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
|
|---|---|---|---|---|---|---|---|---|
2024-02-18T04:57:19Z | http://arxiv.org/pdf/2402.11457v2 | http://arxiv.org/abs/2402.11457v2 | Shiyu Ni, Keping Bi, Jiafeng Guo, Xueqi Cheng | 2024-06-11T08:08:47Z | cs.CL | Large Language Models (LLMs) have been found to have difficulty knowing they
do not possess certain knowledge and tend to provide specious answers in such
cases. Retrieval Augmentation (RA) has been extensively studied to mitigate
LLMs' hallucinations. However, due to the extra overhead and unassured quality
of retrieval, it may not be optimal to conduct RA all the time. A
straightforward idea is to only conduct retrieval when LLMs are uncertain about
a question. This motivates us to enhance the LLMs' ability to perceive their
knowledge boundaries to help RA. In this paper, we first quantitatively measure
LLMs' such ability and confirm their overconfidence. Then, we study how LLMs'
certainty about a question correlates with their dependence on external
retrieved information. We propose several methods to enhance LLMs' perception
of knowledge boundaries and show that they are effective in reducing
overconfidence. Additionally, equipped with these methods, LLMs can achieve
comparable or even better performance of RA with much fewer retrieval calls. | cs.CL | When Do LLMs Need Retrieval Augmentation? Mitigating LLMs' Overconfidence Helps Retrieval Augmentation |
2024-02-18T04:45:01Z | http://arxiv.org/pdf/2402.11456v2 | http://arxiv.org/abs/2402.11456v2 | Sebastian Antony Joseph, Lily Chen, Jan Trienes, Hannah Louisa Göke, Monika Coers, Wei Xu, Byron C Wallace, Junyi Jessy Li | 2024-06-05T01:27:00Z | cs.CL | Plain language summarization with LLMs can be useful for improving textual
accessibility of technical content. But how factual are these summaries in a
high-stakes domain like medicine? This paper presents FactPICO, a factuality
benchmark for plain language summarization of medical texts describing
randomized controlled trials (RCTs), which are the basis of evidence-based
medicine and can directly inform patient treatment. FactPICO consists of 345
plain language summaries of RCT abstracts generated from three LLMs (i.e.,
GPT-4, Llama-2, and Alpaca), with fine-grained evaluation and natural language
rationales from experts. We assess the factuality of critical elements of RCTs
in those summaries: Populations, Interventions, Comparators, Outcomes (PICO),
as well as the reported findings concerning these. We also evaluate the
correctness of the extra information (e.g., explanations) added by LLMs. Using
FactPICO, we benchmark a range of existing factuality metrics, including the
newly devised ones based on LLMs. We find that plain language summarization of
medical evidence is still challenging, especially when balancing between
simplicity and factuality, and that existing metrics correlate poorly with
expert judgments on the instance level. | cs.CL | FactPICO: Factuality Evaluation for Plain Language Summarization of Medical Evidence |
2024-02-18T04:41:25Z | http://arxiv.org/pdf/2402.11455v1 | http://arxiv.org/abs/2402.11455v1 | Hanqing Wang, Bowen Ping, Shuo Wang, Xu Han, Yun Chen, Zhiyuan Liu, Maosong Sun | 2024-02-18T04:41:25Z | cs.CL | LoRA employs lightweight modules to customize large language models (LLMs)
for each downstream task or domain, where different learned additional modules
represent diverse skills. Combining existing LoRAs to address new tasks can
enhance the reusability of learned LoRAs, particularly beneficial for tasks
with limited annotated data. Most prior works on LoRA combination primarily
rely on task-level weights for each involved LoRA, making different examples
and tokens share the same LoRA weights. However, in generative tasks, different
tokens may necessitate diverse skills to manage. Taking the Chinese math task
as an example, understanding the problem description may depend more on the
Chinese LoRA, while the calculation part may rely more on the math LoRA. To
this end, we propose LoRA-Flow, which utilizes dynamic weights to adjust the
impact of different LoRAs. The weights at each step are determined by a fusion
gate with extremely few parameters, which can be learned with only 200 training
examples. Experiments across six generative tasks demonstrate that our method
consistently outperforms baselines with task-level fusion weights. This
underscores the necessity of introducing dynamic fusion weights for LoRA
combination. | cs.CL | LoRA-Flow: Dynamic LoRA Fusion for Large Language Models in Generative Tasks |
2024-02-18T04:28:28Z | http://arxiv.org/pdf/2402.11453v3 | http://arxiv.org/abs/2402.11453v3 | Zhiyu Yang, Zihan Zhou, Shuo Wang, Xin Cong, Xu Han, Yukun Yan, Zhenghao Liu, Zhixing Tan, Pengyuan Liu, Dong Yu, Zhiyuan Liu, Xiaodong Shi, Maosong Sun | 2024-03-19T14:44:22Z | cs.CL | Scientific data visualization plays a crucial role in research by enabling
the direct display of complex information and assisting researchers in
identifying implicit patterns. Despite its importance, the use of Large
Language Models (LLMs) for scientific data visualization remains rather
unexplored. In this study, we introduce MatPlotAgent, an efficient
model-agnostic LLM agent framework designed to automate scientific data
visualization tasks. Leveraging the capabilities of both code LLMs and
multi-modal LLMs, MatPlotAgent consists of three core modules: query
understanding, code generation with iterative debugging, and a visual feedback
mechanism for error correction. To address the lack of benchmarks in this
field, we present MatPlotBench, a high-quality benchmark consisting of 100
human-verified test cases. Additionally, we introduce a scoring approach that
utilizes GPT-4V for automatic evaluation. Experimental results demonstrate that
MatPlotAgent can improve the performance of various LLMs, including both
commercial and open-source models. Furthermore, the proposed evaluation method
shows a strong correlation with human-annotated scores. | cs.CL | MatPlotAgent: Method and Evaluation for LLM-Based Agentic Scientific Data Visualization |
2024-02-18T04:28:16Z | http://arxiv.org/pdf/2402.11452v1 | http://arxiv.org/abs/2402.11452v1 | Zhaorun Chen, Zhuokai Zhao, Zhihong Zhu, Ruiqi Zhang, Xiang Li, Bhiksha Raj, Huaxiu Yao | 2024-02-18T04:28:16Z | cs.CL | Recent advancements in large language models (LLMs) have shown promise in
multi-step reasoning tasks, yet their reliance on extensive manual labeling to
provide procedural feedback remains a significant impediment. To address this
challenge, in this paper, we propose a novel self-supervised framework AutoPRM
that efficiently enhances the fine-tuning of LLMs for intricate reasoning
challenges. Specifically, AutoPRM first decomposes complex problems into more
manageable subquestions with a controllable granularity switch, then
sequentially apply reinforcement learning to iteratively improve the
subquestion solver. Additionally, we propose context-guided-decoding to avoid
reward tampering and guide the subquestion solver towards the solution of the
holistic problem. Extensive experiments show that AutoPRM significantly
improves performance on mathematical and commonsense reasoning tasks over SOTA.
More encouragingly, AutoPRM can be easily integrated with other orthogonal
reasoning pipelines. | cs.CL | AutoPRM: Automating Procedural Supervision for Multi-Step Reasoning via Controllable Question Decomposition |
2024-02-18T04:19:44Z | http://arxiv.org/pdf/2402.11451v2 | http://arxiv.org/abs/2402.11451v2 | Yubo Ma, Zhibin Gou, Junheng Hao, Ruochen Xu, Shuohang Wang, Liangming Pan, Yujiu Yang, Yixin Cao, Aixin Sun, Hany Awadalla, Weizhu Chen | 2024-02-21T03:04:49Z | cs.CL, cs.AI | Scientific reasoning poses an excessive challenge for even the most advanced
Large Language Models (LLMs). To make this task more practical and solvable for
LLMs, we introduce a new task setting named tool-augmented scientific
reasoning. This setting supplements LLMs with scalable toolsets, and shifts the
focus from pursuing an omniscient problem solver to a proficient tool-user. To
facilitate the research of such setting, we construct a tool-augmented training
corpus named MathFunc which encompasses over 30,000 samples and roughly 6,000
tools. Building on MathFunc, we develop SciAgent to retrieve, understand and,
if necessary, use tools for scientific problem solving. Additionally, we craft
a benchmark, SciToolBench, spanning five scientific domains to evaluate LLMs'
abilities with tool assistance. Extensive experiments on SciToolBench confirm
the effectiveness of SciAgent. Notably, SciAgent-Mistral-7B surpasses other
LLMs with the same size by more than 13% in absolute accuracy. Furthermore,
SciAgent-DeepMath-7B shows much superior performance than ChatGPT. | cs.CL | SciAgent: Tool-augmented Language Models for Scientific Reasoning |
2024-02-18T04:08:10Z | http://arxiv.org/pdf/2402.11447v1 | http://arxiv.org/abs/2402.11447v1 | Zhichao Xu, Daniel Cohen, Bei Wang, Vivek Srikumar | 2024-02-18T04:08:10Z | cs.CL | By allowing models to predict without task-specific training, in-context
learning (ICL) with pretrained LLMs has enormous potential in NLP. However, a
number of problems persist in ICL. In particular, its performance is sensitive
to the choice and order of in-context examples. Given the same set of
in-context examples with different orderings, model performance may vary
between near random to near state-of-the-art. In this work, we formulate
in-context example ordering as an optimization problem. We examine three
problem settings that differ in the assumptions they make about what is known
about the task. Inspired by the idea of learning from label proportions, we
propose two principles for in-context example ordering guided by model's
probability predictions. We apply our proposed principles to thirteen text
classification datasets and nine different autoregressive LLMs with 700M to 13B
parameters. We demonstrate that our approach outperforms the baselines by
improving the classification accuracy, reducing model miscalibration, and also
by selecting better in-context examples. | cs.CL | In-Context Example Ordering Guided by Label Distributions |
2024-02-18T03:40:06Z | http://arxiv.org/pdf/2402.11443v1 | http://arxiv.org/abs/2402.11443v1 | Siyuan Wang, Zhuohan Long, Zhihao Fan, Zhongyu Wei, Xuanjing Huang | 2024-02-18T03:40:06Z | cs.CL | This paper presents a benchmark self-evolving framework to dynamically
evaluate rapidly advancing Large Language Models (LLMs), aiming for a more
accurate assessment of their capabilities and limitations. We utilize a
multi-agent system to manipulate the context or question of original instances,
reframing new evolving instances with high confidence that dynamically extend
existing benchmarks. Towards a more scalable, robust and fine-grained
evaluation, we implement six reframing operations to construct evolving
instances testing LLMs against diverse queries, data noise and probing their
problem-solving sub-abilities. With this framework, we extend benchmark
datasets of four tasks. Experimental results show a general performance decline
in most LLMs against their original results. This decline under our scalable
and robust evaluations, alongside our fine-grained evaluation, more accurately
reflect models' capabilities. Besides, our framework widens performance
discrepancies both between different models and within the same model across
various tasks, facilitating more informed model selection for specific tasks
(Code and data are available at
https://github.com/NanshineLoong/Self-Evolving-Benchmark). | cs.CL | Benchmark Self-Evolving: A Multi-Agent Framework for Dynamic LLM Evaluation |
2024-02-18T03:38:51Z | http://arxiv.org/pdf/2402.11442v3 | http://arxiv.org/abs/2402.11442v3 | Siyuan Wang, Zhongyu Wei, Yejin Choi, Xiang Ren | 2024-06-21T02:36:55Z | cs.CL | Large language models (LLMs) have achieved impressive human-like performance
across various reasoning tasks. However, their mastery of underlying
inferential rules still falls short of human capabilities. To investigate this,
we propose a logic scaffolding inferential rule generation framework, to
construct an inferential rule base, ULogic, comprising both primitive and
compositional rules across five domains. Our analysis of GPT-series models over
a rule subset reveals significant gaps in LLMs' logic understanding compared to
human performance, especially in compositional and structural complex rules
with certain bias patterns. We further distill these rules into a smaller-scale
inference engine for flexible rule generation and enhancing downstream
reasoning. Through a multi-judger evaluation, our inference engine proves
effective in generating accurate, complex and abstract conclusions and
premises, and improve various commonsense reasoning tasks. Overall, our work
sheds light on LLMs' limitations in grasping inferential rule and suggests ways
to enhance their logical reasoning abilities~\footnote{Code and data are
available at \url{https://github.com/SiyuanWangw/ULogic}.}. | cs.CL | Can LLMs Reason with Rules? Logic Scaffolding for Stress-Testing and Improving LLMs |
2024-02-18T03:36:26Z | http://arxiv.org/pdf/2402.11441v1 | http://arxiv.org/abs/2402.11441v1 | Fali Wang, Runxue Bao, Suhang Wang, Wenchao Yu, Yanchi Liu, Wei Cheng, Haifeng Chen | 2024-02-18T03:36:26Z | cs.CL, cs.AI, cs.LG | Though Large Language Models (LLMs) have shown remarkable open-generation
capabilities across diverse domains, they struggle with knowledge-intensive
tasks. To alleviate this issue, knowledge integration methods have been
proposed to enhance LLMs with domain-specific knowledge graphs using external
modules. However, they suffer from data inefficiency as they require both known
and unknown knowledge for fine-tuning. Thus, we study a novel problem of
integrating unknown knowledge into LLMs efficiently without unnecessary overlap
of known knowledge. Injecting new knowledge poses the risk of forgetting
previously acquired knowledge. To tackle this, we propose a novel
Infuser-Guided Knowledge Integration (InfuserKI) framework that utilizes
transformer internal states to determine whether to enhance the original LLM
output with additional information, thereby effectively mitigating knowledge
forgetting. Evaluations on the UMLS-2.5k and MetaQA domain knowledge graphs
demonstrate that InfuserKI can effectively acquire new knowledge and outperform
state-of-the-art baselines by 9% and 6%, respectively, in reducing knowledge
forgetting. | cs.CL | InfuserKI: Enhancing Large Language Models with Knowledge Graphs via Infuser-Guided Knowledge Integration |
2024-02-18T03:10:39Z | http://arxiv.org/pdf/2402.11436v2 | http://arxiv.org/abs/2402.11436v2 | Wenda Xu, Guanglei Zhu, Xuandong Zhao, Liangming Pan, Lei Li, William Yang Wang | 2024-06-18T04:41:07Z | cs.CL, cs.AI | Recent studies show that large language models (LLMs) improve their
performance through self-feedback on certain tasks while degrade on others. We
discovered that such a contrary is due to LLM's bias in evaluating their own
output. In this paper, we formally define LLM's self-bias - the tendency to
favor its own generation - using two statistics. We analyze six LLMs (GPT-4,
GPT-3.5, Gemini, LLaMA2, Mixtral and DeepSeek) on translation, constrained text
generation, and mathematical reasoning tasks. We find that self-bias is
prevalent in all examined LLMs across multiple languages and tasks. Our
analysis reveals that while the self-refine pipeline improves the fluency and
understandability of model outputs, it further amplifies self-bias. To mitigate
such biases, we discover that larger model size and external feedback with
accurate assessment can significantly reduce bias in the self-refine pipeline,
leading to actual performance improvement in downstream tasks. The code and
data are released at https://github.com/xu1998hz/llm_self_bias. | cs.CL | Pride and Prejudice: LLM Amplifies Self-Bias in Self-Refinement |
2024-02-18T02:52:54Z | http://arxiv.org/pdf/2402.11432v2 | http://arxiv.org/abs/2402.11432v2 | Kang Chen, Zheng Lian, Haiyang Sun, Bin Liu, Jianhua Tao | 2024-06-17T02:57:09Z | cs.CL | Deception detection has attracted increasing attention due to its importance
in real-world scenarios. Its main goal is to detect deceptive behaviors from
multimodal clues such as gestures, facial expressions, prosody, etc. However,
these bases are usually subjective and related to personal habits. Therefore,
we extend deception detection to deception reasoning, further providing
objective evidence to support subjective judgment. Specifically, we provide
potential lies and basic facts and then analyze why this sentence may be a lie
by combining factual inconsistencies and intent behind them. Compared with
deception detection, this task is more applicable to real-world scenarios. For
example, in interrogation, the police should judge whether a person is lying
based on solid evidence. This paper presents our initial attempts at this task,
including constructing a dataset and defining evaluation metrics. Meanwhile,
this task can serve as a benchmark for evaluating the complex reasoning
capability of large language models. Code and data will be made publicly
available. | cs.CL | Can Deception Detection Go Deeper? Dataset, Evaluation, and Benchmark for Deception Reasoning |
2024-02-18T02:41:06Z | http://arxiv.org/pdf/2402.11430v1 | http://arxiv.org/abs/2402.11430v1 | Jun Gao, Huan Zhao, Wei Wang, Changlong Yu, Ruifeng Xu | 2024-02-18T02:41:06Z | cs.CL | In this study, we present EventRL, a reinforcement learning approach
developed to enhance event extraction for large language models (LLMs). EventRL
utilizes outcome supervision with specific reward functions to tackle prevalent
challenges in LLMs, such as instruction following and hallucination, manifested
as the mismatch of event structure and the generation of undefined event types.
