations affects measured
outcomes in sentence comprehension and lexical optimisation analyses. | cs.CL | How to Compute the Probability of a Word |
2024-06-20T17:58:34Z | http://arxiv.org/pdf/2406.14553v1 | http://arxiv.org/abs/2406.14553v1 | Daniil Larionov, Mikhail Seleznyov, Vasiliy Viskov, Alexander Panchenko, Steffen Eger | 2024-06-20T17:58:34Z | cs.CL | State-of-the-art trainable machine translation evaluation metrics like xCOMET
achieve high correlation with human judgment but rely on large encoders (up to
10.7B parameters), making them computationally expensive and inaccessible to
researchers with limited resources. To address this issue, we investigate
whether the knowledge stored in these large encoders can be compressed while
maintaining quality. We employ distillation, quantization, and pruning
techniques to create efficient xCOMET alternatives and introduce a novel data
collection pipeline for efficient black-box distillation. Our experiments show
that, using quantization, xCOMET can be compressed up to three times with no
quality degradation. Additionally, through distillation, we create an
xCOMET-lite metric, which has only 2.6% of xCOMET-XXL parameters, but retains
92.1% of its quality. Besides, it surpasses strong small-scale metrics like
COMET-22 and BLEURT-20 on the WMT22 metrics challenge dataset by 6.4%, despite
using 50% fewer parameters. All code, dataset, and models are available online. | cs.CL | xCOMET-lite: Bridging the Gap Between Efficiency and Quality in Learned MT Evaluation Metrics |
2024-06-20T17:57:51Z | http://arxiv.org/pdf/2406.14550v1 | http://arxiv.org/abs/2406.14550v1 | Shilong Li, Yancheng He, Hangyu Guo, Xingyuan Bu, Ge Bai, Jie Liu, Jiaheng Liu, Xingwei Qu, Yangguang Li, Wanli Ouyang, Wenbo Su, Bo Zheng | 2024-06-20T17:57:51Z | cs.CL, cs.AI | Long-context capabilities are essential for large language models (LLMs) to
tackle complex and long-input tasks. Despite numerous efforts made to optimize
LLMs for long contexts, challenges persist in robustly processing long inputs.
In this paper, we introduce GraphReader, a graph-based agent system designed to
handle long texts by structuring them into a graph and employing an agent to
explore this graph autonomously. Upon receiving a question, the agent first
undertakes a step-by-step analysis and devises a rational plan. It then invokes
a set of predefined functions to read node content and neighbors, facilitating
a coarse-to-fine exploration of the graph. Throughout the exploration, the
agent continuously records new insights and reflects on current circumstances
to optimize the process until it has gathered sufficient information to
generate an answer. Experimental results on the LV-Eval dataset reveal that
GraphReader, using a 4k context window, consistently outperforms GPT-4-128k
across context lengths from 16k to 256k by a large margin. Additionally, our
approach demonstrates superior performance on four challenging single-hop and
multi-hop benchmarks. | cs.CL | GraphReader: Building Graph-based Agent to Enhance Long-Context Abilities of Large Language Models |
2024-06-20T17:55:04Z | http://arxiv.org/pdf/2406.14546v1 | http://arxiv.org/abs/2406.14546v1 | Johannes Treutlein, Dami Choi, Jan Betley, Cem Anil, Samuel Marks, Roger Baker Grosse, Owain Evans | 2024-06-20T17:55:04Z | cs.CL, cs.AI, cs.LG | One way to address safety risks from large language models (LLMs) is to
censor dangerous knowledge from their training data. While this removes the
explicit information, implicit information can remain scattered across various
training documents. Could an LLM infer the censored knowledge by piecing
together these implicit hints? As a step towards answering this question, we
study inductive out-of-context reasoning (OOCR), a type of generalization in
which LLMs infer latent information from evidence distributed across training
documents and apply it to downstream tasks without in-context learning. Using a
suite of five tasks, we demonstrate that frontier LLMs can perform inductive
OOCR. In one experiment we finetune an LLM on a corpus consisting only of
distances between an unknown city and other known cities. Remarkably, without
in-context examples or Chain of Thought, the LLM can verbalize that the unknown
city is Paris and use this fact to answer downstream questions. Further
experiments show that LLMs trained only on individual coin flip outcomes can
verbalize whether the coin is biased, and those trained only on pairs
$(x,f(x))$ can articulate a definition of $f$ and compute inverses. While OOCR
succeeds in a range of cases, we also show that it is unreliable, particularly
for smaller LLMs learning complex structures. Overall, the ability of LLMs to
"connect the dots" without explicit in-context learning poses a potential
obstacle to monitoring and controlling the knowledge acquired by LLMs. | cs.CL | Connecting the Dots: LLMs can Infer and Verbalize Latent Structure from Disparate Training Data |
2024-06-20T17:54:33Z | http://arxiv.org/pdf/2406.14545v1 | http://arxiv.org/abs/2406.14545v1 | Đorđe Klisura, Anthony Rios | 2024-06-20T17:54:33Z | cs.CL | Relational databases are integral to modern information systems, serving as
the foundation for storing, querying, and managing data efficiently and
effectively. Advancements in large language modeling have led to the emergence
of text-to-SQL technologies, significantly enhancing the querying and
extracting of information from these databases and raising concerns about
privacy and security. Our research extracts the database schema elements
underlying a text-to-SQL model. Knowledge of the schema can make attacks such
as SQL injection easier. By asking specially crafted questions, we have
developed a zero-knowledge framework designed to probe various database schema
elements without knowledge of the database itself. The text-to-SQL models then
process these questions to produce an output that we use to uncover the
structure of the database schema. We apply it to specialized text-to-SQL models
fine-tuned on text-SQL pairs and generative language models used for SQL
generation. Overall, we can reconstruct the table names with an F1 of nearly
.75 for fine-tuned models and .96 for generative. | cs.CL | Unmasking Database Vulnerabilities: Zero-Knowledge Schema Inference Attacks in Text-to-SQL Systems |
2024-06-20T17:54:03Z | http://arxiv.org/pdf/2406.14544v1 | http://arxiv.org/abs/2406.14544v1 | Yuxuan Qiao, Haodong Duan, Xinyu Fang, Junming Yang, Lin Chen, Songyang Zhang, Jiaqi Wang, Dahua Lin, Kai Chen | 2024-06-20T17:54:03Z | cs.CV, cs.CL | Vision Language Models (VLMs) demonstrate remarkable proficiency in
addressing a wide array of visual questions, which requires strong perception
and reasoning faculties. Assessing these two competencies independently is
crucial for model refinement, despite the inherent difficulty due to the
intertwined nature of seeing and reasoning in existing VLMs. To tackle this
issue, we present Prism, an innovative framework designed to disentangle the
perception and reasoning processes involved in visual question solving. Prism
comprises two distinct stages: a perception stage that utilizes a VLM to
extract and articulate visual information in textual form, and a reasoning
stage that formulates responses based on the extracted visual information using
a Large Language Model (LLM). This modular design enables the systematic
comparison and assessment of both proprietary and open-source VLM for their
perception and reasoning strengths. Our analytical framework provides several
valuable insights, underscoring Prism's potential as a cost-effective solution
for vision-language tasks. By combining a streamlined VLM focused on perception
with a powerful LLM tailored for reasoning, Prism achieves superior results in
general vision-language tasks while substantially cutting down on training and
operational expenses. Quantitative evaluations show that Prism, when configured
with a vanilla 2B LLaVA and freely accessible GPT-3.5, delivers performance on
par with VLMs $10 \times$ larger on the rigorous multimodal benchmark MMStar.
The project is released at: https://github.com/SparksJoe/Prism. | cs.CV | Prism: A Framework for Decoupling and Assessing the Capabilities of VLMs |
2024-06-20T17:45:54Z | http://arxiv.org/pdf/2406.14532v1 | http://arxiv.org/abs/2406.14532v1 | Amrith Setlur, Saurabh Garg, Xinyang Geng, Naman Garg, Virginia Smith, Aviral Kumar | 2024-06-20T17:45:54Z | cs.LG, cs.CL | Training on model-generated synthetic data is a promising approach for
finetuning LLMs, but it remains unclear when it helps or hurts. In this paper,
we investigate this question for math reasoning via an empirical study,
followed by building a conceptual understanding of our observations. First, we
find that while the typical approach of finetuning a model on synthetic correct
or positive problem-solution pairs generated by capable models offers modest
performance gains, sampling more correct solutions from the finetuned learner
itself followed by subsequent fine-tuning on this self-generated data
$\textbf{doubles}$ the efficiency of the same synthetic problems. At the same
time, training on model-generated positives can amplify various spurious
correlations, resulting in flat or even inverse scaling trends as the amount of
data increases. Surprisingly, we find that several of these issues can be
addressed if we also utilize negative responses, i.e., model-generated
responses that are deemed incorrect by a final answer verifier. Crucially,
these negatives must be constructed such that the training can appropriately
recover the utility or advantage of each intermediate step in the negative
response. With this per-step scheme, we are able to attain consistent gains
over only positive data, attaining performance similar to amplifying the amount
of synthetic data by $\mathbf{8 \times}$. We show that training on per-step
negatives can help to unlearn spurious correlations in the positive data, and
is equivalent to advantage-weighted reinforcement learning (RL), implying that
it inherits robustness benefits of RL over imitating positive data alone. | cs.LG | RL on Incorrect Synthetic Data Scales the Efficiency of LLM Math Reasoning by Eight-Fold |
2024-06-20T17:27:14Z | http://arxiv.org/pdf/2406.14517v1 | http://arxiv.org/abs/2406.14517v1 | Yapei Chang, Kalpesh Krishna, Amir Houmansadr, John Wieting, Mohit Iyyer | 2024-06-20T17:27:14Z | cs.LG, cs.AI, cs.CL, cs.CR | The most effective techniques to detect LLM-generated text rely on inserting
a detectable signature -- or watermark -- during the model's decoding process.
Most existing watermarking methods require access to the underlying LLM's
logits, which LLM API providers are loath to share due to fears of model
distillation. As such, these watermarks must be implemented independently by
each LLM provider. In this paper, we develop PostMark, a modular post-hoc
watermarking procedure in which an input-dependent set of words (determined via
a semantic embedding) is inserted into the text after the decoding process has
completed. Critically, PostMark does not require logit access, which means it
can be implemented by a third party. We also show that PostMark is more robust
to paraphrasing attacks than existing watermarking methods: our experiments
cover eight baseline algorithms, five base LLMs, and three datasets. Finally,
we evaluate the impact of PostMark on text quality using both automated and
human assessments, highlighting the trade-off between quality and robustness to
paraphrasing. We release our code, outputs, and annotations at
https://github.com/lilakk/PostMark. | cs.LG | PostMark: A Robust Blackbox Watermark for Large Language Models |
2024-06-20T17:15:46Z | http://arxiv.org/pdf/2406.14511v1 | http://arxiv.org/abs/2406.14511v1 | Somin Wadhwa, Silvio Amir, Byron C. Wallace | 2024-06-20T17:15:46Z | cs.CL | Eliciting "chain of thought" (CoT) rationales -- sequences of token that
convey a "reasoning" process -- has been shown to consistently improve LLM
performance on tasks like question answering. More recent efforts have shown
that such rationales can also be used for model distillation: Including CoT
sequences (elicited from a large "teacher" model) in addition to target labels
when fine-tuning a small student model yields (often substantial) improvements.