We evaluate EventRL against existing methods like Few-Shot Prompting (FSP)
(based on GPT4) and Supervised Fine-Tuning (SFT) across various LLMs, including
GPT-4, LLaMa, and CodeLLaMa models. Our findings show that EventRL
significantly outperforms these conventional approaches by improving the
performance in identifying and structuring events, particularly in handling
novel event types. The study emphasizes the critical role of reward function
selection and demonstrates the benefits of incorporating code data for better
event extraction. While increasing model size leads to higher accuracy,
maintaining the ability to generalize is essential to avoid overfitting. | cs.CL | EventRL: Enhancing Event Extraction with Outcome Supervision for Large Language Models |
2024-02-18T01:46:46Z | http://arxiv.org/pdf/2402.11422v1 | http://arxiv.org/abs/2402.11422v1 | Peng Xing, Yinghui Li, Shirong Ma, Xinnian Liang, Haojing Huang, Yangning Li, Hai-Tao Zheng, Wenhao Jiang, Ying Shen | 2024-02-18T01:46:46Z | cs.CL | Chinese Spelling Correction (CSC) aims to detect and correct spelling errors
in given sentences. Recently, multi-domain CSC has gradually attracted the
attention of researchers because it is more practicable. In this paper, we
focus on the key flaw of the CSC model when adapting to multi-domain scenarios:
the tendency to forget previously acquired knowledge upon learning new
domain-specific knowledge (i.e., catastrophic forgetting). To address this, we
propose a novel model-agnostic Multi-stage Knowledge Transfer (MKT) framework,
which utilizes a continuously evolving teacher model for knowledge transfer in
each domain, rather than focusing solely on new domain knowledge. It deserves
to be mentioned that we are the first to apply continual learning methods to
the multi-domain CSC task. Experiments prove the effectiveness of our proposed
method, and further analyses demonstrate the importance of overcoming
catastrophic forgetting for improving the model performance. | cs.CL | Mitigating Catastrophic Forgetting in Multi-domain Chinese Spelling Correction by Multi-stage Knowledge Transfer Framework |
2024-02-18T01:40:34Z | http://arxiv.org/pdf/2402.11420v1 | http://arxiv.org/abs/2402.11420v1 | Yinghui Li, Shang Qin, Jingheng Ye, Shirong Ma, Yangning Li, Libo Qin, Xuming Hu, Wenhao Jiang, Hai-Tao Zheng, Philip S. Yu | 2024-02-18T01:40:34Z | cs.CL | Recently, Large Language Models (LLMs) have been widely studied by
researchers for their roles in various downstream NLP tasks. As a fundamental
task in the NLP field, Chinese Grammatical Error Correction (CGEC) aims to
correct all potential grammatical errors in the input sentences. Previous
studies have shown that LLMs' performance as correctors on CGEC remains
unsatisfactory due to its challenging task focus. To promote the CGEC field to
better adapt to the era of LLMs, we rethink the roles of LLMs in the CGEC task
so that they can be better utilized and explored in CGEC. Considering the rich
grammatical knowledge stored in LLMs and their powerful semantic understanding
capabilities, we utilize LLMs as explainers to provide explanation information
for the CGEC small models during error correction to enhance performance. We
also use LLMs as evaluators to bring more reasonable CGEC evaluations, thus
alleviating the troubles caused by the subjectivity of the CGEC task. In
particular, our work is also an active exploration of how LLMs and small models
better collaborate in downstream tasks. Extensive experiments and detailed
analyses on widely used datasets verify the effectiveness of our thinking
intuition and the proposed methods. | cs.CL | Rethinking the Roles of Large Language Models in Chinese Grammatical Error Correction |
2024-02-18T01:20:00Z | http://arxiv.org/pdf/2402.11417v1 | http://arxiv.org/abs/2402.11417v1 | Yifan Yang, Jiajun Zhou, Ngai Wong, Zheng Zhang | 2024-02-18T01:20:00Z | cs.CL, cs.AI, cs.LG | Various parameter-efficient fine-tuning (PEFT) techniques have been proposed
to enable computationally efficient fine-tuning while maintaining model
performance. However, existing PEFT methods are still limited by the growing
number of trainable parameters with the rapid deployment of Large Language
Models (LLMs). To address this challenge, we present LoRETTA, an
ultra-parameter-efficient framework that significantly reduces trainable
parameters through tensor-train decomposition. Specifically, we propose two
methods, named {LoRETTA}$_{adp}$ and {LoRETTA}$_{rep}$. The former employs
tensorized adapters, offering a high-performance yet lightweight approach for
the fine-tuning of LLMs. The latter emphasizes fine-tuning via weight
parameterization with a set of small tensor factors. LoRETTA achieves
comparable or better performance than most widely used PEFT methods with up to
$100\times$ fewer parameters on the LLaMA-2-7B models. Furthermore, empirical
results demonstrate that the proposed method effectively improves training
efficiency, enjoys better multi-task learning performance, and enhances the
anti-overfitting capability. Plug-and-play codes built upon the Huggingface
framework and PEFT library will be released. | cs.CL | LoRETTA: Low-Rank Economic Tensor-Train Adaptation for Ultra-Low-Parameter Fine-Tuning of Large Language Models |
2024-02-18T01:03:25Z | http://arxiv.org/pdf/2402.11414v1 | http://arxiv.org/abs/2402.11414v1 | Liqiang Jing, Jingxuan Zuo, Yue Zhang | 2024-02-18T01:03:25Z | cs.CL | Multimodal summarization aims to generate a concise summary based on the
input text and image. However, the existing methods potentially suffer from
unfactual output. To evaluate the factuality of multimodal summarization
models, we propose two fine-grained and explainable evaluation frameworks
(FALLACIOUS) for different application scenarios, i.e. reference-based
factuality evaluation framework and reference-free factuality evaluation
framework. Notably, the reference-free factuality evaluation framework doesn't
need ground truth and hence it has a wider application scenario. To evaluate
the effectiveness of the proposed frameworks, we compute the correlation
between our frameworks and the other metrics. The experimental results show the
effectiveness of our proposed method. We will release our code and dataset via
github. | cs.CL | Fine-grained and Explainable Factuality Evaluation for Multimodal Summarization |
2024-02-18T00:56:16Z | http://arxiv.org/pdf/2402.11411v1 | http://arxiv.org/abs/2402.11411v1 | Yiyang Zhou, Chenhang Cui, Rafael Rafailov, Chelsea Finn, Huaxiu Yao | 2024-02-18T00:56:16Z | cs.LG, cs.CL, cs.CV | Instruction-following Vision Large Language Models (VLLMs) have achieved
significant progress recently on a variety of tasks. These approaches merge
strong pre-trained vision models and large language models (LLMs). Since these
components are trained separately, the learned representations need to be
aligned with joint training on additional image-language pairs. This procedure
is not perfect and can cause the model to hallucinate - provide answers that do
not accurately reflect the image, even when the core LLM is highly factual and
the vision backbone has sufficiently complete representations. In this work, we
frame the hallucination problem as an alignment issue, tackle it with
preference tuning. Specifically, we propose POVID to generate feedback data
with AI models. We use ground-truth instructions as the preferred response and
a two-stage approach to generate dispreferred data. First, we prompt GPT-4V to
inject plausible hallucinations into the correct answer. Second, we distort the
image to trigger the inherent hallucination behavior of the VLLM. This is an
automated approach, which does not rely on human data generation or require a
perfect expert, which makes it easily scalable. Finally, both of these
generation strategies are integrated into an RLHF pipeline via Direct
Preference Optimization. In experiments across broad benchmarks, we show that
we can not only reduce hallucinations, but improve model performance across
standard benchmarks, outperforming prior approaches. Our data and code are
available at https://github.com/YiyangZhou/POVID. | cs.LG | Aligning Modalities in Vision Large Language Models via Preference Fine-tuning |
2024-02-18T00:32:33Z | http://arxiv.org/pdf/2402.11409v2 | http://arxiv.org/abs/2402.11409v2 | Zhichao Xu, Jiepu Jiang | 2024-04-16T16:34:02Z | cs.CL | Empathy is critical for effective and satisfactory conversational
communication. Prior efforts to measure conversational empathy mostly focus on
expressed communicative intents -- that is, the way empathy is expressed. Yet,
these works ignore the fact that conversation is also a collaboration involving
both speakers and listeners. In contrast, we propose a multi-dimensional
empathy evaluation framework to measure both expressed intents from the
speaker's perspective and perceived empathy from the listener's perspective. We
apply our proposed framework to analyze our internal customer-service dialogue.
We find the two dimensions (expressed intent types and perceived empathy) are
inter-connected, and perceived empathy has a high correlation with dialogue
satisfaction levels.
To reduce the annotation cost, we explore different options to automatically
measure conversational empathy: prompting LLMs and training language
model-based classifiers. Our experiments show that prompting methods with even
popular models like GPT-4 and Flan family models perform relatively poorly on
both public and our internal datasets. In contrast, instruction-finetuned
classifiers based on Flan-T5 family models outperform prior works and
competitive baselines. We conduct a detailed ablation study to give more
insights into instruction finetuning method's strong performance. | cs.CL | Multi-dimensional Evaluation of Empathetic Dialog Responses |
2024-02-18T00:04:40Z | http://arxiv.org/pdf/2402.11406v2 | http://arxiv.org/abs/2402.11406v2 | Min Zhang, Jianfeng He, Taoran Ji, Chang-Tien Lu | 2024-02-26T16:43:37Z | cs.CL | The fairness and trustworthiness of Large Language Models (LLMs) are
receiving increasing attention. Implicit hate speech, which employs indirect
language to convey hateful intentions, occupies a significant portion of
practice. However, the extent to which LLMs effectively address this issue
remains insufficiently examined. This paper delves into the capability of LLMs
to detect implicit hate speech (Classification Task) and express confidence in
their responses (Calibration Task). Our evaluation meticulously considers
various prompt patterns and mainstream uncertainty estimation methods. Our
findings highlight that LLMs exhibit two extremes: (1) LLMs display excessive
sensitivity towards groups or topics that may cause fairness issues, resulting
in misclassifying benign statements as hate speech. (2) LLMs' confidence scores
for each method excessively concentrate on a fixed range, remaining unchanged
regardless of the dataset's complexity. Consequently, the calibration
performance is heavily reliant on primary classification accuracy. These
discoveries unveil new limitations of LLMs, underscoring the need for caution
when optimizing models to ensure they do not veer towards extremes. This serves
as a reminder to carefully consider sensitivity and confidence in the pursuit
of model fairness. | cs.CL | Don't Go To Extremes: Revealing the Excessive Sensitivity and Calibration Limitations of LLMs in Implicit Hate Speech Detection |
2024-02-17T22:50:38Z | http://arxiv.org/pdf/2402.11399v2 | http://arxiv.org/abs/2402.11399v2 | Abe Bohan Hou, Jingyu Zhang, Yichen Wang, Daniel Khashabi, Tianxing He | 2024-06-08T04:24:27Z | cs.CL, cs.CR, cs.CY, cs.LG | Recent watermarked generation algorithms inject detectable signatures during
language generation to facilitate post-hoc detection. While token-level
watermarks are vulnerable to paraphrase attacks, SemStamp (Hou et al., 2023)
applies watermark on the semantic representation of sentences and demonstrates
promising robustness. SemStamp employs locality-sensitive hashing (LSH) to
partition the semantic space with arbitrary hyperplanes, which results in a
suboptimal tradeoff between robustness and speed. We propose k-SemStamp, a
simple yet effective enhancement of SemStamp, utilizing k-means clustering as
an alternative of LSH to partition the embedding space with awareness of
inherent semantic structure. Experimental results indicate that k-SemStamp
saliently improves its robustness and sampling efficiency while preserving the
generation quality, advancing a more effective tool for machine-generated text
detection. | cs.CL | k-SemStamp: A Clustering-Based Semantic Watermark for Detection of Machine-Generated Text |
2024-02-17T22:46:44Z | http://arxiv.org/pdf/2402.11398v2 | http://arxiv.org/abs/2402.11398v2 | Shaochen Xu, Zihao Wu, Huaqin Zhao, Peng Shu, Zhengliang Liu, Wenxiong Liao, Sheng Li, Andrea Sikora, Tianming Liu, Xiang Li | 2024-02-20T22:23:42Z | cs.CL, cs.AI | In this study, we leverage LLM to enhance the semantic analysis and develop
similarity metrics for texts, addressing the limitations of traditional
unsupervised NLP metrics like ROUGE and BLEU. We develop a framework where LLMs
such as GPT-4 are employed for zero-shot text identification and label
generation for radiology reports, where the labels are then used as
measurements for text similarity. By testing the proposed framework on the
MIMIC data, we find that GPT-4 generated labels can significantly improve the
semantic similarity assessment, with scores more closely aligned with clinical
ground truth than traditional NLP metrics. Our work demonstrates the
possibility of conducting semantic analysis of the text data using
semi-quantitative reasoning results by the LLMs for highly specialized domains.
While the framework is implemented for radiology report similarity analysis,
its concept can be extended to other specialized domains as well. | cs.CL | Reasoning before Comparison: LLM-Enhanced Semantic Similarity Metrics for Domain Specialized Text Analysis |
2024-02-17T22:37:17Z | http://arxiv.org/pdf/2402.14833v1 | http://arxiv.org/abs/2402.14833v1 | Jiayi Liu, Tinghan Yang, Jennifer Neville | 2024-02-17T22:37:17Z | cs.CL, cs.AI, cs.LG | Large language models (LLMs) have become pivotal in recent research. However,
during the inference process, LLMs still require substantial resources. In this
paper, we propose CliqueParcel, a method designed to improve the efficiency of
LLMs via prompt batching. Existing strategies to optimize inference efficiency
often compromise on output quality, leading to a discounted output problem.
This issue might result in reduced accuracy or outputs that are less detailed.
CliqueParcel is our answer to this challenge. While ensuring accuracy and
minimizing deviations from the original outputs (i.e., faithfulness), our
method significantly improves efficiency during inference.
To lay the groundwork, we first redefine efficiency measurements by excluding
the reduction in running time due to shorter lengths. Then, we provide a
comprehensive trade-off between efficiency and faithfulness to clarify the
nature of the 'discounted output' problem. Within the CliqueParcel framework,
we suggest multiple batching sub-methods and discuss the specific scenarios in
which they can be applied. During evaluation, CliqueParcel is tested on eight
widely recognized datasets, which can be classified into three types: reading
comprehension, open-source question-answering, and reasoning. Our experiments
explore the performance of CliqueParcel, including efficiency, faithfulness,
and the trade-off between them. This work provides novel insights into
inference efficiency and demonstrates promising performance. | cs.CL | CliqueParcel: An Approach For Batching LLM Prompts That Jointly Optimizes Efficiency And Faithfulness |
2024-02-17T18:31:21Z | http://arxiv.org/pdf/2402.11359v3 | http://arxiv.org/abs/2402.11359v3 | Shaokun Zhang, Jieyu Zhang, Jiale Liu, Linxin Song, Chi Wang, Ranjay Krishna, Qingyun Wu | 2024-06-07T23:49:18Z | cs.AI, cs.CL | Researchers and practitioners have recently reframed powerful Large Language
Models (LLMs) as agents, enabling them to automate complex tasks largely via
the use of specialized functions. To facilitate the development of LLM agents,
we present a novel paradigm of training LLM agents without modifying the LLM
weights, which is particularly useful when the LLMs are difficult or
inaccessible for modifications. Inspired by how humans continuously forge tools
to adapt to real-world tasks, rather than change our biological structure to
fit a static set of tools, we propose to progressively forge agent's functions
to better solve the downstream tasks instead of modifying the LLM weights. By
treating the functions as learnable `agent parameters' and leveraging the
fundamental idea of model training in artificial intelligence, we develop
AgentOptimizer that employs the LLM to update agents' functions and devise an
agent training algorithm with two strategies, roll-back, and early-stop, to
streamline the training process. With extensive experiments, we showcase that
the agent training paradigm could significantly improve the performance of
representative LLM agents in various downstream tasks. We also study the
behavior of the agent training regarding aspects like the learning curve and
domain transferability. | cs.AI | Offline Training of Language Model Agents with Functions as Learnable Weights |
2024-02-17T18:12:02Z | http://arxiv.org/pdf/2402.11355v3 | http://arxiv.org/abs/2402.11355v3 | Matan Avitan, Ryan Cotterell, Yoav Goldberg, Shauli Ravfogel | 2024-05-07T17:58:17Z | cs.CL, cs.CY, cs.LG | Interventions targeting the representation space of language models (LMs)
have emerged as an effective means to influence model behavior. Such methods
are employed, for example, to eliminate or alter the encoding of demographic
information such as gender within the model's representations and, in so doing,
create a counterfactual representation. However, because the intervention
operates within the representation space, understanding precisely what aspects
of the text it modifies poses a challenge. In this paper, we give a method to
convert representation counterfactuals into string counterfactuals. We
demonstrate that this approach enables us to analyze the linguistic alterations
corresponding to a given representation space intervention and to interpret the
features utilized to encode a specific concept. Moreover, the resulting
counterfactuals can be used to mitigate bias in classification through data
augmentation. | cs.CL | Natural Language Counterfactuals through Representation Surgery |
2024-02-17T18:05:53Z | http://arxiv.org/pdf/2402.11353v1 | http://arxiv.org/abs/2402.11353v1 | Eunkyung Jo, Yuin Jeong, SoHyun Park, Daniel A. Epstein, Young-Ho Kim | 2024-02-17T18:05:53Z | cs.HC, cs.AI, cs.CL, H.5.2; I.2.7 | Recent large language models (LLMs) offer the potential to support public
health monitoring by facilitating health disclosure through open-ended
conversations but rarely preserve the knowledge gained about individuals across
repeated interactions. Augmenting LLMs with long-term memory (LTM) presents an
opportunity to improve engagement and self-disclosure, but we lack an
understanding of how LTM impacts people's interaction with LLM-driven chatbots
in public health interventions. We examine the case of CareCall -- an
LLM-driven voice chatbot with LTM -- through the analysis of 1,252 call logs
and interviews with nine users. We found that LTM enhanced health disclosure
and fostered positive perceptions of the chatbot by offering familiarity.
However, we also observed challenges in promoting self-disclosure through LTM,
particularly around addressing chronic health conditions and privacy concerns.
We discuss considerations for LTM integration in LLM-driven chatbots for public
health monitoring, including carefully deciding what topics need to be
remembered in light of public health goals. | cs.HC | Understanding the Impact of Long-Term Memory on Self-Disclosure with Large Language Model-Driven Chatbots for Public Health Intervention |
2024-02-17T17:52:24Z | http://arxiv.org/pdf/2402.11349v2 | http://arxiv.org/abs/2402.11349v2 | Bruce W. Lee, JaeHyuk Lim | 2024-06-06T17:20:21Z | cs.CL, cs.AI | We argue that language-only models don't learn the physical manifestation of
language. We present an empirical investigation of visual-auditory properties
of language through a series of tasks, termed H-Test. These tasks highlight a
fundamental gap between human linguistic understanding and the sensory-deprived
linguistic understanding of LLMs. In support of our hypothesis, 1. deliberate
reasoning (Chain-of-Thought), 2. few-shot examples, or 3. stronger LLM from the
same model family (LLaMA 2 13B -> LLaMA 2 70B) has no significant effect on
H-Test performance.
We bring in the philosophical case of Mary, who learns about the world in a
sensory-deprived environment as a useful conceptual framework to understand how
language-only models learn about the world (Jackson, 1986). Our experiments
show that some of the strongest proprietary LLMs stay near random chance
baseline accuracy of 50%, highlighting the limitations of linguistic knowledge
acquired in the absence of sensory experience. Our code and data are available
at <github.com/brucewlee/h-test>. | cs.CL | Language Models Don't Learn the Physical Manifestation of Language |
2024-02-17T17:47:10Z | http://arxiv.org/pdf/2402.11347v1 | http://arxiv.org/abs/2402.11347v1 | Wendi Cui, Jiaxin Zhang, Zhuohang Li, Hao Sun, Damien Lopez, Kamalika Das, Bradley Malin, Sricharan Kumar | 2024-02-17T17:47:10Z | cs.CL | Crafting an ideal prompt for Large Language Models (LLMs) is a challenging
task that demands significant resources and expert human input. Existing work
treats the optimization of prompt instruction and in-context learning examples
as distinct problems, leading to sub-optimal prompt performance. This research
addresses this limitation by establishing a unified in-context prompt
optimization framework, which aims to achieve joint optimization of the prompt
instruction and examples. However, formulating such optimization in the
discrete and high-dimensional natural language space introduces challenges in
terms of convergence and computational efficiency. To overcome these issues, we
present PhaseEvo, an efficient automatic prompt optimization framework that
combines the generative capability of LLMs with the global search proficiency
of evolution algorithms. Our framework features a multi-phase design
incorporating innovative LLM-based mutation operators to enhance search
efficiency and accelerate convergence. We conduct an extensive evaluation of
our approach across 35 benchmark tasks. The results demonstrate that PhaseEvo
significantly outperforms the state-of-the-art baseline methods by a large
margin whilst maintaining good efficiency. | cs.CL | PhaseEvo: Towards Unified In-Context Prompt Optimization for Large Language Models |
2024-02-17T16:34:50Z | http://arxiv.org/pdf/2402.11324v1 | http://arxiv.org/abs/2402.11324v1 | Jiateng Liu, Pengfei Yu, Yuji Zhang, Sha Li, Zixuan Zhang, Heng Ji | 2024-02-17T16:34:50Z | cs.CL | The dynamic nature of real-world information necessitates efficient knowledge
editing (KE) in large language models (LLMs) for knowledge updating. However,
current KE approaches, which typically operate on (subject, relation, object)
triples, ignore the contextual information and the relation among different
knowledge. Such editing methods could thus encounter an uncertain editing
boundary, leaving a lot of relevant knowledge in ambiguity: Queries that could
be answered pre-edit cannot be reliably answered afterward. In this work, we
analyze this issue by introducing a theoretical framework for KE that
highlights an overlooked set of knowledge that remains unchanged and aids in
knowledge deduction during editing, which we name as the deduction anchor. We
further address this issue by proposing a novel task of event-based knowledge
editing that pairs facts with event descriptions. This task manifests not only
a closer simulation of real-world editing scenarios but also a more logically
sound setting, implicitly defining the deduction anchor to address the issue of
indeterminate editing boundaries. We empirically demonstrate the superiority of
event-based editing over the existing setting on resolving uncertainty in
edited models, and curate a new benchmark dataset EvEdit derived from the
CounterFact dataset. Moreover, while we observe that the event-based setting is
significantly challenging for existing approaches, we propose a novel approach
Self-Edit that showcases stronger performance, achieving 55.6% consistency
improvement while maintaining the naturalness of generation. | cs.CL | EVEDIT: Event-based Knowledge Editing with Deductive Editing Boundaries |
2024-02-17T14:37:38Z | http://arxiv.org/pdf/2402.11297v1 | http://arxiv.org/abs/2402.11297v1 | Husein Zolkepli, Aisyah Razak, Kamarul Adha, Ariff Nazhan | 2024-02-17T14:37:38Z | cs.CL | Our contribution introduces a groundbreaking multimodal large language model
designed to comprehend multi-images, multi-audio, and multi-images-multi-audio
within a single multiturn session. Leveraging state-of-the-art models, we
utilize the SigLIP encoder for visual inputs and the Whisper Encoder for audio
inputs. Notably, this multimodal large language model is bilingual, proficient
in understanding both English and Malay simultaneously. We proudly unveil two
versions of this model: TinyLlama with 1.1B parameters, and Mistral with 7B
parameters. With its ability to navigate diverse modalities and languages, our
model represents a significant advancement for the Malaysian context and
beyond.
All models released at
https://huggingface.co/collections/mesolitica/multimodal-malaysian-llm-65c6f893e03f78fa9e5c8859 | cs.CL | MMMModal -- Multi-Images Multi-Audio Multi-turn Multi-Modal |
2024-02-17T14:34:31Z | http://arxiv.org/pdf/2402.11296v1 | http://arxiv.org/abs/2402.11296v1 | Junlong Li, Fan Zhou, Shichao Sun, Yikai Zhang, Hai Zhao, Pengfei Liu | 2024-02-17T14:34:31Z | cs.CL, cs.AI | As a relative quality comparison of model responses, human and Large Language
Model (LLM) preferences serve as common alignment goals in model fine-tuning
and criteria in evaluation. Yet, these preferences merely reflect broad
tendencies, resulting in less explainable and controllable models with
potential safety risks. In this work, we dissect the preferences of human and
32 different LLMs to understand their quantitative composition, using
annotations from real-world user-model conversations for a fine-grained,
scenario-wise analysis. We find that humans are less sensitive to errors, favor
responses that support their stances, and show clear dislike when models admit
their limits. On the contrary, advanced LLMs like GPT-4-Turbo emphasize
correctness, clarity, and harmlessness more. Additionally, LLMs of similar
sizes tend to exhibit similar preferences, regardless of their training
methods, and fine-tuning for alignment does not significantly alter the
preferences of pretrained-only LLMs. Finally, we show that preference-based
evaluation can be intentionally manipulated. In both training-free and
training-based settings, aligning a model with the preferences of judges boosts
scores, while injecting the least preferred properties lowers them. This
results in notable score shifts: up to 0.59 on MT-Bench (1-10 scale) and 31.94
on AlpacaEval 2.0 (0-100 scale), highlighting the significant impact of this
strategic adaptation. Interactive Demo:
https://huggingface.co/spaces/GAIR/Preference-Dissection-Visualization Dataset:
https://huggingface.co/datasets/GAIR/preference-dissection Code:
https://github.com/GAIR-NLP/Preference-Dissection | cs.CL | Dissecting Human and LLM Preferences |
2024-02-17T14:26:57Z | http://arxiv.org/pdf/2402.11295v3 | http://arxiv.org/abs/2402.11295v3 | Yuzhuang Xu, Xu Han, Zonghan Yang, Shuo Wang, Qingfu Zhu, Zhiyuan Liu, Weidong Liu, Wanxiang Che | 2024-05-22T03:55:10Z | cs.CL | Model quantification uses low bit-width values to represent the weight
matrices of existing models to be quantized, which is a promising approach to
reduce both storage and computational overheads of deploying highly anticipated
LLMs. However, current quantization methods suffer severe performance
degradation when the bit-width is extremely reduced, and thus focus on
utilizing 4-bit or 8-bit values to quantize models. This paper boldly quantizes
the weight matrices of LLMs to 1-bit, paving the way for the extremely low
bit-width deployment of LLMs. For this target, we introduce a 1-bit model
compressing framework named OneBit, including a novel 1-bit parameter
representation method to better quantize LLMs as well as an effective parameter
initialization method based on matrix decomposition to improve the convergence
speed of the quantization framework. Sufficient experimental results indicate
that OneBit achieves good performance (at least 81% of the non-quantized
performance on LLaMA models) with robust training processes when only using
1-bit weight matrices. | cs.CL | OneBit: Towards Extremely Low-bit Large Language Models |
2024-02-17T14:19:38Z | http://arxiv.org/pdf/2402.11291v2 | http://arxiv.org/abs/2402.11291v2 | Panagiotis Giadikiaroglou, Maria Lymperaiou, Giorgos Filandrianos, Giorgos Stamou | 2024-04-20T07:04:28Z | cs.CL, cs.AI | Exploring the capabilities of Large Language Models (LLMs) in puzzle solving
unveils critical insights into their potential and challenges in AI, marking a
significant step towards understanding their applicability in complex reasoning
tasks. This survey leverages a unique taxonomy -- dividing puzzles into
rule-based and rule-less categories -- to critically assess LLMs through
various methodologies, including prompting techniques, neuro-symbolic
approaches, and fine-tuning. Through a critical review of relevant datasets and
benchmarks, we assess LLMs' performance, identifying significant challenges in
complex puzzle scenarios. Our findings highlight the disparity between LLM
capabilities and human-like reasoning, particularly in those requiring advanced
logical inference. The survey underscores the necessity for novel strategies
and richer datasets to advance LLMs' puzzle-solving proficiency and contribute
to AI's logical reasoning and creative problem-solving advancements. | cs.CL | Puzzle Solving using Reasoning of Large Language Models: A Survey |
2024-02-17T13:43:39Z | http://arxiv.org/pdf/2402.11282v1 | http://arxiv.org/abs/2402.11282v1 | Qihang Yang, Caimei Yang, Yu Liao, Ziman Zhuang | 2024-02-17T13:43:39Z | cs.CL | In several languages, omitting a verb phrase (VP) in double centre-embedded
structures creates a grammaticality illusion. Similar illusion also exhibited
in Mandarin missing-NP double centre-embedded structures. However, there is no
consensus on its very nature. Instead of treating it as grammaticality
illusion, we argue that ambiguous interpretations of verbs can best account for
this phenomenon in Mandarin. To further support this hypothesis, we conducted
two electroencephalography (EEG) experiments on quasi double centre-embedded
structures whose complexity is reduced by placing the self-embedding relative
clauses into the sentence's subject position. Experiment 1 showed that similar
phenomenon even exhibited in this structure, evidenced by an absence of P600
effect and a presence of N400 effect. In Experiment 2, providing semantic cues
to reduce ambiguity dispelled this illusion, as evidenced by a P600 effect. We
interpret the results under garden-path theory and propose that word-order
difference may account for this cross-linguistic variation. | cs.CL | Grammaticality illusion or ambiguous interpretation? Event-related potentials reveal the nature of the missing-NP effect in Mandarin centre-embedded structures |
2024-02-17T13:41:44Z | http://arxiv.org/pdf/2402.11281v3 | http://arxiv.org/abs/2402.11281v3 | Yixin Yang, Zheng Li, Qingxiu Dong, Heming Xia, Zhifang Sui | 2024-06-20T13:12:31Z | cs.CL | Understanding the deep semantics of images is essential in the era dominated
by social media. However, current research works primarily on the superficial
description of images, revealing a notable deficiency in the systematic
investigation of the inherent deep semantics. In this work, we introduce
DEEPEVAL, a comprehensive benchmark to assess Large Multimodal Models' (LMMs)
capacities of visual deep semantics. DEEPEVAL includes human-annotated dataset
and three progressive subtasks: fine-grained description selection, in-depth
title matching, and deep semantics understanding. Utilizing DEEPEVAL, we
evaluate 9 open-source LMMs and GPT-4V(ision). Our evaluation demonstrates a
substantial gap between the deep semantic comprehension capabilities of
existing LMMs and humans. For example, GPT-4V is 30% behind humans in
understanding deep semantics, even though it achieves human-comparable
performance in image description. Further analysis reveals that LMM performance
on DEEPEVAL varies according to the specific facets of deep semantics explored,
indicating the fundamental challenges remaining in developing LMMs. | cs.CL | Can Large Multimodal Models Uncover Deep Semantics Behind Images? |
2024-02-17T13:37:39Z | http://arxiv.org/pdf/2402.11279v1 | http://arxiv.org/abs/2402.11279v1 | Pei Wang, Yejie Wang, Muxi Diao, Keqing He, Guanting Dong, Weiran Xu | 2024-02-17T13:37:39Z | cs.CL, cs.AI | In the deployment of large language models (LLMs), accurate confidence
estimation is critical for assessing the credibility of model predictions.