In this work we ask: Why and how does this additional training signal help in
model distillation? We perform ablations to interrogate this, and report some
potentially surprising results. Specifically: (1) Placing CoT sequences after
labels (rather than before) realizes consistently better downstream performance
-- this means that no student "reasoning" is necessary at test time to realize
gains. (2) When rationales are appended in this way, they need not be coherent
reasoning sequences to yield improvements; performance increases are robust to
permutations of CoT tokens, for example. In fact, (3) a small number of key
tokens are sufficient to achieve improvements equivalent to those observed when
full rationales are used in model distillation. | cs.CL | Investigating Mysteries of CoT-Augmented Distillation |
2024-06-20T17:12:38Z | http://arxiv.org/pdf/2406.14508v1 | http://arxiv.org/abs/2406.14508v1 | Kobi Hackenburg, Ben M. Tappin, Paul Röttger, Scott Hale, Jonathan Bright, Helen Margetts | 2024-06-20T17:12:38Z | cs.CL, cs.AI, cs.CY, cs.HC | Large language models can now generate political messages as persuasive as
those written by humans, raising concerns about how far this persuasiveness may
continue to increase with model size. Here, we generate 720 persuasive messages
on 10 U.S. political issues from 24 language models spanning several orders of
magnitude in size. We then deploy these messages in a large-scale randomized
survey experiment (N = 25,982) to estimate the persuasive capability of each
model. Our findings are twofold. First, we find evidence of a log scaling law:
model persuasiveness is characterized by sharply diminishing returns, such that
current frontier models are barely more persuasive than models smaller in size
by an order of magnitude or more. Second, mere task completion (coherence,
staying on topic) appears to account for larger models' persuasive advantage.
These findings suggest that further scaling model size will not much increase
the persuasiveness of static LLM-generated messages. | cs.CL | Evidence of a log scaling law for political persuasion with large language models |
2024-06-20T17:06:58Z | http://arxiv.org/pdf/2406.14504v1 | http://arxiv.org/abs/2406.14504v1 | Pushpdeep Singh, Mayur Patidar, Lovekesh Vig | 2024-06-20T17:06:58Z | cs.CL | LLMs are increasingly being deployed for multilingual applications and have
demonstrated impressive translation capabilities between several low and high
resource languages. An aspect of translation that often gets overlooked is that
of cultural adaptation, or modifying source culture references to suit the
target culture. Cultural adaptation has applications across several creative
industries and requires intimate knowledge of source and target cultures during
translation. While specialized translation models still outperform LLMs on the
machine translation task when viewed from the lens of correctness, they are not
sensitive to cultural differences often requiring manual correction. LLMs on
the other hand have a rich reservoir of cultural knowledge embedded within its
parameters that can be potentially exploited for such applications. In this
paper we define the task of cultural adaptation and create an evaluation
framework to benchmark different models for this task. We evaluate the
performance of modern LLMs for cultural adaptation and analyze their cross
cultural knowledge while connecting related concepts across different cultures.
We also analyze possible issues with automatic adaptation including cultural
biases and stereotypes. We hope that this task will offer more insight into the
cultural understanding of LLMs and their creativity in cross-cultural
scenarios. | cs.CL | Translating Across Cultures: LLMs for Intralingual Cultural Adaptation |
2024-06-20T17:06:13Z | http://arxiv.org/pdf/2406.14503v1 | http://arxiv.org/abs/2406.14503v1 | Jingcong Liang, Junlong Wang, Xinyu Zhai, Yungui Zhuang, Yiyang Zheng, Xin Xu, Xiandong Ran, Xiaozheng Dong, Honghui Rong, Yanlun Liu, Hao Chen, Yuhan Wei, Donghai Li, Jiajie Peng, Xuanjing Huang, Chongde Shi, Yansong Feng, Yun Song, Zhongyu Wei | 2024-06-20T17:06:13Z | cs.CL | We give a detailed overview of the CAIL 2023 Argument Mining Track, one of
the Chinese AI and Law Challenge (CAIL) 2023 tracks. The main goal of the track
is to identify and extract interacting argument pairs in trial dialogs. It
mainly uses summarized judgment documents but can also refer to trial
recordings. The track consists of two stages, and we introduce the tasks
designed for each stage; we also extend the data from previous events into a
new dataset -- CAIL2023-ArgMine -- with annotated new cases from various causes
of action. We outline several submissions that achieve the best results,
including their methods for different stages. While all submissions rely on
language models, they have incorporated strategies that may benefit future work
in this field. | cs.CL | Overview of the CAIL 2023 Argument Mining Track |
2024-06-20T17:01:55Z | http://arxiv.org/pdf/2406.14500v1 | http://arxiv.org/abs/2406.14500v1 | Xingmeng Zhao, Tongnian Wang, Anthony Rios | 2024-06-20T17:01:55Z | cs.CL | Radiology report summarization (RRS) is crucial for patient care, requiring
concise "Impressions" from detailed "Findings." This paper introduces a novel
prompting strategy to enhance RRS by first generating a layperson summary. This
approach normalizes key observations and simplifies complex information using
non-expert communication techniques inspired by doctor-patient interactions.
Combined with few-shot in-context learning, this method improves the model's
ability to link general terms to specific findings. We evaluate this approach
on the MIMIC-CXR, CheXpert, and MIMIC-III datasets, benchmarking it against
7B/8B parameter state-of-the-art open-source large language models (LLMs) like
Meta-Llama-3-8B-Instruct. Our results demonstrate improvements in summarization
accuracy and accessibility, particularly in out-of-domain tests, with
improvements as high as 5% for some metrics. | cs.CL | Improving Expert Radiology Report Summarization by Prompting Large Language Models with a Layperson Summary |
2024-06-20T17:00:34Z | http://arxiv.org/pdf/2406.14498v1 | http://arxiv.org/abs/2406.14498v1 | Sheikh Asif Imran, Mohammad Nur Hossain Khan, Subrata Biswas, Bashima Islam | 2024-06-20T17:00:34Z | cs.CL | Integrating inertial measurement units (IMUs) with large language models
(LLMs) advances multimodal AI by enhancing human activity understanding. We
introduce SensorCaps, a dataset of 26,288 IMU-derived activity narrations, and
OpenSQA, an instruction-following dataset with 257,562 question-answer pairs.
Combining LIMU-BERT and Llama, we develop LLaSA, a Large Multimodal Agent
capable of interpreting and responding to activity and motion analysis queries.
Our evaluation demonstrates LLaSA's effectiveness in activity classification
and question answering, highlighting its potential in healthcare, sports
science, and human-computer interaction. These contributions advance
sensor-aware language models and open new research avenues. Our code repository
and datasets can be found on https://github.com/BASHLab/LLaSA. | cs.CL | LLaSA: Large Multimodal Agent for Human Activity Analysis Through Wearable Sensors |
2024-06-20T16:59:52Z | http://arxiv.org/pdf/2406.14497v1 | http://arxiv.org/abs/2406.14497v1 | Zora Zhiruo Wang, Akari Asai, Xinyan Velocity Yu, Frank F. Xu, Yiqing Xie, Graham Neubig, Daniel Fried | 2024-06-20T16:59:52Z | cs.SE, cs.CL | While language models (LMs) have proven remarkably adept at generating code,
many programs are challenging for LMs to generate using their parametric
knowledge alone. Providing external contexts such as library documentation can
facilitate generating accurate and functional code. Despite the success of
retrieval-augmented generation (RAG) in various text-oriented tasks, its
potential for improving code generation remains under-explored. In this work,
we conduct a systematic, large-scale analysis by asking: in what scenarios can
retrieval benefit code generation models? and what challenges remain? We first
curate a comprehensive evaluation benchmark, CodeRAG-Bench, encompassing three
categories of code generation tasks, including basic programming, open-domain,
and repository-level problems. We aggregate documents from five sources for
models to retrieve contexts: competition solutions, online tutorials, library
documentation, StackOverflow posts, and GitHub repositories. We examine
top-performing models on CodeRAG-Bench by providing contexts retrieved from one
or multiple sources. While notable gains are made in final code generation by
retrieving high-quality contexts across various settings, our analysis reveals
room for improvement -- current retrievers still struggle to fetch useful
contexts especially with limited lexical overlap, and generators fail to
improve with limited context lengths or abilities to integrate additional
contexts. We hope CodeRAG-Bench serves as an effective testbed to encourage
further development of advanced code-oriented RAG methods. | cs.SE | CodeRAG-Bench: Can Retrieval Augment Code Generation? |
2024-06-20T16:59:39Z | http://arxiv.org/pdf/2406.14496v1 | http://arxiv.org/abs/2406.14496v1 | Gregor Geigle, Radu Timofte, Goran Glavaš | 2024-06-20T16:59:39Z | cs.CV, cs.CL | Recent Large Vision-Language Models (LVLMs) demonstrate impressive abilities
on numerous image understanding and reasoning tasks. The task of fine-grained
object classification (e.g., distinction between \textit{animal species}),
however, has been probed insufficiently, despite its downstream importance. We
fill this evaluation gap by creating \texttt{FOCI} (\textbf{F}ine-grained
\textbf{O}bject \textbf{C}lass\textbf{I}fication), a difficult multiple-choice
benchmark for fine-grained object classification, from existing object
classification datasets: (1) multiple-choice avoids ambiguous answers
associated with casting classification as open-ended QA task; (2) we retain
classification difficulty by mining negative labels with a CLIP model.
\texttt{FOCI}\xspace complements five popular classification datasets with four
domain-specific subsets from ImageNet-21k. We benchmark 12 public LVLMs on
\texttt{FOCI} and show that it tests for a \textit{complementary skill} to
established image understanding and reasoning benchmarks. Crucially, CLIP
models exhibit dramatically better performance than LVLMs. Since the image
encoders of LVLMs come from these CLIP models, this points to inadequate
alignment for fine-grained object distinction between the encoder and the LLM
and warrants (pre)training data with more fine-grained annotation. We release
our code at \url{https://github.com/gregor-ge/FOCI-Benchmark}. | cs.CV | African or European Swallow? Benchmarking Large Vision-Language Models for Fine-Grained Object Classification |
2024-06-20T16:56:11Z | http://arxiv.org/pdf/2406.14492v1 | http://arxiv.org/abs/2406.14492v1 | Gregor Geigle, Radu Timofte, Goran Glavaš | 2024-06-20T16:56:11Z | cs.CV, cs.CL | Large vision-language models (LVLMs) have recently dramatically pushed the
state of the art in image captioning and many image understanding tasks (e.g.,
visual question answering). LVLMs, however, often \textit{hallucinate} and
produce captions that mention concepts that cannot be found in the image. These
hallucinations erode the trustworthiness of LVLMs and are arguably among the
main obstacles to their ubiquitous adoption. Recent work suggests that addition
of grounding objectives -- those that explicitly align image regions or objects
to text spans -- reduces the amount of LVLM hallucination. Although intuitive,
this claim is not empirically justified as the reduction effects have been
established, we argue, with flawed evaluation protocols that (i) rely on data
(i.e., MSCOCO) that has been extensively used in LVLM training and (ii) measure
hallucination via question answering rather than open-ended caption generation.