However, existing methods often fail to overcome the issue of overconfidence on
incorrect answers. In this work, we focus on improving the confidence
estimation of large language models. Considering the fragility of
self-awareness in language models, we introduce a Multi-Perspective Consistency
(MPC) method. We leverage complementary insights from different perspectives
within models (MPC-Internal) and across different models (MPC-Across) to
mitigate the issue of overconfidence arising from a singular viewpoint. The
experimental results on eight publicly available datasets show that our MPC
achieves state-of-the-art performance. Further analyses indicate that MPC can
mitigate the problem of overconfidence and is effectively scalable to other
models. | cs.CL | Multi-Perspective Consistency Enhances Confidence Estimation in Large Language Models |
2024-02-17T13:02:54Z | http://arxiv.org/pdf/2402.11271v2 | http://arxiv.org/abs/2402.11271v2 | Shu Yang, Muhammad Asif Ali, Lu Yu, Lijie Hu, Di Wang | 2024-03-30T22:05:59Z | cs.CL, cs.CY, cs.HC | The increasing significance of large models and their multi-modal variants in
societal information processing has ignited debates on social safety and
ethics. However, there exists a paucity of comprehensive analysis for: (i) the
interactions between human and artificial intelligence systems, and (ii)
understanding and addressing the associated limitations. To bridge this gap, we
propose Model Autophagy Analysis (MONAL) for large models' self-consumption
explanation. MONAL employs two distinct autophagous loops (referred to as
``self-consumption loops'') to elucidate the suppression of human-generated
information in the exchange between human and AI systems. Through comprehensive
experiments on diverse datasets, we evaluate the capacities of generated models
as both creators and disseminators of information. Our key findings reveal (i)
A progressive prevalence of model-generated synthetic information over time
within training datasets compared to human-generated information; (ii) The
discernible tendency of large models, when acting as information transmitters
across multiple iterations, to selectively modify or prioritize specific
contents; and (iii) The potential for a reduction in the diversity of socially
or human-generated information, leading to bottlenecks in the performance
enhancement of large models and confining them to local optima. | cs.CL | MONAL: Model Autophagy Analysis for Modeling Human-AI Interactions |
2024-02-17T12:25:31Z | http://arxiv.org/pdf/2402.11260v1 | http://arxiv.org/abs/2402.11260v1 | Shu Yang, Muhammad Asif Ali, Cheng-Long Wang, Lijie Hu, Di Wang | 2024-02-17T12:25:31Z | cs.CL, cs.AI | Adapting large language models (LLMs) to new domains/tasks and enabling them
to be efficient lifelong learners is a pivotal challenge. In this paper, we
propose MoRAL, i.e., Mixture-of-Experts augmented Low-Rank Adaptation for
Lifelong Learning. MoRAL combines the multi-tasking abilities of MoE with the
fine-tuning abilities of LoRA for effective life-long learning of LLMs. In
contrast to the conventional approaches that use factual triplets as inputs
MoRAL relies on simple question-answer pairs, which is a more practical and
effective strategy for robust and efficient learning. Owing to new data
settings, we introduce a new evaluation benchmark namely: Life Long Learning of
LLM (5L-bench) encompassing a newly curated dataset of question-answer pairs,
and a set of evaluation metrics for rigorous evaluation of MoRAL in open-book
and closed-book settings. Experimental evaluation shows (i) LLMs learn fast in
open-book settings with up to 30.15% improvement in "RA" for Phi-2-2.7B
compared to closed-book (for models fine-tuned with MoRAL); (ii) MoRAL shows
higher performance improvement for models with a greater number of parameters;
(iii) MoRAL is robust to catastrophic forgetting offering better knowledge
retention compared to baselines. | cs.CL | MoRAL: MoE Augmented LoRA for LLMs' Lifelong Learning |
2024-02-17T11:28:08Z | http://arxiv.org/pdf/2402.11254v2 | http://arxiv.org/abs/2402.11254v2 | Ying Mo, Jiahao Liu, Jian Yang, Qifan Wang, Shun Zhang, Jingang Wang, Zhoujun Li | 2024-06-24T08:34:56Z | cs.CL | There has been increasing interest in exploring the capabilities of advanced
large language models (LLMs) in the field of information extraction (IE),
specifically focusing on tasks related to named entity recognition (NER) and
relation extraction (RE). Although researchers are exploring the use of
few-shot information extraction through in-context learning with LLMs, they
tend to focus only on using correct or positive examples for demonstration,
neglecting the potential value of incorporating incorrect or negative examples
into the learning process. In this paper, we present c-ICL, a novel few-shot
technique that leverages both correct and incorrect sample constructions to
create in-context learning demonstrations. This approach enhances the ability
of LLMs to extract entities and relations by utilizing prompts that incorporate
not only the positive samples but also the reasoning behind them. This method
allows for the identification and correction of potential interface errors.
Specifically, our proposed method taps into the inherent contextual information
and valuable information in hard negative samples and the nearest positive
neighbors to the test and then applies the in-context learning demonstrations
based on LLMs. Our experiments on various datasets indicate that c-ICL
outperforms previous few-shot in-context learning methods, delivering
substantial enhancements in performance across a broad spectrum of related
tasks. These improvements are noteworthy, showcasing the versatility of our
approach in miscellaneous scenarios. | cs.CL | C-ICL: Contrastive In-context Learning for Information Extraction |
2024-02-17T11:25:26Z | http://arxiv.org/pdf/2402.11253v3 | http://arxiv.org/abs/2402.11253v3 | Sangkyu Lee, Sungdong Kim, Ashkan Yousefpour, Minjoon Seo, Kang Min Yoo, Youngjae Yu | 2024-06-25T13:39:52Z | cs.LG, cs.AI, cs.CL | Existing approaches for aligning large language models with human preferences
face a trade-off that requires a separate reward model (RM) for on-policy
learning. In this paper, we present a novel alignment framework, SELF-JUDGE
that (1) does on-policy learning and 2) is parameter efficient, as it does not
require an additional RM for evaluating the samples for on-policy learning. To
this end, we propose Judge-augmented Supervised Fine-Tuning (JSFT) to train a
single model to act as both a policy and a judge. Specifically, we view the
pairwise judgment task, choosing the better response from a response pair, as a
special case of the instruction-following task. The resulting model can judge
preferences of on-the-fly responses from current policy initialized from
itself. Experimental results show the efficacy of SELF-JUDGE, outperforming
baselines in preference benchmarks. We also show that the rejecting sampling by
itself can improve performance further without an additional evaluator. | cs.LG | Aligning Large Language Models by On-Policy Self-Judgment |
2024-02-17T11:18:22Z | http://arxiv.org/pdf/2402.11251v1 | http://arxiv.org/abs/2402.11251v1 | Siyin Wang, Shimin Li, Tianxiang Sun, Jinlan Fu, Qinyuan Cheng, Jiasheng Ye, Junjie Ye, Xipeng Qiu, Xuanjing Huang | 2024-02-17T11:18:22Z | cs.CL | In the realm of Large Language Models (LLMs), users commonly employ diverse
decoding strategies and adjust hyperparameters to control the generated text.
However, a critical question emerges: Are LLMs conscious of the existence of
these decoding strategies and capable of regulating themselves? The current
decoding generation process often relies on empirical and heuristic manual
adjustments to hyperparameters based on types of tasks and demands. However,
this process is typically cumbersome, and the decoding hyperparameters may not
always be optimal for each sample. To address the aforementioned challenges, we
propose a novel text generation paradigm termed Hyperparameter Aware Generation
(HAG). By leveraging hyperparameter-aware instruction tuning, the LLM
autonomously determines the optimal decoding strategy and configs based on the
input samples, enabling self-regulation. Our approach eliminates the need for
extensive manual tuning, offering a more autonomous, self-regulate model
behavior. Experimental results spanning six datasets across reasoning,
creativity, translation, and mathematics tasks demonstrate that
hyperparameter-aware instruction tuning empowers the LLMs to self-regulate the
decoding strategy and hyperparameter. HAG extends the current paradigm in the
text generation process, highlighting the feasibility of endowing the LLMs with
self-regulate decoding strategies. | cs.CL | LLM can Achieve Self-Regulation via Hyperparameter Aware Generation |
2024-02-17T10:37:51Z | http://arxiv.org/pdf/2402.11243v1 | http://arxiv.org/abs/2402.11243v1 | Deniz Gorur, Antonio Rago, Francesca Toni | 2024-02-17T10:37:51Z | cs.CL, cs.AI, I.2.7 | Argument mining (AM) is the process of automatically extracting arguments,
their components and/or relations amongst arguments and components from text.
As the number of platforms supporting online debate increases, the need for AM
becomes ever more urgent, especially in support of downstream tasks.
Relation-based AM (RbAM) is a form of AM focusing on identifying agreement
(support) and disagreement (attack) relations amongst arguments. RbAM is a
challenging classification task, with existing methods failing to perform
satisfactorily. In this paper, we show that general-purpose Large Language
Models (LLMs), appropriately primed and prompted, can significantly outperform
the best performing (RoBERTa-based) baseline. Specifically, we experiment with
two open-source LLMs (Llama-2 and Mistral) with ten datasets. | cs.CL | Can Large Language Models perform Relation-based Argument Mining? |
2024-02-17T10:32:43Z | http://arxiv.org/pdf/2403.05561v1 | http://arxiv.org/abs/2403.05561v1 | Claire S. Lee, Noelle Lim, Michael Guerzhoy | 2024-02-17T10:32:43Z | cs.CY, cs.CL | We present a novel task that can elucidate the connection between anxiety and
ADHD; use Transformers to make progress toward solving a task that is not
solvable by keyword-based classifiers; and discuss a method for visualization
of our classifier illuminating the connection between anxiety and ADHD
presentations.
Up to approximately 50% of adults with ADHD may also have an anxiety disorder
and approximately 30\% of adults with anxiety may also have ADHD. Patients
presenting with anxiety may be treated for anxiety without ADHD ever being
considered, possibly affecting treatment. We show how data that bears on ADHD
that is comorbid with anxiety can be obtained from social media data, and show
that Transformers can be used to detect a proxy for possible comorbid ADHD in
people with anxiety symptoms.
We collected data from anxiety and ADHD online forums (subreddits). We
identified posters who first started posting in the Anxiety subreddit and later
started posting in the ADHD subreddit as well. We use this subset of the
posters as a proxy for people who presented with anxiety symptoms and then
became aware that they might have ADHD. We fine-tune a Transformer
architecture-based classifier to classify people who started posting in the
Anxiety subreddit and then started posting in the ADHD subreddit vs. people who
posted in the Anxiety subreddit without later posting in the ADHD subreddit. We
show that a Transformer architecture is capable of achieving reasonable results
(76% correct for RoBERTa vs. under 60% correct for the best keyword-based
model, both with 50% base rate). | cs.CY | Detecting a Proxy for Potential Comorbid ADHD in People Reporting Anxiety Symptoms from Social Media Data |
2024-02-17T08:14:37Z | http://arxiv.org/pdf/2402.11218v2 | http://arxiv.org/abs/2402.11218v2 | Xun Liang, Hanyu Wang, Shichao Song, Mengting Hu, Xunzhi Wang, Zhiyu Li, Feiyu Xiong, Bo Tang | 2024-05-24T04:56:41Z | cs.CL | Controlled Text Generation (CTG) aims to produce texts that exhibit specific
desired attributes. In this study, we introduce a pluggable CTG framework for
Large Language Models (LLMs) named Dynamic Attribute Graphs-based controlled
text generation (DATG). This framework utilizes an attribute scorer to evaluate
the attributes of sentences generated by LLMs and constructs dynamic attribute
graphs. DATG modulates the occurrence of key attribute words and key
anti-attribute words, achieving effective attribute control without
compromising the original capabilities of the model. We conduct experiments
across four datasets in two tasks: toxicity mitigation and sentiment
transformation, employing five LLMs as foundational models. Our findings
highlight a remarkable enhancement in control accuracy, achieving a peak
improvement of 19.29% over baseline methods in the most favorable task across
four datasets. Additionally, we observe a significant decrease in perplexity,
markedly improving text fluency. | cs.CL | Controlled Text Generation for Large Language Model with Dynamic Attribute Graphs |
2024-02-17T08:04:23Z | http://arxiv.org/pdf/2402.11217v1 | http://arxiv.org/abs/2402.11217v1 | Wenxuan Wang, Yihang Su, Jingyuan Huan, Jie Liu, Wenting Chen, Yudi Zhang, Cheng-Yi Li, Kao-Jung Chang, Xiaohan Xin, Linlin Shen, Michael R. Lyu | 2024-02-17T08:04:23Z | cs.CL, cs.CV | The significant breakthroughs of Medical Multi-Modal Large Language Models
(Med-MLLMs) renovate modern healthcare with robust information synthesis and
medical decision support. However, these models are often evaluated on
benchmarks that are unsuitable for the Med-MLLMs due to the intricate nature of
the real-world diagnostic frameworks, which encompass diverse medical
specialties and involve complex clinical decisions. Moreover, these benchmarks
are susceptible to data leakage, since Med-MLLMs are trained on large
assemblies of publicly available data. Thus, an isolated and clinically
representative benchmark is highly desirable for credible Med-MLLMs evaluation.
To this end, we introduce Asclepius, a novel Med-MLLM benchmark that rigorously
and comprehensively assesses model capability in terms of: distinct medical
specialties (cardiovascular, gastroenterology, etc.) and different diagnostic
capacities (perception, disease analysis, etc.). Grounded in 3 proposed core
principles, Asclepius ensures a comprehensive evaluation by encompassing 15
medical specialties, stratifying into 3 main categories and 8 sub-categories of
clinical tasks, and exempting from train-validate contamination. We further
provide an in-depth analysis of 6 Med-MLLMs and compare them with 5 human
specialists, providing insights into their competencies and limitations in
various medical contexts. Our work not only advances the understanding of
Med-MLLMs' capabilities but also sets a precedent for future evaluations and
the safe deployment of these models in clinical environments. We launch and
maintain a leaderboard for community assessment of Med-MLLM capabilities
(https://asclepius-med.github.io/). | cs.CL | Asclepius: A Spectrum Evaluation Benchmark for Medical Multi-Modal Large Language Models |
2024-02-17T06:48:45Z | http://arxiv.org/pdf/2402.11208v1 | http://arxiv.org/abs/2402.11208v1 | Wenkai Yang, Xiaohan Bi, Yankai Lin, Sishuo Chen, Jie Zhou, Xu Sun | 2024-02-17T06:48:45Z | cs.CR, cs.AI, cs.CL | Leveraging the rapid development of Large Language Models LLMs, LLM-based
agents have been developed to handle various real-world applications, including
finance, healthcare, and shopping, etc. It is crucial to ensure the reliability
and security of LLM-based agents during applications. However, the safety
issues of LLM-based agents are currently under-explored. In this work, we take
the first step to investigate one of the typical safety threats, backdoor
attack, to LLM-based agents. We first formulate a general framework of agent
backdoor attacks, then we present a thorough analysis on the different forms of
agent backdoor attacks. Specifically, from the perspective of the final
attacking outcomes, the attacker can either choose to manipulate the final
output distribution, or only introduce malicious behavior in the intermediate
reasoning process, while keeping the final output correct. Furthermore, the
former category can be divided into two subcategories based on trigger
locations: the backdoor trigger can be hidden either in the user query or in an
intermediate observation returned by the external environment. We propose the
corresponding data poisoning mechanisms to implement the above variations of
agent backdoor attacks on two typical agent tasks, web shopping and tool
utilization. Extensive experiments show that LLM-based agents suffer severely
from backdoor attacks, indicating an urgent need for further research on the
development of defenses against backdoor attacks on LLM-based agents. Warning:
This paper may contain biased content. | cs.CR | Watch Out for Your Agents! Investigating Backdoor Threats to LLM-Based Agents |
2024-02-17T06:23:27Z | http://arxiv.org/pdf/2402.12399v2 | http://arxiv.org/abs/2402.12399v2 | Zhiyuan Zeng, Qipeng Guo, Zhaoye Fei, Zhangyue Yin, Yunhua Zhou, Linyang Li, Tianxiang Sun, Hang Yan, Dahua Lin, Xipeng Qiu | 2024-02-21T13:33:12Z | cs.LG, cs.AI, cs.CL | Sparse Mixture of Experts (MoE) models are popular for training large
language models due to their computational efficiency. However, the commonly
used top-$k$ routing mechanism suffers from redundancy computation and memory
costs due to the unbalanced routing. Some experts are overflow, where the
exceeding tokens are dropped. While some experts are vacant, which are padded
with zeros, negatively impacting model performance. To address the dropped
tokens and padding, we propose the Rectify-Router, comprising the Intra-GPU
Rectification and the Fill-in Rectification. The Intra-GPU Rectification
handles dropped tokens, efficiently routing them to experts within the GPU
where they are located to avoid inter-GPU communication. The Fill-in
Rectification addresses padding by replacing padding tokens with the tokens
that have high routing scores. Our experimental results demonstrate that the
Intra-GPU Rectification and the Fill-in Rectification effectively handle
dropped tokens and padding, respectively. Furthermore, the combination of them
achieves superior performance, surpassing the accuracy of the vanilla top-1
router by 4.7%. | cs.LG | Turn Waste into Worth: Rectifying Top-$k$ Router of MoE |
2024-02-17T05:44:40Z | http://arxiv.org/pdf/2402.11203v1 | http://arxiv.org/abs/2402.11203v1 | Yizheng Huang, Jimmy Huang | 2024-02-17T05:44:40Z | cs.IR, cs.AI, cs.CL, cs.LG | The rapid advancement of artificial intelligence (AI) has highlighted ChatGPT
as a pivotal technology in the field of information retrieval (IR).
Distinguished from its predecessors, ChatGPT offers significant benefits that
have attracted the attention of both the industry and academic communities.
While some view ChatGPT as a groundbreaking innovation, others attribute its
success to the effective integration of product development and market
strategies. The emergence of ChatGPT, alongside GPT-4, marks a new phase in
Generative AI, generating content that is distinct from training examples and
exceeding the capabilities of the prior GPT-3 model by OpenAI. Unlike the
traditional supervised learning approach in IR tasks, ChatGPT challenges
existing paradigms, bringing forth new challenges and opportunities regarding
text quality assurance, model bias, and efficiency. This paper seeks to examine
the impact of ChatGPT on IR tasks and offer insights into its potential future
developments. | cs.IR | Exploring ChatGPT for Next-generation Information Retrieval: Opportunities and Challenges |
2024-02-17T05:22:56Z | http://arxiv.org/pdf/2402.11199v2 | http://arxiv.org/abs/2402.11199v2 | Minh-Vuong Nguyen, Linhao Luo, Fatemeh Shiri, Dinh Phung, Yuan-Fang Li, Thuy-Trang Vu, Gholamreza Haffari | 2024-06-19T05:14:05Z | cs.CL | Large language models (LLMs) demonstrate strong reasoning abilities when
prompted to generate chain-of-thought (CoT) explanations alongside answers.
However, previous research on evaluating LLMs has solely focused on answer
accuracy, neglecting the correctness of the generated CoT. In this paper, we
delve deeper into the CoT reasoning capabilities of LLMs in multi-hop question
answering by utilizing knowledge graphs (KGs). We propose a novel
discriminative and generative CoT evaluation paradigm to assess LLMs' knowledge
of reasoning and the accuracy of the generated CoT. Through experiments
conducted on 5 different families of LLMs across 2 multi-hop question-answering
datasets, we find that LLMs possess sufficient knowledge to perform reasoning.
However, there exists a significant disparity between answer accuracy and
faithfulness of the CoT reasoning generated by LLMs, indicating that they often
arrive at correct answers through incorrect reasoning. | cs.CL | Direct Evaluation of Chain-of-Thought in Multi-hop Reasoning with Knowledge Graphs |
2024-02-17T05:15:12Z | http://arxiv.org/pdf/2402.11197v2 | http://arxiv.org/abs/2402.11197v2 | Hiroyuki Deguchi, Yusuke Sakai, Hidetaka Kamigaito, Taro Watanabe, Hideki Tanaka, Masao Utiyama | 2024-06-11T07:14:46Z | cs.CL | Minimum Bayes risk (MBR) decoding achieved state-of-the-art translation
performance by using COMET, a neural metric that has a high correlation with
human evaluation. However, MBR decoding requires quadratic time since it
computes the expected score between a translation hypothesis and all reference
translations. We propose centroid-based MBR (CBMBR) decoding to improve the
speed of MBR decoding. Our method clusters the reference translations in the
feature space, and then calculates the score using the centroids of each
cluster. The experimental results show that our CBMBR not only improved the
decoding speed of the expected score calculation 5.7 times, but also
outperformed vanilla MBR decoding in translation quality by up to 0.5 COMET in
the WMT'22 En$\leftrightarrow$Ja, En$\leftrightarrow$De, En$\leftrightarrow$Zh,
and WMT'23 En$\leftrightarrow$Ja translation tasks. | cs.CL | Centroid-Based Efficient Minimum Bayes Risk Decoding |
2024-02-17T05:10:18Z | http://arxiv.org/pdf/2402.11194v2 | http://arxiv.org/abs/2402.11194v2 | Pragya Srivastava, Manuj Malik, Vivek Gupta, Tanuja Ganu, Dan Roth | 2024-02-29T09:13:58Z | cs.CL | Large Language Models (LLMs), excel in natural language understanding, but
their capability for complex mathematical reasoning with an amalgamation of
structured tables and unstructured text is uncertain. This study explores LLMs'
mathematical reasoning on four financial tabular question-answering datasets:
TATQA, FinQA, ConvFinQA, and Multihiertt. Through extensive experiments with
various models and prompting techniques, we assess how LLMs adapt to complex
tables and mathematical tasks. We focus on sensitivity to table complexity and
performance variations with an increasing number of arithmetic reasoning steps.
The results provide insights into LLMs' capabilities and limitations in
handling complex mathematical scenarios for semi-structured tables. Ultimately,
we introduce a novel prompting technique tailored to semi-structured documents,
matching or outperforming other baselines in performance while providing a
nuanced understanding of LLMs abilities for such a task. | cs.CL | Evaluating LLMs' Mathematical Reasoning in Financial Document Question Answering |
2024-02-17T05:05:31Z | http://arxiv.org/pdf/2402.11192v2 | http://arxiv.org/abs/2402.11192v2 | Xuan Ren, Biao Wu, Lingqiao Liu | 2024-06-01T03:36:23Z | cs.CL, cs.AI | This paper explores an intriguing observation: fine-tuning a large language
model (LLM) with responses generated by a LLM often yields better results than
using responses generated by humans. We conduct an in-depth investigation to
understand why this occurs. Contrary to the common belief that these instances
is simply due to the more detailed nature of LLM-generated content, our study
identifies another contributing factor: an LLM is inherently more "familiar"
with LLM generated responses. This familiarity is evidenced by lower perplexity
before fine-tuning. We design a series of experiments to understand the impact
of the "familiarity" and our conclusion reveals that this "familiarity"
significantly impacts learning performance. Training with LLM-generated
responses not only enhances performance but also helps maintain the model's
capabilities in other tasks after fine-tuning on a specific task. | cs.CL | I Learn Better If You Speak My Language: Understanding the Superior Performance of Fine-Tuning Large Language Models with LLM-Generated Responses |
2024-02-17T04:54:58Z | http://arxiv.org/pdf/2402.11191v1 | http://arxiv.org/abs/2402.11191v1 | Yang Cao, Xinyi Chen, Xin Zhang, Siying Li | 2024-02-17T04:54:58Z | cs.CL | In this paper, we present a novel method for automatically generating sports
news, which employs a unique algorithm that extracts pivotal moments from live
text broadcasts and uses them to create an initial draft of the news. This
draft is further refined by incorporating key details and background
information from a specially designed sports knowledge graph. This graph
contains 5,893 entities, which are classified into three distinct conceptual
categories, interconnected through four relationship types, and characterized
by 27 unique attributes. In addition, we create a multi-stage learning model by
combining convolutional neural networks and a transformer encoder. This model
expresses entity-task interactions using convolutional neural networks and
enriches entity representations in the query set with the transformer encoder.
It also includes a processor to compute matching scores for incomplete triples,
addressing few-shot knowledge graph completion problem. The efficiency of this
approach has been confirmed through both subjective and objective evaluations
of 50 selected test cases, demonstrating its capability in revolutionizing the
creation of sports news. | cs.CL | Knowledge Graph Assisted Automatic Sports News Writing |
2024-02-17T04:48:55Z | http://arxiv.org/pdf/2402.11190v1 | http://arxiv.org/abs/2402.11190v1 | Xiangjue Dong, Yibo Wang, Philip S. Yu, James Caverlee | 2024-02-17T04:48:55Z | cs.CL | Large Language Models (LLMs) can generate biased responses. Yet previous
direct probing techniques contain either gender mentions or predefined gender
stereotypes, which are challenging to comprehensively collect. Hence, we
propose an indirect probing framework based on conditional generation. This
approach aims to induce LLMs to disclose their gender bias even without
explicit gender or stereotype mentions. We explore three distinct strategies to
disclose explicit and implicit gender bias in LLMs. Our experiments demonstrate
that all tested LLMs exhibit explicit and/or implicit gender bias, even when
gender stereotypes are not present in the inputs. In addition, an increased
model size or model alignment amplifies bias in most cases. Furthermore, we
investigate three methods to mitigate bias in LLMs via Hyperparameter Tuning,
Instruction Guiding, and Debias Tuning. Remarkably, these methods prove
effective even in the absence of explicit genders or stereotypes. | cs.CL | Disclosure and Mitigation of Gender Bias in LLMs |
2024-02-17T04:16:30Z | http://arxiv.org/pdf/2402.11187v1 | http://arxiv.org/abs/2402.11187v1 | Yifei Yang, Zouying Cao, Hai Zhao | 2024-02-17T04:16:30Z | cs.CL, cs.AI | Large language models (LLMs) based on transformer are witnessing a notable
trend of size expansion, which brings considerable costs to both model training
and inference. However, existing methods such as model quantization, knowledge
distillation, and model pruning are constrained by various issues, including
hardware support limitations, the need for extensive training, and alterations
to the internal structure of the model. In this paper, we propose a concise
layer-wise pruning method called \textit{Layer Collapse (LaCo)}, in which rear
model layers collapse into a prior layer, enabling a rapid reduction in model
size while preserving the model structure. Comprehensive experiments show that
our method maintains an average task performance of over 80\% at pruning ratios
of 25-30\%, significantly outperforming existing state-of-the-art structured
pruning methods. We also conduct post-training experiments to confirm that the
proposed pruning method effectively inherits the parameters of the original
model. Finally, we discuss our motivation from the perspective of layer-wise
similarity and evaluate the performance of the pruned LLMs across various
pruning ratios. | cs.CL | LaCo: Large Language Model Pruning via Layer Collapse |
2024-02-17T03:13:42Z | http://arxiv.org/pdf/2402.11178v1 | http://arxiv.org/abs/2402.11178v1 | Haolan Zhan, Zhuang Li, Xiaoxi Kang, Tao Feng, Yuncheng Hua, Lizhen Qu, Yi Ying, Mei Rianto Chandra, Kelly Rosalin, Jureynolds Jureynolds, Suraj Sharma, Shilin Qu, Linhao Luo, Lay-Ki Soon, Zhaleh Semnani Azad, Ingrid Zukerman, Gholamreza Haffari | 2024-02-17T03:13:42Z | cs.CL | Norm violations occur when individuals fail to conform to culturally accepted
behaviors, which may lead to potential conflicts. Remediating norm violations
requires social awareness and cultural sensitivity of the nuances at play. To
equip interactive AI systems with a remediation ability, we offer ReNoVi - a
large-scale corpus of 9,258 multi-turn dialogues annotated with social norms,
as well as define a sequence of tasks to help understand and remediate norm
violations step by step. ReNoVi consists of two parts: 512 human-authored
dialogues (real data), and 8,746 synthetic conversations generated by ChatGPT
through prompt learning. While collecting sufficient human-authored data is
costly, synthetic conversations provide suitable amounts of data to help
mitigate the scarcity of training data, as well as the chance to assess the
alignment between LLMs and humans in the awareness of social norms. We thus
harness the power of ChatGPT to generate synthetic training data for our task.