In this work, in contrast, we offer the first systematic analysis of the effect
of fine-grained object grounding on LVLM hallucination under an evaluation
protocol that more realistically captures LVLM hallucination in open
generation. Our extensive experiments over three backbone LLMs reveal that
grounding objectives have little to no effect on object hallucination in open
caption generation. | cs.CV | Does Object Grounding Really Reduce Hallucination of Large Vision-Language Models? |
2024-06-20T16:55:33Z | http://arxiv.org/pdf/2406.14491v1 | http://arxiv.org/abs/2406.14491v1 | Daixuan Cheng, Yuxian Gu, Shaohan Huang, Junyu Bi, Minlie Huang, Furu Wei | 2024-06-20T16:55:33Z | cs.CL | Unsupervised multitask pre-training has been the critical method behind the
recent success of language models (LMs). However, supervised multitask learning
still holds significant promise, as scaling it in the post-training stage
trends towards better generalization. In this paper, we explore supervised
multitask pre-training by proposing Instruction Pre-Training, a framework that
scalably augments massive raw corpora with instruction-response pairs to
pre-train LMs. The instruction-response pairs are generated by an efficient
instruction synthesizer built on open-source models. In our experiments, we
synthesize 200M instruction-response pairs covering 40+ task categories to
verify the effectiveness of Instruction Pre-Training. In pre-training from
scratch, Instruction Pre-Training not only consistently enhances pre-trained
base models but also benefits more from further instruction tuning. In
continual pre-training, Instruction Pre-Training enables Llama3-8B to be
comparable to or even outperform Llama3-70B. Our model, code, and data are
available at https://github.com/microsoft/LMOps. | cs.CL | Instruction Pre-Training: Language Models are Supervised Multitask Learners |
2024-06-20T16:41:09Z | http://arxiv.org/pdf/2406.14479v1 | http://arxiv.org/abs/2406.14479v1 | Jiachen Jiang, Jinxin Zhou, Zhihui Zhu | 2024-06-20T16:41:09Z | cs.AI, cs.CL, cs.CV, cs.LG | Analyzing the similarity of internal representations within and across
different models has been an important technique for understanding the behavior
of deep neural networks. Most existing methods for analyzing the similarity
between representations of high dimensions, such as those based on Canonical
Correlation Analysis (CCA) and widely used Centered Kernel Alignment (CKA),
rely on statistical properties of the representations for a set of data points.
In this paper, we focus on transformer models and study the similarity of
representations between the hidden layers of individual transformers. In this
context, we show that a simple sample-wise cosine similarity metric is capable
of capturing the similarity and aligns with the complicated CKA. Our
experimental results on common transformers reveal that representations across
layers are positively correlated, albeit the similarity decreases when layers
are far apart. We then propose an aligned training approach to enhance the
similarity between internal representations, with trained models that enjoy the
following properties: (1) the last-layer classifier can be directly applied
right after any hidden layers, yielding intermediate layer accuracies much
higher than those under standard training, (2) the layer-wise accuracies
monotonically increase and reveal the minimal depth needed for the given task,
(3) when served as multi-exit models, they achieve on-par performance with
standard multi-exit architectures which consist of additional classifiers
designed for early exiting in shallow layers. To our knowledge, our work is the
first to show that one common classifier is sufficient for multi-exit models.
We conduct experiments on both vision and NLP tasks to demonstrate the
performance of the proposed aligned training. | cs.AI | On Layer-wise Representation Similarity: Application for Multi-Exit Models with a Single Classifier |
2024-06-20T16:34:07Z | http://arxiv.org/pdf/2406.14473v1 | http://arxiv.org/abs/2406.14473v1 | Xinyi Xu, Zhaoxuan Wu, Rui Qiao, Arun Verma, Yao Shu, Jingtan Wang, Xinyuan Niu, Zhenfeng He, Jiangwei Chen, Zijian Zhou, Gregory Kang Ruey Lau, Hieu Dao, Lucas Agussurja, Rachael Hwee Ling Sim, Xiaoqiang Lin, Wenyang Hu, Zhongxiang Dai, Pang Wei Koh, Bryan Kian Hsiang Low | 2024-06-20T16:34:07Z | cs.LG, cs.CL | This position paper proposes a data-centric viewpoint of AI research,
focusing on large language models (LLMs). We start by making the key
observation that data is instrumental in the developmental (e.g., pretraining
and fine-tuning) and inferential stages (e.g., in-context learning) of LLMs,
and yet it receives disproportionally low attention from the research
community. We identify four specific scenarios centered around data, covering
data-centric benchmarks and data curation, data attribution, knowledge
transfer, and inference contextualization. In each scenario, we underscore the
importance of data, highlight promising research directions, and articulate the
potential impacts on the research community and, where applicable, the society
as a whole. For instance, we advocate for a suite of data-centric benchmarks
tailored to the scale and complexity of data for LLMs. These benchmarks can be
used to develop new data curation methods and document research efforts and
results, which can help promote openness and transparency in AI and LLM
research. | cs.LG | Data-Centric AI in the Age of Large Language Models |
2024-06-20T16:26:03Z | http://arxiv.org/pdf/2406.14464v1 | http://arxiv.org/abs/2406.14464v1 | Roman Aperdannier, Sigurd Schacht, Alexander Piazza | 2024-06-20T16:26:03Z | cs.SD, cs.CL, eess.AS | Speaker diarization provides the answer to the question "who spoke when?" for
an audio file. This information can be used to complete audio transcripts for
further processing steps. Most speaker diarization systems assume that the
audio file is available as a whole. However, there are scenarios in which the
speaker labels are needed immediately after the arrival of an audio segment.
Speaker diarization with a correspondingly low latency is referred to as online
speaker diarization. This paper provides an overview. First the history of
online speaker diarization is briefly presented. Next a taxonomy and datasets
for training and evaluation are given. In the sections that follow, online
diarization methods and systems are discussed in detail. This paper concludes
with the presentation of challenges that still need to be solved by future
research in the field of online speaker diarization. | cs.SD | A Review of Common Online Speaker Diarization Methods |
2024-06-20T16:24:07Z | http://arxiv.org/pdf/2406.14462v1 | http://arxiv.org/abs/2406.14462v1 | Salvatore Giorgi, Tingting Liu, Ankit Aich, Kelsey Isman, Garrick Sherman, Zachary Fried, João Sedoc, Lyle H. Ungar, Brenda Curtis | 2024-06-20T16:24:07Z | cs.CL | Large language models (LLMs) are increasingly being used in human-centered
social scientific tasks, such as data annotation, synthetic data creation, and
engaging in dialog. However, these tasks are highly subjective and dependent on
human factors, such as one's environment, attitudes, beliefs, and lived
experiences. Thus, employing LLMs (which do not have such human factors) in
these tasks may result in a lack of variation in data, failing to reflect the
diversity of human experiences. In this paper, we examine the role of prompting
LLMs with human-like personas and asking the models to answer as if they were a
specific human. This is done explicitly, with exact demographics, political
beliefs, and lived experiences, or implicitly via names prevalent in specific
populations. The LLM personas are then evaluated via (1) subjective annotation
task (e.g., detecting toxicity) and (2) a belief generation task, where both
tasks are known to vary across human factors. We examine the impact of explicit
vs. implicit personas and investigate which human factors LLMs recognize and
respond to. Results show that LLM personas show mixed results when reproducing
known human biases, but generate generally fail to demonstrate implicit biases.
We conclude that LLMs lack the intrinsic cognitive mechanisms of human thought,
while capturing the statistical patterns of how people speak, which may
restrict their effectiveness in complex social science applications. | cs.CL | Explicit and Implicit Large Language Model Personas Generate Opinions but Fail to Replicate Deeper Perceptions and Biases |
2024-06-20T16:18:04Z | http://arxiv.org/pdf/2406.14459v1 | http://arxiv.org/abs/2406.14459v1 | Shijie Han, Zhenyu Zhang, Andrei Arsene Simion | 2024-06-20T16:18:04Z | cs.CL | Language models like BERT excel at sentence classification tasks due to
extensive pre-training on general data, but their robustness to parameter
corruption is unexplored. To understand this better, we look at what happens if
a language model is "broken", in the sense that some of its parameters are
corrupted and then recovered by fine-tuning. Strategically corrupting BERT
variants at different levels, we find corrupted models struggle to fully
recover their original performance, with higher corruption causing more severe
degradation. Notably, bottom-layer corruption affecting fundamental linguistic
features is more detrimental than top-layer corruption. Our insights contribute
to understanding language model robustness and adaptability under adverse
conditions, informing strategies for developing resilient NLP systems against
parameter perturbations. | cs.CL | Healing Powers of BERT: How Task-Specific Fine-Tuning Recovers Corrupted Language Models |
2024-06-20T15:59:07Z | http://arxiv.org/pdf/2406.14434v1 | http://arxiv.org/abs/2406.14434v1 | Weihao Liu, Ning Wu, Wenbiao Ding, Shining Liang, Ming Gong, Dongmei Zhang | 2024-06-20T15:59:07Z | cs.CL | In the era of large language models (LLMs), building multilingual large
language models (MLLMs) that can serve users worldwide holds great
significance. However, existing research seldom focuses on the truthfulness of
MLLMs. Meanwhile, contemporary multilingual aligning technologies struggle to
balance massive languages and often exhibit serious truthfulness gaps across
different languages, especially those that differ greatly from English. In our
work, we construct a benchmark for truthfulness evaluation in multilingual
scenarios and explore the ways to align facts across languages to enhance the
truthfulness of MLLMs. Furthermore, we propose Fact-aware Multilingual
Selective Synergy (FaMSS) to optimize the data allocation across a large number
of languages and different data types. Experimental results demonstrate that
our approach can effectively reduce the multilingual representation disparity
and enhance the multilingual capabilities of LLMs. | cs.CL | Towards Truthful Multilingual Large Language Models: Benchmarking and Alignment Strategies |
2024-06-20T15:49:28Z | http://arxiv.org/pdf/2406.14425v2 | http://arxiv.org/abs/2406.14425v2 | Gayane Ghazaryan, Erik Arakelyan, Pasquale Minervini, Isabelle Augenstein | 2024-06-25T13:48:41Z | cs.CL, cs.AI, cs.LG | Question Answering (QA) datasets have been instrumental in developing and
evaluating Large Language Model (LLM) capabilities. However, such datasets are
scarce for languages other than English due to the cost and difficulties of
collection and manual annotation. This means that producing novel models and
measuring the performance of multilingual LLMs in low-resource languages is
challenging. To mitigate this, we propose $\textbf{S}$yn$\textbf{DAR}$in, a
method for generating and validating QA datasets for low-resource languages. We
utilize parallel content mining to obtain $\textit{human-curated}$ paragraphs
between English and the target language. We use the English data as context to
$\textit{generate}$ synthetic multiple-choice (MC) question-answer pairs, which
are automatically translated and further validated for quality. Combining these
with their designated non-English $\textit{human-curated}$ paragraphs form the
final QA dataset. The method allows to maintain the content quality, reduces
the likelihood of factual errors, and circumvents the need for costly
annotation. To test the method, we created a QA dataset with $1.2$K samples for
the Armenian language. The human evaluation shows that $98\%$ of the generated
English data maintains quality and diversity in the question types and topics,
while the translation validation pipeline can filter out $\sim70\%$ of data
with poor quality. We use the dataset to benchmark state-of-the-art LLMs,
showing their inability to achieve human accuracy with some model performances
closer to random chance. This shows that the generated dataset is non-trivial
and can be used to evaluate reasoning capabilities in low-resource language. | cs.CL | SynDARin: Synthesising Datasets for Automated Reasoning in Low-Resource Languages |
2024-06-20T15:31:05Z | http://arxiv.org/pdf/2406.14408v2 | http://arxiv.org/abs/2406.14408v2 | Xiaohan Lin, Qingxing Cao, Yinya Huang, Haiming Wang, Jianqiao Lu, Zhengying Liu, Linqi Song, Xiaodan Liang | 2024-06-21T02:51:41Z | cs.AI, cs.CL, cs.LG | Formal verification (FV) has witnessed growing significance with current
emerging program synthesis by the evolving large language models (LLMs).
However, current formal verification mainly resorts to symbolic verifiers or
hand-craft rules, resulting in limitations for extensive and flexible
verification. On the other hand, formal languages for automated theorem
proving, such as Isabelle, as another line of rigorous verification, are
maintained with comprehensive rules and theorems. In this paper, we propose
FVEL, an interactive Formal Verification Environment with LLMs. Specifically,
FVEL transforms a given code to be verified into Isabelle, and then conducts
verification via neural automated theorem proving with an LLM. The joined
paradigm leverages the rigorous yet abundant formulated and organized rules in
Isabelle and is also convenient for introducing and adjusting cutting-edge
LLMs. To achieve this goal, we extract a large-scale FVELER3. The FVELER
dataset includes code dependencies and verification processes that are
formulated in Isabelle, containing 758 theories, 29,125 lemmas, and 200,646
proof steps in total with in-depth dependencies. We benchmark FVELER in the
FVEL environment by first fine-tuning LLMs with FVELER and then evaluating them
on Code2Inv and SV-COMP. The results show that FVEL with FVELER fine-tuned
Llama3- 8B solves 17.39% (69 -> 81) more problems, and Mistral-7B 12% (75 ->
84) more problems in SV-COMP. And the proportion of proof errors is reduced.
Project page: https://fveler.github.io/. | cs.AI | FVEL: Interactive Formal Verification Environment with Large Language Models via Theorem Proving |
2024-06-20T15:12:41Z | http://arxiv.org/pdf/2406.14394v1 | http://arxiv.org/abs/2406.14394v1 | Viet Dac Lai, Michael Krumdick, Charles Lovering, Varshini Reddy, Craig Schmidt, Chris Tanner | 2024-06-20T15:12:41Z | cs.CL | The financial domain frequently deals with large numbers of long documents
that are essential for daily operations. Significant effort is put towards
automating financial data analysis. However, a persistent challenge, not
limited to the finance domain, is the scarcity of datasets that accurately
reflect real-world tasks for model evaluation. Existing datasets are often
constrained by size, context, or relevance to practical applications. Moreover,
LLMs are currently trained on trillions of tokens of text, limiting access to
novel data or documents that models have not encountered during training for
unbiased evaluation. We propose SEC-QA, a continuous dataset generation
framework with two key features: 1) the semi-automatic generation of
Question-Answer (QA) pairs spanning multiple long context financial documents,
which better represent real-world financial scenarios; 2) the ability to
continually refresh the dataset using the most recent public document
collections, not yet ingested by LLMs. Our experiments show that current
retrieval augmented generation methods systematically fail to answer these
challenging multi-document questions. In response, we introduce a QA system
based on program-of-thought that improves the ability to perform complex
information retrieval and quantitative reasoning pipelines, thereby increasing
QA accuracy. | cs.CL | SEC-QA: A Systematic Evaluation Corpus for Financial QA |
2024-06-20T15:12:27Z | http://arxiv.org/pdf/2406.14393v1 | http://arxiv.org/abs/2406.14393v1 | Zhihui Xie, Jiahui Gao, Lei Li, Zhenguo Li, Qi Liu, Lingpeng Kong | 2024-06-20T15:12:27Z | cs.LG, cs.CL | The widespread adoption of large language models (LLMs) has raised concerns
about their safety and reliability, particularly regarding their vulnerability
to adversarial attacks. In this paper, we propose a novel perspective that
attributes this vulnerability to reward misspecification during the alignment
process. We introduce a metric ReGap to quantify the extent of reward
misspecification and demonstrate its effectiveness and robustness in detecting
harmful backdoor prompts. Building upon these insights, we present ReMiss, a
system for automated red teaming that generates adversarial prompts against
various target aligned LLMs. ReMiss achieves state-of-the-art attack success
rates on the AdvBench benchmark while preserving the human readability of the
generated prompts. Detailed analysis highlights the unique advantages brought
by the proposed reward misspecification objective compared to previous methods. | cs.LG | Jailbreaking as a Reward Misspecification Problem |
2024-06-20T14:42:58Z | http://arxiv.org/pdf/2406.14373v2 | http://arxiv.org/abs/2406.14373v2 | Gordon Dai, Weijia Zhang, Jinhan Li, Siqi Yang, Chidera Onochie lbe, Srihas Rao, Arthur Caetano, Misha Sra | 2024-07-01T22:06:13Z | cs.AI, cs.CL, cs.CY, cs.HC, cs.MA | The emergence of Large Language Models (LLMs) and advancements in Artificial
Intelligence (AI) offer an opportunity for computational social science
research at scale. Building upon prior explorations of LLM agent design, our
work introduces a simulated agent society where complex social relationships
dynamically form and evolve over time. Agents are imbued with psychological
drives and placed in a sandbox survival environment. We conduct an evaluation
of the agent society through the lens of Thomas Hobbes's seminal Social
Contract Theory (SCT). We analyze whether, as the theory postulates, agents
seek to escape a brutish "state of nature" by surrendering rights to an
absolute sovereign in exchange for order and security. Our experiments unveil
an alignment: Initially, agents engage in unrestrained conflict, mirroring
Hobbes's depiction of the state of nature. However, as the simulation
progresses, social contracts emerge, leading to the authorization of an
absolute sovereign and the establishment of a peaceful commonwealth founded on
mutual cooperation. This congruence between our LLM agent society's
evolutionary trajectory and Hobbes's theoretical account indicates LLMs'
capability to model intricate social dynamics and potentially replicate forces
that shape human societies. By enabling such insights into group behavior and
emergent societal phenomena, LLM-driven multi-agent simulations, while unable
to simulate all the nuances of human behavior, may hold potential for advancing
our understanding of social structures, group dynamics, and complex human
systems. | cs.AI | Artificial Leviathan: Exploring Social Evolution of LLM Agents Through the Lens of Hobbesian Social Contract Theory |
2024-06-20T14:36:15Z | http://arxiv.org/pdf/2406.16953v1 | http://arxiv.org/abs/2406.16953v1 | Justine Frija, Juliette Millet, Emilie Bequignon, Ala Covali, Guillaume Cathelain, Josselin Houenou, Helene Benzaquen, Pierre Alexis Geoffroy, Emmanuel Bacry, Mathieu Grajoszex, Marie-Pia d Ortho | 2024-06-20T14:36:15Z | eess.SP, cs.CL | Obstructive sleep apnea (OSA) is frequent and responsible for cardiovascular
complications and excessive daytime sleepiness. It is underdiagnosed due to the
difficulty to access the gold standard for diagnosis, polysomnography (PSG).
Alternative methods using smartphone sensors could be useful to increase
diagnosis. The objective is to assess the performances of Apneal, an
application that records the sound using a smartphone's microphone and
movements thanks to a smartphone's accelerometer and gyroscope, to estimate
patients' AHI. In this article, we perform a monocentric proof-of-concept study
with a first manual scoring step, and then an automatic detection of
respiratory events from the recorded signals using a sequential deep-learning
model which was released internally at Apneal at the end of 2022 (version 0.1
of Apneal automatic scoring of respiratory events), in adult patients during
in-hospital polysomnography.46 patients (women 34 per cent, mean BMI 28.7 kg
per m2) were included. For AHI superior to 15, sensitivity of manual scoring
was 0.91, and positive predictive value (PPV) 0.89. For AHI superior to 30,
sensitivity was 0.85, PPV 0.94. We obtained an AUC-ROC of 0.85 and an AUC-PR of
0.94 for the identification of AHI superior to 15, and AUC-ROC of 0.95 and
AUC-PR of 0.93 for AHI superior to 30. Promising results are obtained for the
automatic annotations of events.This article shows that manual scoring of
smartphone-based signals is possible and accurate compared to PSG-based
scorings. Automatic scoring method based on a deep learning model provides
promising results. A larger multicentric validation study, involving subjects
with different SAHS severity is required to confirm these results. | eess.SP | Validation of a new, minimally-invasive, software smartphone device to predict sleep apnea and its severity: transversal study |
2024-06-20T14:31:09Z | http://arxiv.org/pdf/2406.14358v1 | http://arxiv.org/abs/2406.14358v1 | Yuannan Li, Shan Xu, Jia Liu | 2024-06-20T14:31:09Z | q-bio.NC, cs.AI, cs.CL | The ability to manipulate logical-mathematical symbols (LMS), encompassing
tasks such as calculation, reasoning, and programming, is a cognitive skill
arguably unique to humans. Considering the relatively recent emergence of this
ability in human evolutionary history, it has been suggested that LMS
processing may build upon more fundamental cognitive systems, possibly through
neuronal recycling. Previous studies have pinpointed two primary candidates,
natural language processing and spatial cognition. Existing comparisons between
these domains largely relied on task-level comparison, which may be confounded
by task idiosyncrasy. The present study instead compared the neural correlates
at the domain level with both automated meta-analysis and synthesized maps
based on three representative LMS tasks, reasoning, calculation, and mental
programming. Our results revealed a more substantial cortical overlap between
LMS processing and spatial cognition, in contrast to language processing.
Furthermore, in regions activated by both spatial and language processing, the
multivariate activation pattern for LMS processing exhibited greater
multivariate similarity to spatial cognition than to language processing. A
hierarchical clustering analysis further indicated that typical LMS tasks were
indistinguishable from spatial cognition tasks at the neural level, suggesting
an inherent connection between these two cognitive processes. Taken together,
our findings support the hypothesis that spatial cognition is likely the basis
of LMS processing, which may shed light on the limitations of large language
models in logical reasoning, particularly those trained exclusively on textual
data without explicit emphasis on spatial content. | q-bio.NC | The neural correlates of logical-mathematical symbol systems processing resemble that of spatial cognition more than natural language processing |
2024-06-20T14:04:59Z | http://arxiv.org/pdf/2406.14336v1 | http://arxiv.org/abs/2406.14336v1 | Erum Haris, Anthony G. Cohn, John G. Stell | 2024-06-20T14:04:59Z | cs.CL, cs.AI | Navigating historical narratives poses a challenge in unveiling the spatial
intricacies of past landscapes. The proposed work addresses this challenge
within the context of the English Lake District, employing the Corpus of the
Lake District Writing. The method utilizes a generative pre-trained transformer
model to extract spatial relations from the textual descriptions in the corpus.
The study applies this large language model to understand the spatial
dimensions inherent in historical narratives comprehensively. The outcomes are
presented as semantic triples, capturing the nuanced connections between
entities and locations, and visualized as a network, offering a graphical
representation of the spatial narrative. The study contributes to a deeper
comprehension of the English Lake District's spatial tapestry and provides an
approach to uncovering spatial relations within diverse historical contexts. | cs.CL | Exploring Spatial Representations in the Historical Lake District Texts with LLM-based Relation Extraction |
2024-06-20T14:04:53Z | http://arxiv.org/pdf/2406.14335v1 | http://arxiv.org/abs/2406.14335v1 | Francesco De Santis, Philippe Bich, Gabriele Ciravegna, Pietro Barbiero, Danilo Giordano, Tania Cerquitelli | 2024-06-20T14:04:53Z | cs.CL, cs.AI | Despite their success, Large-Language Models (LLMs) still face criticism as
their lack of interpretability limits their controllability and reliability.