To ensure the quality of both human-authored and synthetic data, we follow a
quality control protocol during data collection. Our experimental results
demonstrate the importance of remediating norm violations in socio-cultural
conversations, as well as the improvement in performance obtained from
synthetic data. | cs.CL | RENOVI: A Benchmark Towards Remediating Norm Violations in Socio-Cultural Conversations |
2024-02-17T02:55:35Z | http://arxiv.org/pdf/2402.11177v1 | http://arxiv.org/abs/2402.11177v1 | Huaiyuan Ying, Sheng Yu | 2024-02-17T02:55:35Z | cs.CL, cs.IR | Electronic health records (EHRs) hold significant value for research and
applications. As a new way of information extraction, question answering (QA)
can extract more flexible information than conventional methods and is more
accessible to clinical researchers, but its progress is impeded by the scarcity
of annotated data. In this paper, we propose a novel approach that
automatically generates training data for transfer learning of QA models. Our
pipeline incorporates a preprocessing module to handle challenges posed by
extraction types that are not readily compatible with extractive QA frameworks,
including cases with discontinuous answers and many-to-one relationships. The
obtained QA model exhibits excellent performance on subtasks of information
extraction in EHRs, and it can effectively handle few-shot or zero-shot
settings involving yes-no questions. Case studies and ablation studies
demonstrate the necessity of each component in our design, and the resulting
model is deemed suitable for practical use. | cs.CL | A Question Answering Based Pipeline for Comprehensive Chinese EHR Information Extraction |
2024-02-17T02:54:32Z | http://arxiv.org/pdf/2402.11176v2 | http://arxiv.org/abs/2402.11176v2 | Yougang Lyu, Lingyong Yan, Shuaiqiang Wang, Haibo Shi, Dawei Yin, Pengjie Ren, Zhumin Chen, Maarten de Rijke, Zhaochun Ren | 2024-04-17T11:45:00Z | cs.CL, cs.AI | Despite their success at many natural language processing (NLP) tasks, large
language models still struggle to effectively leverage knowledge for
knowledge-intensive tasks, manifesting limitations such as generating
incomplete, non-factual, or illogical answers. These limitations stem from
inadequate knowledge awareness of LLMs during vanilla fine-tuning. To address
these problems, we propose a knowledge-aware fine-tuning (KnowTuning) method to
improve fine-grained and coarse-grained knowledge awareness of LLMs. We devise
a fine-grained knowledge augmentation stage to train LLMs to identify difficult
fine-grained knowledge in answers. We also propose a coarse-grained knowledge
comparison stage to train LLMs to distinguish between reliable and unreliable
knowledge, in three aspects: completeness, factuality, and logicality.
Extensive experiments on both generic and medical question answering (QA)
datasets confirm the effectiveness of KnowTuning, through automatic and human
evaluations, across various sizes of LLMs. We further verify that KnowTuning
generates more facts with less factual error rate under fine-grained facts
evaluation. | cs.CL | KnowTuning: Knowledge-aware Fine-tuning for Large Language Models |
2024-02-17T02:50:33Z | http://arxiv.org/pdf/2402.11175v2 | http://arxiv.org/abs/2402.11175v2 | Yuxia Wang, Jonibek Mansurov, Petar Ivanov, Jinyan Su, Artem Shelmanov, Akim Tsvigun, Osama Mohanned Afzal, Tarek Mahmoud, Giovanni Puccetti, Thomas Arnold, Alham Fikri Aji, Nizar Habash, Iryna Gurevych, Preslav Nakov | 2024-06-27T05:42:12Z | cs.CL | The advent of Large Language Models (LLMs) has brought an unprecedented surge
in machine-generated text (MGT) across diverse channels. This raises legitimate
concerns about its potential misuse and societal implications. The need to
identify and differentiate such content from genuine human-generated text is
critical in combating disinformation, preserving the integrity of education and
scientific fields, and maintaining trust in communication. In this work, we
address this problem by introducing a new benchmark based on a multilingual,
multi-domain, and multi-generator corpus of MGTs -- M4GT-Bench. The benchmark
is compiled of three tasks: (1) mono-lingual and multi-lingual binary MGT
detection; (2) multi-way detection where one need to identify, which particular
model generated the text; and (3) mixed human-machine text detection, where a
word boundary delimiting MGT from human-written content should be determined.
On the developed benchmark, we have tested several MGT detection baselines and
also conducted an evaluation of human performance. We see that obtaining good
performance in MGT detection usually requires an access to the training data
from the same domain and generators. The benchmark is available at
https://github.com/mbzuai-nlp/M4GT-Bench. | cs.CL | M4GT-Bench: Evaluation Benchmark for Black-Box Machine-Generated Text Detection |
2024-02-17T02:25:57Z | http://arxiv.org/pdf/2402.11167v1 | http://arxiv.org/abs/2402.11167v1 | Fan Huang, Haewoon Kwak, Jisun An | 2024-02-17T02:25:57Z | cs.CL, cs.AI | The robustness of AI-content detection models against cultivated attacks
(e.g., paraphrasing or word switching) remains a significant concern. This
study proposes a novel token-ensemble generation strategy to challenge the
robustness of current AI-content detection approaches. We explore the ensemble
attack strategy by completing the prompt with the next token generated from
random candidate LLMs. We find the token-ensemble approach significantly drops
the performance of AI-content detection models (The code and test sets will be
released). Our findings reveal that token-ensemble generation poses a vital
challenge to current detection models and underlines the need for advancing
detection technologies to counter sophisticated adversarial strategies. | cs.CL | Token-Ensemble Text Generation: On Attacking the Automatic AI-Generated Text Detection |
2024-02-17T02:21:44Z | http://arxiv.org/pdf/2402.11166v1 | http://arxiv.org/abs/2402.11166v1 | Jian Wu, Linyi Yang, Yuliang Ji, Wenhao Huang, Börje F. Karlsson, Manabu Okumura | 2024-02-17T02:21:44Z | cs.CL | Multi-hop QA (MHQA) involves step-by-step reasoning to answer complex
questions and find multiple relevant supporting facts. However, Existing large
language models'(LLMs) reasoning ability in multi-hop question answering
remains exploration, which is inadequate in answering multi-hop questions.
Moreover, it is unclear whether LLMs follow a desired reasoning chain to reach
the right final answer. In this paper, we propose a \textbf{gen}erative
question \textbf{dec}omposition method (GenDec) from the perspective of
explainable QA by generating independent and complete sub-questions based on
incorporating additional extracted evidence for enhancing LLMs' reasoning
ability in RAG. To demonstrate the impact, generalization, and robustness of
Gendec, we conduct two experiments, the first is combining GenDec with small QA
systems on paragraph retrieval and QA tasks. We secondly examine the reasoning
capabilities of various state-of-the-art LLMs including GPT-4 and GPT-3.5
combined with GenDec. We experiment on the HotpotQA, 2WikihopMultiHopQA,
MuSiQue, and PokeMQA datasets. | cs.CL | GenDec: A robust generative Question-decomposition method for Multi-hop reasoning |
2024-02-17T02:07:49Z | http://arxiv.org/pdf/2402.11163v1 | http://arxiv.org/abs/2402.11163v1 | Jinhao Jiang, Kun Zhou, Wayne Xin Zhao, Yang Song, Chen Zhu, Hengshu Zhu, Ji-Rong Wen | 2024-02-17T02:07:49Z | cs.CL | In this paper, we aim to improve the reasoning ability of large language
models (LLMs) over knowledge graphs (KGs) to answer complex questions. Inspired
by existing methods that design the interaction strategy between LLMs and KG,
we propose an autonomous LLM-based agent framework, called KG-Agent, which
enables a small LLM to actively make decisions until finishing the reasoning
process over KGs. In KG-Agent, we integrate the LLM, multifunctional toolbox,
KG-based executor, and knowledge memory, and develop an iteration mechanism
that autonomously selects the tool then updates the memory for reasoning over
KG. To guarantee the effectiveness, we leverage program language to formulate
the multi-hop reasoning process over the KG, and synthesize a code-based
instruction dataset to fine-tune the base LLM. Extensive experiments
demonstrate that only using 10K samples for tuning LLaMA-7B can outperform
state-of-the-art methods using larger LLMs or more data, on both in-domain and
out-domain datasets. Our code and data will be publicly released. | cs.CL | KG-Agent: An Efficient Autonomous Agent Framework for Complex Reasoning over Knowledge Graph |
2024-02-17T01:56:19Z | http://arxiv.org/pdf/2402.11161v1 | http://arxiv.org/abs/2402.11161v1 | Zongxia Li, Ishani Mondal, Yijun Liang, Huy Nghiem, Jordan Lee Boyd-Graber | 2024-02-17T01:56:19Z | cs.CL, cs.AI | Question answering (QA) can only make progress if we know if an answer is
correct, but for many of the most challenging and interesting QA examples,
current answer correctness (AC) metrics do not align with human judgments,
particularly verbose, free form answers from large language models (LLM). There
are two challenges: a lack of data and that models are too big. LLM based
scorers correlate better with humans, but this expensive task has only been
tested on limited QA datasets. We rectify these issues by providing clear
guidelines for evaluating machine QA adopted from human QA contests. We also
introduce Precise ANswer correctness Determination and Adjudication (PANDA), a
small, efficient, deterministic AC classifier (812 KB) that more accurately
evaluates answer correctness. | cs.CL | PANDA (Pedantic ANswer-correctness Determination and Adjudication):Improving Automatic Evaluation for Question Answering and Text Generation |
2024-02-17T01:27:29Z | http://arxiv.org/pdf/2402.11159v3 | http://arxiv.org/abs/2402.11159v3 | Yejun Yoon, Seunghyun Yoon, Kunwoo Park | 2024-06-07T00:56:09Z | cs.CL, cs.CV | This paper addresses the critical challenge of assessing the
representativeness of news thumbnail images, which often serve as the first
visual engagement for readers when an article is disseminated on social media.
We focus on whether a news image represents the actors discussed in the news
text. To serve the challenge, we introduce NewsTT, a manually annotated dataset
of 1000 news thumbnail images and text pairs. We found that the pretrained
vision and language models, such as BLIP-2, struggle with this task. Since news
subjects frequently involve named entities or proper nouns, the pretrained
models could have a limited capability to match news actors' visual and textual
appearances. We hypothesize that learning to contrast news text with its
counterfactual, of which named entities are replaced, can enhance the
cross-modal matching ability of vision and language models. We propose
CFT-CLIP, a contrastive learning framework that updates vision and language
bi-encoders according to the hypothesis. We found that our simple method can
boost the performance for assessing news thumbnail representativeness,
supporting our assumption. Code and data can be accessed at
https://github.com/ssu-humane/news-images-acl24. | cs.CL | Assessing News Thumbnail Representativeness: Counterfactual text can enhance the cross-modal matching ability |
2024-02-17T00:20:06Z | http://arxiv.org/pdf/2402.11142v1 | http://arxiv.org/abs/2402.11142v1 | Sizhe Zhou, Yu Meng, Bowen Jin, Jiawei Han | 2024-02-17T00:20:06Z | cs.CL | Relation extraction (RE), a crucial task in NLP, aims to identify semantic
relationships between entities mentioned in texts. Despite significant
advancements in this field, existing models typically rely on extensive
annotated data for training, which can be both costly and time-consuming to
acquire. Moreover, these models often struggle to adapt to new or unseen
relationships. In contrast, few-shot learning settings, which aim to reduce
annotation requirements, may offer incomplete and biased supervision for
understanding target relation semantics, leading to degraded and unstable
performance. To provide the model with accurate and explicit descriptions of
the relations types and meanwhile minimize the annotation requirements, we
study the definition only zero-shot RE setting where only relation definitions
expressed in natural language are used to train a RE model. Motivated by the
strong synthetic data generation power of LLMs, we propose a framework REPaL
which consists of three stages: (1) We utilize LLMs to generate initial seed
instances based on relation definitions and an unlabeled corpora. (2) We
fine-tune a bidirectional Small Language Model (SLM) using these initial seeds
to learn the relations for the target domain. (3) We enhance pattern coverage
and mitigate bias resulting from the limited number of initial seeds by
incorporating feedback acquired from SLM's predictions on unlabeled corpora. To
accomplish this, we leverage the multi-turn conversation ability of LLMs to
generate new instances in follow-up dialogues. Experiments on two datasets show
REPaL achieves better zero-shot performance with large margins over baseline
methods. | cs.CL | Grasping the Essentials: Tailoring Large Language Models for Zero-Shot Relation Extraction |
2024-02-17T00:13:36Z | http://arxiv.org/pdf/2402.11140v1 | http://arxiv.org/abs/2402.11140v1 | Sijia Chen, Baochun Li, Di Niu | 2024-02-17T00:13:36Z | cs.CL, cs.AI, cs.LG | The reasoning performance of Large Language Models (LLMs) on a wide range of
problems critically relies on chain-of-thought prompting, which involves
providing a few chain of thought demonstrations as exemplars in prompts. Recent
work, e.g., Tree of Thoughts, has pointed out the importance of exploration and
self-evaluation in reasoning step selection for complex problem solving. In
this paper, we present Boosting of Thoughts (BoT), an automated prompting
framework for problem solving with LLMs by iteratively exploring and
self-evaluating many trees of thoughts in order to acquire an ensemble of
trial-and-error reasoning experiences, which will serve as a new form of
prompting to solve the complex problem. Starting from a simple prompt without
requiring examples, BoT iteratively explores and evaluates a large collection
of reasoning steps, and more importantly, uses error analysis obtained from the
LLM on them to explicitly revise prompting, which in turn enhances reasoning
step generation, until a final answer is attained. Our experiments with GPT-4
and Llama2 across extensive complex mathematical problems demonstrate that BoT
consistently achieves higher or comparable problem-solving rates than other
advanced prompting approaches. | cs.CL | Boosting of Thoughts: Trial-and-Error Problem Solving with Large Language Models |
2024-02-17T00:09:32Z | http://arxiv.org/pdf/2402.11138v2 | http://arxiv.org/abs/2402.11138v2 | Tianyi Lorena Yan, Fei Wang, James Y. Huang, Wenxuan Zhou, Fan Yin, Aram Galstyan, Wenpeng Yin, Muhao Chen | 2024-06-06T06:03:34Z | cs.CL, cs.AI, cs.LG | Instruction tuning has been used as a promising approach to improve the
performance of large language models (LLMs) on unseen tasks. However, current
LLMs exhibit limited robustness to unseen instructions, generating inconsistent
outputs when the same instruction is phrased with slightly varied forms or
language styles. This behavior indicates LLMs' lack of robustness to textual
variations and generalizability to unseen instructions, potentially leading to
trustworthiness issues. Accordingly, we propose Contrastive Instruction Tuning,
which maximizes the similarity between the hidden representations of
semantically equivalent instruction-instance pairs while minimizing the
similarity between semantically different ones. To facilitate this approach, we
augment the existing FLAN collection by paraphrasing task instructions.
Experiments on the PromptBench benchmark show that CoIN consistently improves
LLMs' robustness to unseen instructions with variations across character, word,
sentence, and semantic levels by an average of +2.5% in accuracy. Code is
available at https://github.com/luka-group/CoIN. | cs.CL | Contrastive Instruction Tuning |
2024-02-16T23:44:38Z | http://arxiv.org/pdf/2402.14830v1 | http://arxiv.org/abs/2402.14830v1 | Arindam Mitra, Hamed Khanpour, Corby Rosset, Ahmed Awadallah | 2024-02-16T23:44:38Z | cs.CL, cs.AI | Mathematical word problem-solving has long been recognized as a complex task
for small language models (SLMs). A recent study hypothesized that the smallest
model size, needed to achieve over 80% accuracy on the GSM8K benchmark, is 34
billion parameters. To reach this level of performance with smaller models,
researcher often train SLMs to generate Python code or use tools to help avoid
calculation errors. Additionally, they employ ensembling, where outputs of up
to 100 model runs are combined to arrive at a more accurate result. Result
selection is done using consensus, majority vote or a separate a verifier model
used in conjunction with the SLM. Ensembling provides a substantial boost in
accuracy but at a significant cost increase with multiple calls to the model
(e.g., Phi-GSM uses top-48 to boost the performance from 68.2 to 81.5).
In this work, we present Orca-Math, a 7-billion-parameter SLM based on the
Mistral-7B, which achieves 86.81% on GSM8k without the need for multiple model
calls or the use of verifiers, code execution or any other external tools. Our
approach has the following key elements: (1) A high quality synthetic dataset
of 200K math problems created using a multi-agent setup where agents
collaborate to create the data, (2) An iterative learning techniques that
enables the SLM to practice solving problems, receive feedback on its solutions
and learn from preference pairs incorporating the SLM solutions and the
feedback. When trained with Supervised Fine-Tuning alone, Orca-Math achieves
81.50% on GSM8k pass@1 metric. With iterative preference learning, Orca-Math
achieves 86.81% pass@1. Orca-Math surpasses the performance of significantly
larger models such as LLAMA-2-70B, WizardMath-70B, Gemini-Pro, ChatGPT-3.5. It
also significantly outperforms other smaller models while using much smaller
data (hundreds of thousands vs. millions of problems). | cs.CL | Orca-Math: Unlocking the potential of SLMs in Grade School Math |
2024-02-16T23:36:43Z | http://arxiv.org/pdf/2402.11131v1 | http://arxiv.org/abs/2402.11131v1 | Nikhil Bhendawade, Irina Belousova, Qichen Fu, Henry Mason, Mohammad Rastegari, Mahyar Najibi | 2024-02-16T23:36:43Z | cs.CL, cs.AI, cs.LG | Speculative decoding is a prominent technique to speed up the inference of a
large target language model based on predictions of an auxiliary draft model.
While effective, in application-specific settings, it often involves
fine-tuning both draft and target models to achieve high acceptance rates. As
the number of downstream tasks grows, these draft models add significant
complexity to inference systems. We propose Speculative Streaming, a
single-model speculative decoding method that fuses drafting into the target
model by changing the fine-tuning objective from next token prediction to
future n-gram prediction. Speculative Streaming speeds up decoding by 1.8 -
3.1X in a diverse set of tasks, such as Summarization, Structured Queries, and
Meaning Representation, without sacrificing generation quality. Additionally,
Speculative Streaming is parameter-efficient. It achieves on-par/higher
speed-ups than Medusa-style architectures while using ~10000X fewer extra
parameters, making it well-suited for resource-constrained devices. | cs.CL | Speculative Streaming: Fast LLM Inference without Auxiliary Models |
2024-02-16T23:28:02Z | http://arxiv.org/pdf/2402.11129v1 | http://arxiv.org/abs/2402.11129v1 | Haoyu Wang, Tuo Zhao, Jing Gao | 2024-02-16T23:28:02Z | cs.CL | Retrieval-augmented Large Language Models (LLMs) offer substantial benefits
in enhancing performance across knowledge-intensive scenarios. However, these
methods often face challenges with complex inputs and encounter difficulties
due to noisy knowledge retrieval, notably hindering model effectiveness. To
address this issue, we introduce BlendFilter, a novel approach that elevates
retrieval-augmented LLMs by integrating query generation blending with
knowledge filtering. BlendFilter proposes the blending process through its
query generation method, which integrates both external and internal knowledge
augmentation with the original query, ensuring comprehensive information
gathering. Additionally, our distinctive knowledge filtering module capitalizes
on the intrinsic capabilities of the LLM, effectively eliminating extraneous
data. We conduct extensive experiments on three open-domain question answering
benchmarks, and the findings clearly indicate that our innovative BlendFilter
surpasses state-of-the-art baselines significantly. | cs.CL | BlendFilter: Advancing Retrieval-Augmented Large Language Models via Query Generation Blending and Knowledge Filtering |
2024-02-16T23:18:19Z | http://arxiv.org/pdf/2403.14633v3 | http://arxiv.org/abs/2403.14633v3 | Smriti Singh, Shuvam Keshari, Vinija Jain, Aman Chadha | 2024-04-16T17:14:16Z | cs.CY, cs.AI, cs.CL | Socioeconomic bias in society exacerbates disparities, influencing access to
opportunities and resources based on individuals' economic and social
backgrounds. This pervasive issue perpetuates systemic inequalities, hindering
the pursuit of inclusive progress as a society. In this paper, we investigate
the presence of socioeconomic bias, if any, in large language models. To this
end, we introduce a novel dataset SilverSpoon, consisting of 3000 samples that
illustrate hypothetical scenarios that involve underprivileged people
performing ethically ambiguous actions due to their circumstances, and ask
whether the action is ethically justified. Further, this dataset has a
dual-labeling scheme and has been annotated by people belonging to both ends of
the socioeconomic spectrum. Using SilverSpoon, we evaluate the degree of
socioeconomic bias expressed in large language models and the variation of this
degree as a function of model size. We also perform qualitative analysis to
analyze the nature of this bias. Our analysis reveals that while humans
disagree on which situations require empathy toward the underprivileged, most
large language models are unable to empathize with the socioeconomically
underprivileged regardless of the situation. To foster further research in this
domain, we make SilverSpoon and our evaluation harness publicly available. | cs.CY | Born With a Silver Spoon? Investigating Socioeconomic Bias in Large Language Models |
2024-02-16T23:08:55Z | http://arxiv.org/pdf/2402.11122v1 | http://arxiv.org/abs/2402.11122v1 | Zihao Lin, Mohammad Beigi, Hongxuan Li, Yufan Zhou, Yuxiang Zhang, Qifan Wang, Wenpeng Yin, Lifu Huang | 2024-02-16T23:08:55Z | cs.CL, cs.AI | Memory Editing (ME) has emerged as an efficient method to modify erroneous
facts or inject new facts into Large Language Models (LLMs). Two mainstream ME
methods exist: parameter-modifying ME and parameter-preserving ME (integrating
extra modules while preserving original parameters). Regrettably, previous
studies on ME evaluation have two critical limitations: (i) evaluating LLMs
with single edit only, neglecting the need for continuous editing, and (ii)
evaluations focusing solely on basic factual triples, overlooking broader LLM
capabilities like logical reasoning and reading understanding. This study
addresses these limitations with contributions threefold: (i) We explore how ME
affects a wide range of fundamental capabilities of LLMs under sequential
editing. Experimental results reveal an intriguing phenomenon: Most
parameter-modifying ME consistently degrade performance across all tasks after
a few sequential edits. In contrast, parameter-preserving ME effectively
maintains LLMs' fundamental capabilities but struggles to accurately recall
edited knowledge presented in a different format. (ii) We extend our evaluation
to different editing settings, such as layers to edit, model size, instruction
tuning, etc. Experimental findings indicate several strategies that can
potentially mitigate the adverse effects of ME. (iii) We further explain why
parameter-modifying ME damages LLMs from three dimensions: parameter changes
after editing, language modeling capability, and the in-context learning
capability. Our in-depth study advocates more careful use of ME in real-world
scenarios. | cs.CL | Navigating the Dual Facets: A Comprehensive Evaluation of Sequential Memory Editing in Large Language Models |
2024-02-16T22:34:53Z | http://arxiv.org/pdf/2402.11114v2 | http://arxiv.org/abs/2402.11114v2 | Zihao He, Siyi Guo, Ashwin Rao, Kristina Lerman | 2024-06-17T23:41:37Z | cs.CL, cs.CY, cs.SI | Language models (LMs) are known to represent the perspectives of some social
groups better than others, which may impact their performance, especially on
subjective tasks such as content moderation and hate speech detection. To
explore how LMs represent different perspectives, existing research focused on
positional alignment, i.e., how closely the models mimic the opinions and
stances of different groups, e.g., liberals or conservatives. However, human
communication also encompasses emotional and moral dimensions. We define the
problem of affective alignment, which measures how LMs' emotional and moral
tone represents those of different groups. By comparing the affect of responses
generated by 36 LMs to the affect of Twitter messages, we observe significant
misalignment of LMs with both ideological groups. This misalignment is larger
than the partisan divide in the U.S. Even after steering the LMs towards
specific ideological perspectives, the misalignment and liberal tendencies of
the model persist, suggesting a systemic bias within LMs. | cs.CL | Whose Emotions and Moral Sentiments Do Language Models Reflect? |
2024-02-16T22:24:13Z | http://arxiv.org/pdf/2402.11111v1 | http://arxiv.org/abs/2402.11111v1 | Alexis Chevalier, Jiayi Geng, Alexander Wettig, Howard Chen, Sebastian Mizera, Toni Annala, Max Jameson Aragon, Arturo Rodríguez Fanlo, Simon Frieder, Simon Machado, Akshara Prabhakar, Ellie Thieu, Jiachen T. Wang, Zirui Wang, Xindi Wu, Mengzhou Xia, Wenhan Jia, Jiatong Yu, Jun-Jie Zhu, Zhiyong Jason Ren, Sanjeev Arora, Danqi Chen | 2024-02-16T22:24:13Z | cs.CL | NLP has recently made exciting progress toward training language models (LMs)
with strong scientific problem-solving skills. However, model development has
not focused on real-life use-cases of LMs for science, including applications
in education that require processing long scientific documents. To address
this, we introduce TutorEval and TutorChat. TutorEval is a diverse
question-answering benchmark consisting of questions about long chapters from
STEM textbooks, written by experts. TutorEval helps measure real-life usability
of LMs as scientific assistants, and it is the first benchmark combining long
contexts, free-form generation, and multi-disciplinary scientific knowledge.
Moreover, we show that fine-tuning base models with existing dialogue datasets
leads to poor performance on TutorEval. Therefore, we create TutorChat, a
dataset of 80,000 long synthetic dialogues about textbooks. We use TutorChat to
fine-tune Llemma models with 7B and 34B parameters. These LM tutors specialized
in math have a 32K-token context window, and they excel at TutorEval while
performing strongly on GSM8K and MATH. Our datasets build on open-source
materials, and we release our models, data, and evaluations. | cs.CL | Language Models as Science Tutors |
2024-02-16T22:12:53Z | http://arxiv.org/pdf/2402.11100v2 | http://arxiv.org/abs/2402.11100v2 | Yinghui Li, Qingyu Zhou, Yuanzhen Luo, Shirong Ma, Yangning Li, Hai-Tao Zheng, Xuming Hu, Philip S. Yu | 2024-06-09T17:55:05Z | cs.CL | Recently, Large Language Models (LLMs) make remarkable evolutions in language
understanding and generation. Following this, various benchmarks for measuring
all kinds of capabilities of LLMs have sprung up. In this paper, we challenge
the reasoning and understanding abilities of LLMs by proposing a FaLlacy
Understanding Benchmark (FLUB) containing cunning texts that are easy for
humans to understand but difficult for models to grasp. Specifically, the
cunning texts that FLUB focuses on mainly consist of the tricky, humorous, and
misleading texts collected from the real internet environment. And we design
three tasks with increasing difficulty in the FLUB benchmark to evaluate the
fallacy understanding ability of LLMs. Based on FLUB, we investigate the
performance of multiple representative and advanced LLMs, reflecting our FLUB
is challenging and worthy of more future study. Interesting discoveries and
valuable insights are achieved in our extensive experiments and detailed
analyses. We hope that our benchmark can encourage the community to improve
LLMs' ability to understand fallacies. Our data and codes are available at
https://github.com/THUKElab/FLUB. | cs.CL | When LLMs Meet Cunning Texts: A Fallacy Understanding Benchmark for Large Language Models |
2024-02-16T21:49:17Z | http://arxiv.org/pdf/2403.15397v1 | http://arxiv.org/abs/2403.15397v1 | Gabriel Nicholas, Paul Friedl | 2024-02-16T21:49:17Z | cs.CY, cs.AI, cs.CL | On July 20, 2023, a group of 27 scholars and digital rights advocates with
expertise in law, computer science, political science, and other disciplines
gathered for the Large Language Models, Law and Policy Roundtable, co-hosted by
the NYU School of Law's Information Law Institute and the Center for Democracy
& Technology. The roundtable convened to discuss how law and policy can help
address some of the larger societal problems posed by large language models
(LLMs). The discussion focused on three policy topic areas in particular:
1. Truthfulness: What risks do LLMs pose in terms of generating mis- and
disinformation? How can these risks be mitigated from a technical and/or
regulatory perspective?
2. Privacy: What are the biggest privacy risks involved in the creation,
deployment, and use of LLMs? How can these risks be mitigated from a technical
and/or regulatory perspective?
3. Market concentration: What threats do LLMs pose concerning market/power
concentration? How can these risks be mitigated from a technical and/or
regulatory perspective?
In this paper, we provide a detailed summary of the day's proceedings. We
first recap what we deem to be the most important contributions made during the
issue framing discussions. We then provide a list of potential legal and
regulatory interventions generated during the brainstorming discussions. | cs.CY | Regulating Large Language Models: A Roundtable Report |
2024-02-16T21:47:30Z | http://arxiv.org/pdf/2402.11094v2 | http://arxiv.org/abs/2402.11094v2 | Matthew Freestone, Shubhra Kanti Karmaker Santu | 2024-03-02T14:01:04Z | cs.CL, I.2.7 | Learning meaningful word embeddings is key to training a robust language
model. The recent rise of Large Language Models (LLMs) has provided us with
many new word/sentence/document embedding models. Although LLMs have shown
remarkable advancement in various NLP tasks, it is still unclear whether the
performance improvement is merely because of scale or whether underlying
embeddings they produce significantly differ from classical encoding models
like Sentence-BERT (SBERT) or Universal Sentence Encoder (USE). This paper
systematically investigates this issue by comparing classical word embedding
techniques against LLM-based word embeddings in terms of their latent vector
semantics. Our results show that LLMs tend to cluster semantically related
words more tightly than classical models. LLMs also yield higher average
accuracy on the Bigger Analogy Test Set (BATS) over classical methods. Finally,
some LLMs tend to produce word embeddings similar to SBERT, a relatively
lighter classical model. | cs.CL | Word Embeddings Revisited: Do LLMs Offer Something New? |
2024-02-16T21:10:33Z | http://arxiv.org/pdf/2402.11078v3 | http://arxiv.org/abs/2402.11078v3 | Govind Gangadhar, Karl Stratos | 2024-06-03T05:39:10Z | cs.LG, cs.AI, cs.CL | Standard fine-tuning is considered not as effective as specialized methods
for model editing due to its comparatively poor performance. However, it is
simple, agnostic to the architectural details of the model being edited, and
able to leverage advances in standard training techniques with no additional
work (e.g., black-box PEFT for computational efficiency), making it an
appealing choice for a model editor. In this work, we show that standard
fine-tuning alone can yield competitive model editing performance with two
minor modifications. First, we optimize the conditional likelihood rather than
the full likelihood. Second, in addition to the typical practice of training on
randomly paraphrased edit prompts to encourage generalization, we also train on
random or similar unedited facts to encourage locality. Our experiments on the
ZsRE and CounterFact datasets demonstrate that these simple modifications allow
standard fine-tuning to match or outperform highly specialized editors in terms
of edit score. | cs.LG | Model Editing by Standard Fine-Tuning |
2024-02-16T20:59:57Z | http://arxiv.org/pdf/2402.11073v3 | http://arxiv.org/abs/2402.11073v3 | Jingwei Ni, Minjing Shi, Dominik Stammbach, Mrinmaya Sachan, Elliott Ash, Markus Leippold | 2024-06-02T18:35:25Z | cs.CL, cs.AI | With the rise of generative AI, automated fact-checking methods to combat
misinformation are becoming more and more important. However, factual claim
detection, the first step in a fact-checking pipeline, suffers from two key
issues that limit its scalability and generalizability: (1) inconsistency in
definitions of the task and what a claim is, and (2) the high cost of manual
annotation. To address (1), we review the definitions in related work and
propose a unifying definition of factual claims that focuses on verifiability.