Traditional post-hoc interpretation methods, based on attention and
gradient-based analysis, offer limited insight into the model's decision-making
processes. In the image field, Concept-based models have emerged as
explainable-by-design architectures, employing human-interpretable features as
intermediate representations. However, these methods have not been yet adapted
to textual data, mainly because they require expensive concept annotations,
which are impractical for real-world text data. This paper addresses this
challenge by proposing a self-supervised Interpretable Concept Embedding Models
(ICEMs). We leverage the generalization abilities of LLMs to predict the
concepts labels in a self-supervised way, while we deliver the final
predictions with an interpretable function. The results of our experiments show
that ICEMs can be trained in a self-supervised way achieving similar
performance to fully supervised concept-based models and end-to-end black-box
ones. Additionally, we show that our models are (i) interpretable, offering
meaningful logical explanations for their predictions; (ii) interactable,
allowing humans to modify intermediate predictions through concept
interventions; and (iii) controllable, guiding the LLMs' decoding process to
follow a required decision-making path. | cs.CL | Self-supervised Interpretable Concept-based Models for Text Classification |
2024-06-20T13:56:52Z | http://arxiv.org/pdf/2406.14326v1 | http://arxiv.org/abs/2406.14326v1 | Mingyi Jia, Junwen Duan, Yan Song, Jianxin Wang | 2024-06-20T13:56:52Z | cs.CL | Electronic Medical Records (EMRs), while integral to modern healthcare,
present challenges for clinical reasoning and diagnosis due to their complexity
and information redundancy. To address this, we proposed medIKAL (Integrating
Knowledge Graphs as Assistants of LLMs), a framework that combines Large
Language Models (LLMs) with knowledge graphs (KGs) to enhance diagnostic
capabilities. medIKAL assigns weighted importance to entities in medical
records based on their type, enabling precise localization of candidate
diseases within KGs. It innovatively employs a residual network-like approach,
allowing initial diagnosis by the LLM to be merged into KG search results.
Through a path-based reranking algorithm and a fill-in-the-blank style prompt
template, it further refined the diagnostic process. We validated medIKAL's
effectiveness through extensive experiments on a newly introduced open-sourced
Chinese EMR dataset, demonstrating its potential to improve clinical diagnosis
in real-world settings. | cs.CL | medIKAL: Integrating Knowledge Graphs as Assistants of LLMs for Enhanced Clinical Diagnosis on EMRs |
2024-06-20T13:54:32Z | http://arxiv.org/pdf/2406.14322v2 | http://arxiv.org/abs/2406.14322v2 | Lynn Chua, Badih Ghazi, Yangsibo Huang, Pritish Kamath, Ravi Kumar, Daogao Liu, Pasin Manurangsi, Amer Sinha, Chiyuan Zhang | 2024-07-03T14:05:20Z | cs.CL, cs.CR, cs.LG | Large language models (LLMs) have emerged as powerful tools for tackling
complex tasks across diverse domains, but they also raise privacy concerns when
fine-tuned on sensitive data due to potential memorization. While differential
privacy (DP) offers a promising solution by ensuring models are 'almost
indistinguishable' with or without any particular privacy unit, current
evaluations on LLMs mostly treat each example (text record) as the privacy
unit. This leads to uneven user privacy guarantees when contributions per user
vary. We therefore study user-level DP motivated by applications where it
necessary to ensure uniform privacy protection across users. We present a
systematic evaluation of user-level DP for LLM fine-tuning on natural language
generation tasks. Focusing on two mechanisms for achieving user-level DP
guarantees, Group Privacy and User-wise DP-SGD, we investigate design choices
like data selection strategies and parameter tuning for the best
privacy-utility tradeoff. | cs.CL | Mind the Privacy Unit! User-Level Differential Privacy for Language Model Fine-Tuning |
2024-06-20T13:52:30Z | http://arxiv.org/pdf/2406.14319v1 | http://arxiv.org/abs/2406.14319v1 | Chuangtao Chen, Grace Li Zhang, Xunzhao Yin, Cheng Zhuo, Ulf Schlichtmann, Bing Li | 2024-06-20T13:52:30Z | cs.AI, cs.CL | In this paper, we introduce a novel low-latency inference framework for large
language models (LLMs) inference which enables LLMs to perform inferences with
incomplete prompts. By reallocating computational processes to prompt input
phase, we achieve a substantial reduction in latency, thereby significantly
enhancing the interactive experience for users of LLMs. The framework adeptly
manages the visibility of the streaming prompt to the model, allowing it to
infer from incomplete prompts or await additional prompts. Compared with
traditional inference methods that utilize complete prompts, our approach
demonstrates an average reduction of 59% in response latency on the MMLU-Pro
dataset, while maintaining comparable accuracy. Additionally, our framework
facilitates collaborative inference and output across different models. By
employing an LLM for inference and a small language model (SLM) for output, we
achieve an average 68% reduction in response latency, alongside a 5.5%
improvement in accuracy on the MMLU-Pro dataset compared with the SLM baseline.
For long prompts exceeding 20 sentences, the response latency can be reduced by
up to 93%. | cs.AI | LiveMind: Low-latency Large Language Models with Simultaneous Inference |
2024-06-20T13:52:25Z | http://arxiv.org/pdf/2406.14318v1 | http://arxiv.org/abs/2406.14318v1 | Zhili Shen, Zihang Xi, Ying He, Wei Tong, Jingyu Hua, Sheng Zhong | 2024-06-20T13:52:25Z | cs.CR, cs.AI, cs.CL | The rapid adoption of online chatbots represents a significant advancement in
artificial intelligence. However, this convenience brings considerable privacy
concerns, as prompts can inadvertently contain sensitive information exposed to
large language models (LLMs). Limited by high computational costs, reduced task
usability, and excessive system modifications, previous works based on local
deployment, embedding perturbation, and homomorphic encryption are inapplicable
to online prompt-based LLM applications.
To address these issues, this paper introduces Prompt Privacy Sanitizer
(i.e., ProSan), an end-to-end prompt privacy protection framework that can
produce anonymized prompts with contextual privacy removed while maintaining
task usability and human readability. It can also be seamlessly integrated into
the online LLM service pipeline. To achieve high usability and dynamic
anonymity, ProSan flexibly adjusts its protection targets and strength based on
the importance of the words and the privacy leakage risk of the prompts.
Additionally, ProSan is capable of adapting to diverse computational resource
conditions, ensuring privacy protection even for mobile devices with limited
computing power. Our experiments demonstrate that ProSan effectively removes
private information across various tasks, including question answering, text
summarization, and code generation, with minimal reduction in task performance. | cs.CR | The Fire Thief Is Also the Keeper: Balancing Usability and Privacy in Prompts |
2024-06-20T13:46:10Z | http://arxiv.org/pdf/2406.14314v2 | http://arxiv.org/abs/2406.14314v2 | Omri Berkovitch, Sapir Caduri, Noam Kahlon, Anatoly Efros, Avi Caciularu, Ido Dagan | 2024-06-30T12:33:48Z | cs.CL, cs.AI | Autonomous agents that interact with graphical user interfaces (GUIs) hold
significant potential for enhancing user experiences. To further improve these
experiences, agents need to be personalized and proactive. By effectively
comprehending user intentions through their actions and interactions with GUIs,
agents will be better positioned to achieve these goals. This paper introduces
the task of goal identification from observed UI trajectories, aiming to infer
the user's intended task based on their GUI interactions. We propose a novel
evaluation metric to assess whether two task descriptions are paraphrases
within a specific UI environment. By Leveraging the inverse relation with the
UI automation task, we utilized the Android-In-The-Wild and Mind2Web datasets
for our experiments. Using our metric and these datasets, we conducted several
experiments comparing the performance of humans and state-of-the-art models,
specifically GPT-4 and Gemini-1.5 Pro. Our results show that Gemini performs
better than GPT but still underperforms compared to humans, indicating
significant room for improvement. | cs.CL | Identifying User Goals from UI Trajectories |
2024-06-20T13:43:38Z | http://arxiv.org/pdf/2406.14313v1 | http://arxiv.org/abs/2406.14313v1 | Riya Sawhney, Indrajit Bhattacharya, Mausam | 2024-06-20T13:43:38Z | cs.CL, cs.AI | Real-world KBQA applications require models that are (1) robust -- e.g., can
differentiate between answerable and unanswerable questions, and (2)
low-resource -- do not require large training data. Towards this goal, we
propose the novel task of few-shot transfer for KBQA with unanswerable
questions. We present FUn-FuSIC that extends the state-of-the-art (SoTA)
few-shot transfer model for answerable-only KBQA to handle unanswerability. It
iteratively prompts an LLM to generate logical forms for the question by
providing feedback using a diverse suite of syntactic, semantic and execution
guided checks, and adapts self-consistency to assess confidence of the LLM to
decide answerability. Experiments over newly constructed datasets show that
FUn-FuSIC outperforms suitable adaptations of the SoTA model for KBQA with
unanswerability, and the SoTA model for answerable-only few-shot-transfer KBQA. | cs.CL | Robust Few-shot Transfer Learning for Knowledge Base Question Answering with Unanswerable Questions |
2024-06-20T13:43:03Z | http://arxiv.org/pdf/2406.14312v1 | http://arxiv.org/abs/2406.14312v1 | Abul Hasan, Jinge Wu, Quang Ngoc Nguyen, Salomé Andres, Imane Guellil, Huayu Zhang, Arlene Casey, Beatrice Alex, Bruce Guthrie, Honghan Wu | 2024-06-20T13:43:03Z | cs.CL, cs.AI | This study introduces a novel knowledge enhanced tokenisation mechanism,
K-Tokeniser, for clinical text processing. Technically, at initialisation
stage, K-Tokeniser populates global representations of tokens based on semantic
types of domain concepts (such as drugs or diseases) from either a domain
ontology like Unified Medical Language System or the training data of the task
related corpus. At training or inference stage, sentence level localised
context will be utilised for choosing the optimal global token representation
to realise the semantic-based tokenisation. To avoid pretraining using the new
tokeniser, an embedding initialisation approach is proposed to generate
representations for new tokens. Using three transformer-based language models,
a comprehensive set of experiments are conducted on four real-world datasets
for evaluating K-Tokeniser in a wide range of clinical text analytics tasks
including clinical concept and relation extraction, automated clinical coding,
clinical phenotype identification, and clinical research article
classification. Overall, our models demonstrate consistent improvements over
their counterparts in all tasks. In particular, substantial improvements are
observed in the automated clinical coding task with 13\% increase on Micro
$F_1$ score. Furthermore, K-Tokeniser also shows significant capacities in
facilitating quicker converge of language models. Specifically, using
K-Tokeniser, the language models would only require 50\% of the training data
to achieve the best performance of the baseline tokeniser using all training
data in the concept extraction task and less than 20\% of the data for the
automated coding task. It is worth mentioning that all these improvements
require no pre-training process, making the approach generalisable. | cs.CL | Infusing clinical knowledge into tokenisers for language models |
2024-06-20T13:37:10Z | http://arxiv.org/pdf/2406.14307v2 | http://arxiv.org/abs/2406.14307v2 | Chao Hui Huang | 2024-07-01T19:38:03Z | q-bio.GN, cs.CL, cs.CV | In this paper, we introduce QuST-LLM, an innovative extension of QuPath that
utilizes the capabilities of large language models (LLMs) to analyze and
interpret spatial transcriptomics (ST) data. In addition to simplifying the
intricate and high-dimensional nature of ST data by offering a comprehensive
workflow that includes data loading, region selection, gene expression
analysis, and functional annotation, QuST-LLM employs LLMs to transform complex
ST data into understandable and detailed biological narratives based on gene
ontology annotations, thereby significantly improving the interpretability of
ST data. Consequently, users can interact with their own ST data using natural
language. Hence, QuST-LLM provides researchers with a potent functionality to
unravel the spatial and functional complexities of tissues, fostering novel
insights and advancements in biomedical research. QuST-LLM is a part of QuST
project. The source code is hosted on GitHub and documentation is available at
(https://github.com/huangch/qust). | q-bio.GN | QuST-LLM: Integrating Large Language Models for Comprehensive Spatial Transcriptomics Analysis |
2024-06-20T13:09:29Z | http://arxiv.org/pdf/2406.14284v1 | http://arxiv.org/abs/2406.14284v1 | Pramit Bhattacharyya, Arnab Bhattacharya | 2024-06-20T13:09:29Z | cs.CL, cs.AI | Bangla (Bengali) is the fifth most spoken language globally and, yet, the
problem of automatic grammar correction in Bangla is still in its nascent
stage. This is mostly due to the need for a large corpus of grammatically
incorrect sentences, with their corresponding correct counterparts. The present
state-of-the-art techniques to curate a corpus for grammatically wrong
sentences involve random swapping, insertion and deletion of words.
However,these steps may not always generate grammatically wrong sentences in
Bangla. In this work, we propose a pragmatic approach to generate grammatically
wrong sentences in Bangla. We first categorize the different kinds of errors in
Bangla into 5 broad classes and 12 finer classes. We then use these to generate
grammatically wrong sentences systematically from a correct sentence. This
approach can generate a large number of wrong sentences and can, thus, mitigate
the challenge of lacking a large corpus for neural networks. We provide a
dataset, Vaiyakarana, consisting of 92,830 grammatically incorrect sentences as
well as 18,426 correct sentences. We also collected 619 human-generated
sentences from essays written by Bangla native speakers. This helped us to
understand errors that are more frequent. We evaluated our corpus against
neural models and LLMs and also benchmark it against human evaluators who are
native speakers of Bangla. Our analysis shows that native speakers are far more
accurate than state-of-the-art models to detect whether the sentence is
grammatically correct. Our methodology of generating erroneous sentences can be
applied for most other Indian languages as well. | cs.CL | VAIYAKARANA : A Benchmark for Automatic Grammar Correction in Bangla |
2024-06-20T13:07:38Z | http://arxiv.org/pdf/2406.14282v1 | http://arxiv.org/abs/2406.14282v1 | Junjie Wang, Mingyang Chen, Binbin Hu, Dan Yang, Ziqi Liu, Yue Shen, Peng Wei, Zhiqiang Zhang, Jinjie Gu, Jun Zhou, Jeff Z. Pan, Wen Zhang, Huajun Chen | 2024-06-20T13:07:38Z | cs.CL, cs.AI | Improving the performance of large language models (LLMs) in complex
question-answering (QA) scenarios has always been a research focal point.
Recent studies have attempted to enhance LLMs' performance by combining
step-wise planning with external retrieval. While effective for advanced models
like GPT-3.5, smaller LLMs face challenges in decomposing complex questions,
necessitating supervised fine-tuning. Previous work has relied on manual
annotation and knowledge distillation from teacher LLMs, which are
time-consuming and not accurate enough. In this paper, we introduce a novel
framework for enhancing LLMs' planning capabilities by using planning data
derived from knowledge graphs (KGs). LLMs fine-tuned with this data have
improved planning capabilities, better equipping them to handle complex QA
tasks that involve retrieval. Evaluations on multiple datasets, including our
newly proposed benchmark, highlight the effectiveness of our framework and the
benefits of KG-derived planning data. | cs.CL | Learning to Plan for Retrieval-Augmented Large Language Models from Knowledge Graphs |
2024-06-20T12:59:27Z | http://arxiv.org/pdf/2406.14277v1 | http://arxiv.org/abs/2406.14277v1 | Minsang Kim, Cheoneum Park, Seungjun Baek | 2024-06-20T12:59:27Z | cs.CL, cs.AI | Retrieval-augmented generation (RAG) has received much attention for
Open-domain question-answering (ODQA) tasks as a means to compensate for the
parametric knowledge of large language models (LLMs). While previous approaches
focused on processing retrieved passages to remove irrelevant context, they
still rely heavily on the quality of retrieved passages which can degrade if
the question is ambiguous or complex. In this paper, we propose a simple yet
efficient method called question and passage augmentation via LLMs for
open-domain QA. Our method first decomposes the original questions into
multiple-step sub-questions. By augmenting the original question with detailed
sub-questions and planning, we are able to make the query more specific on what
needs to be retrieved, improving the retrieval performance. In addition, to
compensate for the case where the retrieved passages contain distracting
information or divided opinions, we augment the retrieved passages with
self-generated passages by LLMs to guide the answer extraction. Experimental
results show that the proposed scheme outperforms the previous state-of-the-art
and achieves significant performance gain over existing RAG methods. | cs.CL | Augmenting Query and Passage for Retrieval-Augmented Generation using LLMs for Open-Domain Question Answering |
2024-06-20T12:58:26Z | http://arxiv.org/pdf/2406.14275v1 | http://arxiv.org/abs/2406.14275v1 | Xiangru Tang, Xingyao Zhang, Yanjun Shao, Jie Wu, Yilun Zhao, Arman Cohan, Ming Gong, Dongmei Zhang, Mark Gerstein | 2024-06-20T12:58:26Z | cs.CL, cs.AI | Large language models (LLMs) excel at a variety of natural language
processing tasks, yet they struggle to generate personalized content for
individuals, particularly in real-world scenarios like scientific writing.
Addressing this challenge, we introduce Step-Back Profiling to personalize LLMs
by distilling user history into concise profiles, including essential traits
and preferences of users. Regarding our experiments, we construct a
Personalized Scientific Writing (PSW) dataset to study multiuser
personalization. PSW requires the models to write scientific papers given
specialized author groups with diverse academic backgrounds. As for the
results, we demonstrate the effectiveness of capturing user characteristics via
Step-Back Profiling for collaborative writing. Moreover, our approach
outperforms the baselines by up to 3.6 points on the general personalization
benchmark (LaMP), including 7 personalization LLM tasks. Our extensive ablation
studies validate the contributions of different components in our method and
provide insights into our task definition. Our dataset and code are available
at \url{https://github.com/gersteinlab/step-back-profiling}. | cs.CL | Step-Back Profiling: Distilling User History for Personalized Scientific Writing |
2024-06-20T12:46:12Z | http://arxiv.org/pdf/2406.14267v1 | http://arxiv.org/abs/2406.14267v1 | Rochelle Choenni, Sara Rajaee, Christof Monz, Ekaterina Shutova | 2024-06-20T12:46:12Z | cs.CL, cs.AI | While multilingual language models (MLMs) have been trained on 100+
languages, they are typically only evaluated across a handful of them due to a
lack of available test data in most languages. This is particularly problematic
when assessing MLM's potential for low-resource and unseen languages. In this
paper, we present an analysis of existing evaluation frameworks in multilingual
NLP, discuss their limitations, and propose several directions for more robust
and reliable evaluation practices. Furthermore, we empirically study to what
extent machine translation offers a {reliable alternative to human translation}
for large-scale evaluation of MLMs across a wide set of languages. We use a
SOTA translation model to translate test data from 4 tasks to 198 languages and
use them to evaluate three MLMs. We show that while the selected subsets of
high-resource test languages are generally sufficiently representative of a
wider range of high-resource languages, we tend to overestimate MLMs' ability
on low-resource languages. Finally, we show that simpler baselines can achieve
relatively strong performance without having benefited from large-scale
multilingual pretraining. | cs.CL | On the Evaluation Practices in Multilingual NLP: Can Machine Translation Offer an Alternative to Human Translations? |
2024-06-20T11:51:00Z | http://arxiv.org/pdf/2406.14230v1 | http://arxiv.org/abs/2406.14230v1 | Han Jiang, Xiaoyuan Yi, Zhihua Wei, Shu Wang, Xing Xie | 2024-06-20T11:51:00Z | cs.CL, cs.AI, cs.CY | Warning: this paper contains model outputs exhibiting unethical information.
Large Language Models (LLMs) have achieved significant breakthroughs, but their
generated unethical content poses potential risks. Measuring value alignment of
LLMs becomes crucial for their regulation and responsible deployment. Numerous
datasets have been constructed to assess social bias, toxicity, and ethics in
LLMs, but they suffer from evaluation chronoeffect, that is, as models rapidly
evolve, existing data becomes leaked or undemanding, overestimating
ever-developing LLMs. To tackle this problem, we propose GETA, a novel
generative evolving testing approach that dynamically probes the underlying
moral baselines of LLMs. Distinct from previous adaptive testing methods that
rely on static datasets with limited difficulty, GETA incorporates an
iteratively-updated item generator which infers each LLM's moral boundaries and
generates difficulty-tailored testing items, accurately reflecting the true
alignment extent. This process theoretically learns a joint distribution of
item and model response, with item difficulty and value conformity as latent
variables, where the generator co-evolves with the LLM, addressing
chronoeffect. We evaluate various popular LLMs with diverse capabilities and
demonstrate that GETA can create difficulty-matching testing items and more
accurately assess LLMs' values, better consistent with their performance on
unseen OOD and i.i.d. items, laying the groundwork for future evaluation
paradigms. | cs.CL | Raising the Bar: Investigating the Values of Large Language Models via Generative Evolving Testing |
2024-06-20T11:27:29Z | http://arxiv.org/pdf/2406.14213v1 | http://arxiv.org/abs/2406.14213v1 | Alsu Sagirova, Mikhail Burtsev | 2024-06-20T11:27:29Z | cs.CL | Even though Transformers are extensively used for Natural Language Processing
tasks, especially for machine translation, they lack an explicit memory to
store key concepts of processed texts. This paper explores the properties of
the content of symbolic working memory added to the Transformer model decoder.
Such working memory enhances the quality of model predictions in machine
translation task and works as a neural-symbolic representation of information
that is important for the model to make correct translations. The study of
memory content revealed that translated text keywords are stored in the working
memory, pointing to the relevance of memory content to the processed text.
Also, the diversity of tokens and parts of speech stored in memory correlates
with the complexity of the corpora for machine translation task. | cs.CL | Complexity of Symbolic Representation in Working Memory of Transformer Correlates with the Complexity of a Task |
2024-06-20T10:59:02Z | http://arxiv.org/pdf/2406.14197v1 | http://arxiv.org/abs/2406.14197v1 | Franz Nowak, Anej Svete, Alexandra Butoi, Ryan Cotterell | 2024-06-20T10:59:02Z | cs.CL, cs.FL | The performance of modern language models (LMs) has been improved by
chain-of-thought (CoT) reasoning, i.e., the process of generating intermediate
results that guide the model towards a final answer. A possible explanation for
this improvement is that CoT reasoning extends an LM's computational power, as
RNNs and transformers with additional scratch space are known to be Turing
complete. Comparing LMs to Turing machines, however, introduces a category
error - Turing machines decide language membership, whereas LMs define
distributions over strings. To bridge this gap, we formalize CoT reasoning in a
probabilistic setting. We present several results on the representational
capacity of recurrent and transformer LMs with CoT reasoning, showing that they
can represent the same family of distributions over strings as probabilistic
Turing machines. | cs.CL | On the Representational Capacity of Neural Language Models with Chain-of-Thought Reasoning |
2024-06-20T10:52:14Z | http://arxiv.org/pdf/2406.14192v1 | http://arxiv.org/abs/2406.14192v1 | Zhaochen Su, Jun Zhang, Tong Zhu, Xiaoye Qu, Juntao Li, Min Zhang, Yu Cheng | 2024-06-20T10:52:14Z | cs.CL, cs.AI | Reasoning about time is essential for Large Language Models (LLMs) to
understand the world. Previous works focus on solving specific tasks, primarily
on time-sensitive question answering. While these methods have proven
effective, they cannot generalize to a wider spectrum of temporal reasoning
tasks. Therefore, we propose a crucial question: Can we build a universal
framework to handle a variety of temporal reasoning tasks? To that end, we
systematically study 38 temporal reasoning tasks. Based on the observation that
19 tasks are directly related to mathematics, we first leverage the available
mathematical dataset to set a solid foundation for temporal reasoning. However,
the in-depth study indicates that focusing solely on mathematical enhancement
falls short of addressing pure temporal reasoning tasks. To mitigate this
limitation, we propose a simple but effective self-critic temporal optimization
method to enhance the model's temporal reasoning capabilities without
sacrificing general task abilities. Finally, we develop Timo, a model designed
to excel in temporal reasoning at the 7B and 13B scales. Notably, Timo
outperforms the counterpart LLMs by 10.0 and 7.6 in average accuracy scores and
achieves the new state-of-the-art (SOTA) performance of comparable size.
Extensive experiments further validate our framework's effectiveness and its
generalization across diverse temporal tasks. The code is available at
https://github.com/zhaochen0110/Timo. | cs.CL | Timo: Towards Better Temporal Reasoning for Language Models |
2024-06-20T10:51:06Z | http://arxiv.org/pdf/2406.14191v1 | http://arxiv.org/abs/2406.14191v1 | Miao Su, ZiXuan Li, Zhuo Chen, Long Bai, Xiaolong Jin, Jiafeng Guo | 2024-06-20T10:51:06Z | cs.CL, cs.AI, cs.LG | Knowledge Base Question Answering (KBQA) has been a long-standing field to
answer questions based on knowledge bases. Recently, the evolving dynamics of
knowledge have attracted a growing interest in Temporal Knowledge Graph
Question Answering (TKGQA), an emerging task to answer temporal questions.
However, this field grapples with ambiguities in defining temporal questions
and lacks a systematic categorization of existing methods for TKGQA. In
response, this paper provides a thorough survey from two perspectives: the
taxonomy of temporal questions and the methodological categorization for TKGQA.
Specifically, we first establish a detailed taxonomy of temporal questions
engaged in prior studies. Subsequently, we provide a comprehensive review of
TKGQA techniques of two categories: semantic parsing-based and TKG
embedding-based. Building on this review, the paper outlines potential research
directions aimed at advancing the field of TKGQA. This work aims to serve as a
comprehensive reference for TKGQA and to stimulate further research. | cs.CL | Temporal Knowledge Graph Question Answering: A Survey |
2024-06-20T10:49:08Z | http://arxiv.org/pdf/2406.14189v1 | http://arxiv.org/abs/2406.14189v1 | Yaguang Li, Xin Chen | 2024-06-20T10:49:08Z | cs.CL | Generative models reliant on sequential autoregression have been at the
forefront of language generation for an extensive period, particularly
following the introduction of widely acclaimed transformers. Despite its
excellent performance, there are always some issues that we face today. For
example, problems such as hallucinations and getting trapped in a logic loop
may occur. To enhance the performance of existing systems, this paper
introduces a new method for generating sequences in natural language, which
involves generating the targeted sentence in a tree-traversing order. The paper
includes an illustration of the theoretical basis and validity of the approach,
as well as a comparison of its fundamentals with the diffusion model in graphic
generation. Finally, a module called SenTree is introduced for generating an
approximating binary tree. It is already available at
https://github.com/arklyg/sentree. Additionally, a joint training framework
based on this approach is proposed, incorporating the intrinsics of generative
adversarial networks. | cs.CL | In Tree Structure Should Sentence Be Generated |
2024-06-20T10:34:46Z | http://arxiv.org/pdf/2406.14177v1 | http://arxiv.org/abs/2406.14177v1 | Sara Papi, Marco Gaido, Matteo Negri, Luisa Bentivogli | 2024-06-20T10:34:46Z | cs.CL, cs.AI, cs.SD, eess.AS | This paper describes the FBK's participation in the Simultaneous Translation
Evaluation Campaign at IWSLT 2024. For this year's submission in the
speech-to-text translation (ST) sub-track, we propose SimulSeamless, which is
realized by combining AlignAtt and SeamlessM4T in its medium configuration. The
SeamlessM4T model is used "off-the-shelf" and its simultaneous inference is
enabled through the adoption of AlignAtt, a SimulST policy based on
cross-attention that can be applied without any retraining or adaptation of the
underlying model for the simultaneous task. We participated in all the Shared
Task languages (English->{German, Japanese, Chinese}, and Czech->English),
achieving acceptable or even better results compared to last year's
submissions. SimulSeamless, covering more than 143 source languages and 200
target languages, is released at: https://github.com/hlt-mt/FBK-fairseq/. | cs.CL | SimulSeamless: FBK at IWSLT 2024 Simultaneous Speech Translation |
2024-06-20T10:23:38Z | http://arxiv.org/pdf/2406.14171v1 | http://arxiv.org/abs/2406.14171v1 | Peijia Guo, Ziguang Li, Haibo Hu, Chao Huang, Ming Li, Rui Zhang | 2024-06-20T10:23:38Z | cs.AI, cs.CL | We conceptualize the process of understanding as information compression, and
propose a method for ranking large language models (LLMs) based on lossless
data compression. We demonstrate the equivalence of compression length under
arithmetic coding with cumulative negative log probabilities when using a large
language model as a prior, that is, the pre-training phase of the model is
essentially the process of learning the optimal coding length. At the same
time, the evaluation metric compression ratio can be obtained without actual
compression, which greatly saves overhead. In this paper, we use five large
language models as priors for compression, then compare their performance on
challenging natural language processing tasks, including sentence completion,
question answering, and coreference resolution. Experimental results show that
compression ratio and model performance are positively correlated, so it can be
used as a general metric to evaluate large language models. | cs.AI | Ranking LLMs by compression |
2024-06-20T10:13:08Z | http://arxiv.org/pdf/2406.14167v1 | http://arxiv.org/abs/2406.14167v1 | Mariia Fedorova, Andrey Kutuzov, Yves Scherrer | 2024-06-20T10:13:08Z | cs.CL | We use contextualized word definitions generated by large language models as
semantic representations in the task of diachronic lexical semantic change
detection (LSCD). In short, generated definitions are used as `senses', and the
change score of a target word is retrieved by comparing their distributions in
two time periods under comparison. On the material of five datasets and three
languages, we show that generated definitions are indeed specific and general
enough to convey a signal sufficient to rank sets of words by the degree of
their semantic change over time. Our approach is on par with or outperforms
prior non-supervised sense-based LSCD methods. At the same time, it preserves
interpretability and allows to inspect the reasons behind a specific shift in
terms of discrete definitions-as-senses. This is another step in the direction
of explainable semantic change modeling. | cs.CL | Definition generation for lexical semantic change detection |
2024-06-20T10:08:17Z | http://arxiv.org/pdf/2406.14164v1 | http://arxiv.org/abs/2406.14164v1 | Panagiotis Kaliosis, John Pavlopoulos, Foivos Charalampakos, Georgios Moschovis, Ion Androutsopoulos | 2024-06-20T10:08:17Z | cs.AI, cs.CL | Diagnostic Captioning (DC) automatically generates a diagnostic text from one
or more medical images (e.g., X-rays, MRIs) of a patient. Treated as a draft,
the generated text may assist clinicians, by providing an initial estimation of
the patient's condition, speeding up and helping safeguard the diagnostic
process. The accuracy of a diagnostic text, however, strongly depends on how
well the key medical conditions depicted in the images are expressed. We
propose a new data-driven guided decoding method that incorporates medical
information, in the form of existing tags capturing key conditions of the
image(s), into the beam search of the diagnostic text generation process. We
evaluate the proposed method on two medical datasets using four DC systems that
range from generic image-to-text systems with CNN encoders and RNN decoders to
pre-trained Large Language Models. The latter can also be used in few- and
zero-shot learning scenarios. In most cases, the proposed mechanism improves
performance with respect to all evaluation measures. We provide an open-source
implementation of the proposed method at https://github.com/nlpaueb/dmmcs. | cs.AI | A Data-Driven Guided Decoding Mechanism for Diagnostic Captioning |
2024-06-20T10:04:09Z | http://arxiv.org/pdf/2406.14162v1 | http://arxiv.org/abs/2406.14162v1 | Jingwei Ni, Tobias Schimanski, Meihong Lin, Mrinmaya Sachan, Elliott Ash, Markus Leippold | 2024-06-20T10:04:09Z | cs.IR, cs.AI, cs.CL | Retrieval Augmented Generation (RAG) is widely employed to ground responses
to queries on domain-specific documents. But do RAG implementations leave out
important information or excessively include irrelevant information? To allay
these concerns, it is necessary to annotate domain-specific benchmarks to
evaluate information retrieval (IR) performance, as relevance definitions vary
across queries and domains. Furthermore, such benchmarks should be
cost-efficiently annotated to avoid annotation selection bias. In this paper,
we propose DIRAS (Domain-specific Information Retrieval Annotation with
Scalability), a manual-annotation-free schema that fine-tunes open-sourced LLMs
to annotate relevance labels with calibrated relevance probabilities. Extensive
evaluation shows that DIRAS fine-tuned models achieve GPT-4-level performance
on annotating and ranking unseen (query, document) pairs, and is helpful for
real-world RAG development. | cs.IR | DIRAS: Efficient LLM-Assisted Annotation of Document Relevance in Retrieval Augmented Generation |
2024-06-20T09:53:23Z | http://arxiv.org/pdf/2406.14155v1 | http://arxiv.org/abs/2406.14155v1 | Dominik Stammbach, Philine Widmer, Eunjung Cho, Caglar Gulcehre, Elliott Ash | 2024-06-20T09:53:23Z | cs.CL | Large language models such as ChatGPT often exhibit striking political
biases. If users query them about political information, they might take a
normative stance and reinforce such biases. To overcome this, we align LLMs
with diverse political viewpoints from 100,000 comments written by candidates
running for national parliament in Switzerland. Such aligned models are able to
generate more accurate political viewpoints from Swiss parties compared to
commercial models such as ChatGPT. We also propose a procedure to generate
balanced overviews from multiple viewpoints using such models. | cs.CL | Aligning Large Language Models with Diverse Political Viewpoints |
2024-06-20T09:52:10Z | http://arxiv.org/pdf/2406.14154v1 | http://arxiv.org/abs/2406.14154v1 | David Hartmann, Amin Oueslati, Dimitri Staufer | 2024-06-20T09:52:10Z | cs.CY, cs.CL, cs.HC, cs.LG | Online platforms face the challenge of moderating an ever-increasing volume
of content, including harmful hate speech. In the absence of clear legal
definitions and a lack of transparency regarding the role of algorithms in
shaping decisions on content moderation, there is a critical need for external
accountability. Our study contributes to filling this gap by systematically
evaluating four leading cloud-based content moderation services through a
third-party audit, highlighting issues such as biases against minorities and
vulnerable groups that may arise through over-reliance on these services. Using
a black-box audit approach and four benchmark data sets, we measure performance
in explicit and implicit hate speech detection as well as counterfactual
fairness through perturbation sensitivity analysis and present disparities in
performance for certain target identity groups and data sets. Our analysis
reveals that all services had difficulties detecting implicit hate speech,
which relies on more subtle and codified messages. Moreover, our results point
to the need to remove group-specific bias. It seems that biases towards some
groups, such as Women, have been mostly rectified, while biases towards other
groups, such as LGBTQ+ and PoC remain. | cs.CY | Watching the Watchers: A Comparative Fairness Audit of Cloud-based Content Moderation Services |
2024-06-20T09:35:22Z | http://arxiv.org/pdf/2406.14144v1 | http://arxiv.org/abs/2406.14144v1 | Jianhui Chen, Xiaozhi Wang, Zijun Yao, Yushi Bai, Lei Hou, Juanzi Li | 2024-06-20T09:35:22Z | cs.CL, cs.AI, cs.LG | Large language models (LLMs) excel in various capabilities but also pose
safety risks such as generating harmful content and misinformation, even after
safety alignment. In this paper, we explore the inner mechanisms of safety
alignment from the perspective of mechanistic interpretability, focusing on
identifying and analyzing safety neurons within LLMs that are responsible for
safety behaviors. We propose generation-time activation contrasting to locate
these neurons and dynamic activation patching to evaluate their causal effects.
Experiments on multiple recent LLMs show that: (1) Safety neurons are sparse
and effective. We can restore $90$% safety performance with intervention only
on about $5$% of all the neurons. (2) Safety neurons encode transferrable
mechanisms. They exhibit consistent effectiveness on different red-teaming
datasets. The finding of safety neurons also interprets "alignment tax". We
observe that the identified key neurons for safety and helpfulness
significantly overlap, but they require different activation patterns of the
shared neurons. Furthermore, we demonstrate an application of safety neurons in
detecting unsafe outputs before generation. Our findings may promote further
research on understanding LLM alignment. The source codes will be publicly
released to facilitate future research. | cs.CL | Finding Safety Neurons in Large Language Models |
2024-06-20T09:27:33Z | http://arxiv.org/pdf/2406.14137v1 | http://arxiv.org/abs/2406.14137v1 | Shujin Wu, Yi R. Fung, Sha Li, Yixin Wan, Kai-Wei Chang, Heng Ji | 2024-06-20T09:27:33Z | cs.CL | Large vision-language models (LVLMs), while proficient in following
instructions and responding to diverse questions, invariably generate detailed
responses even when questions are ambiguous or unanswerable, leading to
hallucinations and bias issues. Thus, it is essential for LVLMs to proactively
engage with humans to ask for clarifications or additional information for
better responses. In this study, we aim to shift LVLMs from passive answer
providers to proactive engaged partners. We begin by establishing a
three-tiered hierarchy for questions of invalid, ambiguous, and personalizable
nature to measure the proactive engagement capabilities of LVLMs. Utilizing
this hierarchy, we create PIE, (ProactIve Engagement Evaluation) through GPT-4o
and human annotators, consisting of 853 questions across six distinct,
fine-grained question types that are verified by human annotators and
accompanied with well-defined metrics. Our evaluations on \benchmark indicate
poor performance of existing LVLMs, with the best-performing open-weights model
only achieving an Aggregate Align Rate (AAR) of 0.28. In response, we introduce
MACAROON, self-iMaginAtion for ContrAstive pReference OptimizatiON, which
instructs LVLMs to autonomously generate contrastive response pairs for
unlabeled questions given the task description and human-crafted criteria.
Then, the self-imagined data is formatted for conditional reinforcement
learning. Experimental results show MACAROON effectively improves LVLMs'
capabilities to be proactively engaged (0.84 AAR) while maintaining comparable
performance on general tasks. | cs.CL | MACAROON: Training Vision-Language Models To Be Your Engaged Partners |
2024-06-20T09:14:19Z | http://arxiv.org/pdf/2406.14129v1 | http://arxiv.org/abs/2406.14129v1 | Yifan Du, Kun Zhou, Yuqi Huo, Yifan Li, Wayne Xin Zhao, Haoyu Lu, Zijia Zhao, Bingning Wang, Weipeng Chen, Ji-Rong Wen | 2024-06-20T09:14:19Z | cs.CV, cs.CL, cs.MM | With the rapid development of video Multimodal Large Language Models (MLLMs),
numerous benchmarks have been proposed to assess their video understanding
capability. However, due to the lack of rich events in the videos, these
datasets may suffer from the short-cut bias that the answers can be deduced
from a few frames, without the need to watch the entire video. To address this
issue, we introduce Event-Bench, an event-oriented long video understanding
benchmark built on existing datasets and human annotations. Event-Bench
includes six event-related tasks and 2,190 test instances to comprehensively
evaluate video event understanding ability. Additionally, we propose Video
Instruction Merging~(VIM), a cost-effective method that enhances video MLLMs
using merged, event-intensive video instructions, addressing the scarcity of
human-annotated, event-intensive data. Extensive experiments show that the
best-performing model, GPT-4o, achieves an overall accuracy of 53.33,
significantly outperforming the best open-source model by 41.42%. Leveraging an
effective instruction synthesis method and an adaptive model architecture, VIM
surpasses both state-of-the-art open-source models and GPT-4V on the
Event-Bench. All code, data, and models are publicly available at
https://github.com/RUCAIBox/Event-Bench. | cs.CV | Towards Event-oriented Long Video Understanding |
2024-06-20T09:03:18Z | http://arxiv.org/pdf/2406.14117v1 | http://arxiv.org/abs/2406.14117v1 | Shuoqi Sun, Shengyao Zhuang, Shuai Wang, Guido Zuccon | 2024-06-20T09:03:18Z | cs.IR, cs.CL | We provide a systematic understanding of the impact of specific components
and wordings used in prompts on the effectiveness of rankers based on zero-shot
Large Language Models (LLMs). Several zero-shot ranking methods based on LLMs
have recently been proposed. Among many aspects, methods differ across (1) the
ranking algorithm they implement, e.g., pointwise vs. listwise, (2) the
backbone LLMs used, e.g., GPT3.5 vs. FLAN-T5, (3) the components and wording
used in prompts, e.g., the use or not of role-definition (role-playing) and the
actual words used to express this. It is currently unclear whether performance
differences are due to the underlying ranking algorithm, or because of spurious
factors such as better choice of words used in prompts. This confusion risks to
undermine future research. Through our large-scale experimentation and
analysis, we find that ranking algorithms do contribute to differences between
methods for zero-shot LLM ranking. However, so do the LLM backbones -- but even
more importantly, the choice of prompt components and wordings affect the
ranking. In fact, in our experiments, we find that, at times, these latter
elements have more impact on the ranker's effectiveness than the actual ranking
algorithms, and that differences among ranking methods become more blurred when
prompt variations are considered. | cs.IR | An Investigation of Prompt Variations for Zero-shot LLM-based Rankers |
2024-06-20T08:58:58Z | http://arxiv.org/pdf/2406.14115v1 | http://arxiv.org/abs/2406.14115v1 | Ziche Liu, Rui Ke, Feng Jiang, Haizhou Li | 2024-06-20T08:58:58Z | cs.CL | Data selection for fine-tuning Large Language Models (LLMs) aims to select a
high-quality subset from a given candidate dataset to train a Pending Fine-tune
Model (PFM) into a Selective-Enhanced Model (SEM). It can improve the model
performance and accelerate the training process. Although a few surveys have
investigated related works of data selection, there is a lack of comprehensive
comparison between existing methods due to their various experimental settings.
To address this issue, we first propose a three-stage scheme for data selection
and comprehensively review existing works according to this scheme. Then, we
design a unified comparing method with ratio-based efficiency indicators and
ranking-based feasibility indicators to overcome the difficulty of comparing
various models with diverse experimental settings. After an in-depth
comparative analysis, we find that the more targeted method with data-specific
and model-specific quality labels has higher efficiency, but the introduction
of additional noise information should be avoided when designing selection
algorithms. Finally, we summarize the trends in data selection and highlight
the short-term and long-term challenges to guide future research. | cs.CL | Take the essence and discard the dross: A Rethinking on Data Selection for Fine-Tuning Large Language Models |
2024-06-20T08:40:21Z | http://arxiv.org/pdf/2406.14106v1 | http://arxiv.org/abs/2406.14106v1 | Yuncong Li, Tianhua Xu, Sheng-hua Zhong, Haiqin Yang | 2024-06-20T08:40:21Z | cs.AI, cs.CL | Event Coreference Resolution (ECR) is the task of clustering event mentions
that refer to the same real-world event. Despite significant advancements, ECR
research faces two main challenges: limited generalizability across domains due
to narrow dataset evaluations, and difficulties in comparing models within
diverse ECR pipelines. To address these issues, we develop EasyECR, the first
open-source library designed to standardize data structures and abstract ECR
pipelines for easy implementation and fair evaluation. More specifically,
EasyECR integrates seven representative pipelines and ten popular benchmark
datasets, enabling model evaluations on various datasets and promoting the
development of robust ECR pipelines. By conducting extensive evaluation via our
EasyECR, we find that, \lowercase\expandafter{\romannumeral1}) the
representative ECR pipelines cannot generalize across multiple datasets, hence
evaluating ECR pipelines on multiple datasets is necessary,
\lowercase\expandafter{\romannumeral2}) all models in ECR pipelines have a
great effect on pipeline performance, therefore, when one model in ECR
pipelines are compared, it is essential to ensure that the other models remain
consistent. Additionally, reproducing ECR results is not trivial, and the
developed library can help reduce this discrepancy. The experimental results
provide valuable baselines for future research. | cs.AI | EasyECR: A Library for Easy Implementation and Evaluation of Event Coreference Resolution Models |
2024-06-20T08:24:57Z | http://arxiv.org/pdf/2406.14099v2 | http://arxiv.org/abs/2406.14099v2 | Federico Ruggeri, Eleonora Misino, Arianna Muti, Katerina Korre, Paolo Torroni, Alberto Barrón-Cedeño | 2024-07-02T15:38:17Z | cs.CL | We introduce the Guideline-Centered annotation process, a novel data
annotation methodology focused on reporting the annotation guidelines
associated with each data sample. We identify three main limitations of the
standard prescriptive annotation process and describe how the
Guideline-Centered methodology overcomes them by reducing the loss of
information in the annotation process and ensuring adherence to guidelines.
Additionally, we discuss how the Guideline-Centered enables the reuse of
annotated data across multiple tasks at the cost of a single human-annotation
process. | cs.CL | Let Guidelines Guide You: A Prescriptive Guideline-Centered Data Annotation Methodology |
2024-06-20T08:13:30Z | http://arxiv.org/pdf/2406.14092v1 | http://arxiv.org/abs/2406.14092v1 | Jing Xu, Minglin Wu, Xixin Wu, Helen Meng | 2024-06-20T08:13:30Z | cs.CL, eess.AS | Self-supervised (SSL) models have shown great performance in various
downstream tasks. However, they are typically developed for limited languages,
and may encounter new languages in real-world. Developing a SSL model for each
new language is costly. Thus, it is vital to figure out how to efficiently
adapt existed SSL models to a new language without impairing its original
abilities. We propose adaptation methods which integrate LoRA to existed SSL
models to extend new language. We also develop preservation strategies which
include data combination and re-clustering to retain abilities on existed
languages. Applied to mHuBERT, we investigate their effectiveness on speech
re-synthesis task. Experiments show that our adaptation methods enable mHuBERT
to be applied to a new language (Mandarin) with MOS value increased about 1.6
and the relative value of WER reduced up to 61.72%. Also, our preservation
strategies ensure that the performance on both existed and new languages
remains intact. | cs.CL | Seamless Language Expansion: Enhancing Multilingual Mastery in Self-Supervised Models |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.