To address (2), we introduce AFaCTA (Automatic Factual Claim deTection
Annotator), a novel framework that assists in the annotation of factual claims
with the help of large language models (LLMs). AFaCTA calibrates its annotation
confidence with consistency along three predefined reasoning paths. Extensive
evaluation and experiments in the domain of political speech reveal that AFaCTA
can efficiently assist experts in annotating factual claims and training
high-quality classifiers, and can work with or without expert supervision. Our
analyses also result in PoliClaim, a comprehensive claim detection dataset
spanning diverse political topics. | cs.CL | AFaCTA: Assisting the Annotation of Factual Claim Detection with Reliable LLM Annotators |
2024-02-16T20:48:53Z | http://arxiv.org/pdf/2402.11068v1 | http://arxiv.org/abs/2402.11068v1 | Guangya Wan, Yuqi Wu, Mengxuan Hu, Zhixuan Chu, Sheng Li | 2024-02-16T20:48:53Z | cs.CL, cs.AI | Causal discovery (CD) and Large Language Models (LLMs) represent two emerging
fields of study with significant implications for artificial intelligence.
Despite their distinct origins, CD focuses on uncovering cause-effect
relationships from data, and LLMs on processing and generating humanlike text,
the convergence of these domains offers novel insights and methodologies for
understanding complex systems. This paper presents a comprehensive survey of
the integration of LLMs, such as GPT4, into CD tasks. We systematically review
and compare existing approaches that leverage LLMs for various CD tasks and
highlight their innovative use of metadata and natural language to infer causal
structures. Our analysis reveals the strengths and potential of LLMs in both
enhancing traditional CD methods and as an imperfect expert, alongside the
challenges and limitations inherent in current practices. Furthermore, we
identify gaps in the literature and propose future research directions aimed at
harnessing the full potential of LLMs in causality research. To our knowledge,
this is the first survey to offer a unified and detailed examination of the
synergy between LLMs and CD, setting the stage for future advancements in the
field. | cs.CL | Bridging Causal Discovery and Large Language Models: A Comprehensive Survey of Integrative Approaches and Future Directions |
2024-02-16T20:20:43Z | http://arxiv.org/pdf/2402.11060v1 | http://arxiv.org/abs/2402.11060v1 | Chenkai Sun, Ke Yang, Revanth Gangi Reddy, Yi R. Fung, Hou Pong Chan, ChengXiang Zhai, Heng Ji | 2024-02-16T20:20:43Z | cs.CL, cs.AI, cs.IR | The increasing demand for personalized interactions with large language
models (LLMs) calls for the development of methodologies capable of accurately
and efficiently identifying user opinions and preferences. Retrieval
augmentation emerges as an effective strategy, as it can accommodate a vast
number of users without the costs from fine-tuning. Existing research, however,
has largely focused on enhancing the retrieval stage and devoted limited
exploration toward optimizing the representation of the database, a crucial
aspect for tasks such as personalization. In this work, we examine the problem
from a novel angle, focusing on how data can be better represented for more
efficient retrieval in the context of LLM customization. To tackle this
challenge, we introduce Persona-DB, a simple yet effective framework consisting
of a hierarchical construction process to improve generalization across task
contexts and collaborative refinement to effectively bridge knowledge gaps
among users. In the task of response forecasting, Persona-DB demonstrates
superior efficiency in maintaining accuracy with a significantly reduced
retrieval size, a critical advantage in scenarios with extensive histories or
limited context windows. Our experiments also indicate a marked improvement of
over 15% under cold-start scenarios, when users have extremely sparse data.
Furthermore, our analysis reveals the increasing importance of collaborative
knowledge as the retrieval capacity expands. | cs.CL | Persona-DB: Efficient Large Language Model Personalization for Response Prediction with Collaborative Data Refinement |
2024-02-16T20:14:47Z | http://arxiv.org/pdf/2402.11058v3 | http://arxiv.org/abs/2402.11058v3 | Jihyung Kil, Farideh Tavazoee, Dongyeop Kang, Joo-Kyung Kim | 2024-06-03T01:09:38Z | cs.CV, cs.CL | Visual Question Answering (VQA) often involves diverse reasoning scenarios
across Vision and Language (V&L). Most prior VQA studies, however, have merely
focused on assessing the model's overall accuracy without evaluating it on
different reasoning cases. Furthermore, some recent works observe that
conventional Chain-of-Thought (CoT) prompting fails to generate effective
reasoning for VQA, especially for complex scenarios requiring multi-hop
reasoning. In this paper, we propose II-MMR, a novel idea to identify and
improve multi-modal multi-hop reasoning in VQA. In specific, II-MMR takes a VQA
question with an image and finds a reasoning path to reach its answer using two
novel language promptings: (i) answer prediction-guided CoT prompt, or (ii)
knowledge triplet-guided prompt. II-MMR then analyzes this path to identify
different reasoning cases in current VQA benchmarks by estimating how many hops
and what types (i.e., visual or beyond-visual) of reasoning are required to
answer the question. On popular benchmarks including GQA and A-OKVQA, II-MMR
observes that most of their VQA questions are easy to answer, simply demanding
"single-hop" reasoning, whereas only a few questions require "multi-hop"
reasoning. Moreover, while the recent V&L model struggles with such complex
multi-hop reasoning questions even using the traditional CoT method, II-MMR
shows its effectiveness across all reasoning cases in both zero-shot and
fine-tuning settings. | cs.CV | II-MMR: Identifying and Improving Multi-modal Multi-hop Reasoning in Visual Question Answering |
2024-02-16T19:59:45Z | http://arxiv.org/pdf/2402.11051v1 | http://arxiv.org/abs/2402.11051v1 | Runcong Zhao, Qinglin Zhu, Hainiu Xu, Jiazheng Li, Yuxiang Zhou, Yulan He, Lin Gui | 2024-02-16T19:59:45Z | cs.CL, cs.AI | Existing datasets for narrative understanding often fail to represent the
complexity and uncertainty of relationships in real-life social scenarios. To
address this gap, we introduce a new benchmark, Conan, designed for extracting
and analysing intricate character relation graphs from detective narratives.
Specifically, we designed hierarchical relationship categories and manually
extracted and annotated role-oriented relationships from the perspectives of
various characters, incorporating both public relationships known to most
characters and secret ones known to only a few. Our experiments with advanced
Large Language Models (LLMs) like GPT-3.5, GPT-4, and Llama2 reveal their
limitations in inferencing complex relationships and handling longer
narratives. The combination of the Conan dataset and our pipeline strategy is
geared towards understanding the ability of LLMs to comprehend nuanced
relational dynamics in narrative contexts. | cs.CL | Large Language Models Fall Short: Understanding Complex Relationships in Detective Narratives |
2024-02-16T19:28:52Z | http://arxiv.org/pdf/2402.11035v2 | http://arxiv.org/abs/2402.11035v2 | Benjamin Reichman, Larry Heck | 2024-04-16T20:22:21Z | cs.CL, cs.IR | Dense passage retrieval (DPR) is the first step in the retrieval augmented
generation (RAG) paradigm for improving the performance of large language
models (LLM). DPR fine-tunes pre-trained networks to enhance the alignment of
the embeddings between queries and relevant textual data. A deeper
understanding of DPR fine-tuning will be required to fundamentally unlock the
full potential of this approach. In this work, we explore DPR-trained models
mechanistically by using a combination of probing, layer activation analysis,
and model editing. Our experiments show that DPR training decentralizes how
knowledge is stored in the network, creating multiple access pathways to the
same information. We also uncover a limitation in this training style: the
internal knowledge of the pre-trained model bounds what the retrieval model can
retrieve. These findings suggest a few possible directions for dense retrieval:
(1) expose the DPR training process to more knowledge so more can be
decentralized, (2) inject facts as decentralized representations, (3) model and
incorporate knowledge uncertainty in the retrieval process, and (4) directly
map internal model knowledge to a knowledge base. | cs.CL | Retrieval-Augmented Generation: Is Dense Passage Retrieval Retrieving? |
2024-02-16T19:26:09Z | http://arxiv.org/pdf/2402.11034v2 | http://arxiv.org/abs/2402.11034v2 | Jannat Ara Meem, Muhammad Shihab Rashid, Yue Dong, Vagelis Hristidis | 2024-06-03T18:36:59Z | cs.CL | Existing work on Temporal Question Answering (TQA) has predominantly focused
on questions anchored to specific timestamps or events (e.g. "Who was the US
president in 1970?"). Little work has studied questions whose temporal context
is relative to the present time (e.g. "Who was the previous US president?"). We
refer to this problem as Present-Anchored Temporal QA (PATQA). PATQA poses
unique challenges: (1) large language models (LLMs) may have outdated
knowledge, (2) complex temporal relationships (e.g. 'before', 'previous') are
hard to reason, (3) multi-hop reasoning may be required, and (4) the gold
answers of benchmarks must be continuously updated. To address these
challenges, we introduce the PAT-Questions benchmark, which includes single and
multi-hop temporal questions. The answers in PAT-Questions can be automatically
refreshed by re-running SPARQL queries on a knowledge graph, if available. We
evaluate several state-of-the-art LLMs and a SOTA temporal reasoning model
(TEMPREASON-T5) on PAT-Questions through direct prompting and
retrieval-augmented generation (RAG). The results highlight the limitations of
existing solutions in PATQA and motivate the need for new methods to improve
PATQA reasoning capabilities. | cs.CL | PAT-Questions: A Self-Updating Benchmark for Present-Anchored Temporal Question-Answering |
2024-02-16T18:50:24Z | http://arxiv.org/pdf/2402.10893v1 | http://arxiv.org/abs/2402.10893v1 | Moritz Stephan, Alexander Khazatsky, Eric Mitchell, Annie S Chen, Sheryl Hsu, Archit Sharma, Chelsea Finn | 2024-02-16T18:50:24Z | cs.LG, cs.AI, cs.CL | The diversity of contexts in which large language models (LLMs) are deployed
requires the ability to modify or customize default model behaviors to
incorporate nuanced requirements and preferences. A convenient interface to
specify such model adjustments is high-level verbal feedback, such as "Don't
use emojis when drafting emails to my boss." However, while writing high-level
feedback is far simpler than collecting annotations for reinforcement learning
from human feedback (RLHF), we find that simply prompting a model with such
feedback leads to overgeneralization of the feedback to contexts where it is
not relevant. We study the problem of incorporating verbal feedback without
such overgeneralization, inspiring a new method Contextualized Critiques with
Constrained Preference Optimization (C3PO). C3PO uses a piece of high-level
feedback to generate a small synthetic preference dataset specifying how the
feedback should (and should not) be applied. It then fine-tunes the model in
accordance with the synthetic preference data while minimizing the divergence
from the original model for prompts where the feedback does not apply. Our
experimental results indicate that our approach effectively applies verbal
feedback to relevant scenarios while preserving existing behaviors for other
contexts. For both human- and GPT-4-generated high-level feedback, C3PO
effectively adheres to the given feedback comparably to in-context baselines
while reducing overgeneralization by 30%. | cs.LG | RLVF: Learning from Verbal Feedback without Overgeneralization |
2024-02-16T18:49:27Z | http://arxiv.org/pdf/2402.10892v2 | http://arxiv.org/abs/2402.10892v2 | Johnny Tian-Zheng Wei, Ryan Yixiang Wang, Robin Jia | 2024-06-10T19:39:34Z | cs.CR, cs.CL, cs.LG | Detecting whether copyright holders' works were used in LLM pretraining is
poised to be an important problem. This work proposes using data watermarks to
enable principled detection with only black-box model access, provided that the
rightholder contributed multiple training documents and watermarked them before
public release. By applying a randomly sampled data watermark, detection can be
framed as hypothesis testing, which provides guarantees on the false detection
rate. We study two watermarks: one that inserts random sequences, and another
that randomly substitutes characters with Unicode lookalikes. We first show how
three aspects of watermark design -- watermark length, number of duplications,
and interference -- affect the power of the hypothesis test. Next, we study how
a watermark's detection strength changes under model and dataset scaling: while
increasing the dataset size decreases the strength of the watermark, watermarks
remain strong if the model size also increases. Finally, we view SHA hashes as
natural watermarks and show that we can robustly detect hashes from
BLOOM-176B's training data, as long as they occurred at least 90 times.
Together, our results point towards a promising future for data watermarks in
real world use. | cs.CR | Proving membership in LLM pretraining data via data watermarks |
2024-02-16T18:47:21Z | http://arxiv.org/pdf/2402.10891v1 | http://arxiv.org/abs/2402.10891v1 | Dylan Zhang, Justin Wang, Francois Charton | 2024-02-16T18:47:21Z | cs.CL, cs.AI, cs.LG | Instruction tuning -- fine-tuning a large language model (LLM) on pairs of
instructions and desired outcomes -- is an approach that enables pre-trained
language models to perform real-world tasks and follow human instructions. Its
practical success depends on the model learning a broader set of instructions
than those it was trained on. Yet the factors that determine model
generalization to such \emph{unseen tasks} are not well understood. %To
understand the driving factors of generalization, In this paper, we experiment
with string rewrites, a symbolic task that serves as a building block for
Turing complete Markov algorithms while allowing experimental control of
"inputs" and "instructions". We investigate the trade-off between the number of
instructions the model is trained on and the number of training samples
provided for each instruction and observe that the diversity of the instruction
set determines generalization. Generalization emerges once a diverse enough set
of tasks is provided, even though very few examples are provided for each task.
Instruction diversity also ensures robustness with respect to non-uniform
distributions of instructions in the training set. | cs.CL | Instruction Diversity Drives Generalization To Unseen Tasks |
2024-02-16T18:45:58Z | http://arxiv.org/pdf/2402.10890v2 | http://arxiv.org/abs/2402.10890v2 | Ziru Chen, Michael White, Raymond Mooney, Ali Payani, Yu Su, Huan Sun | 2024-06-06T14:55:40Z | cs.CL, cs.AI, cs.LG | In this paper, we examine how large language models (LLMs) solve multi-step
problems under a language agent framework with three components: a generator, a
discriminator, and a planning method. We investigate the practical utility of
two advanced planning methods, iterative correction and tree search. We present
a comprehensive analysis of how discrimination accuracy affects the overall
performance of agents when using these two methods or a simpler method,
re-ranking. Experiments on two tasks, text-to-SQL parsing and mathematical
reasoning, show that: (1) advanced planning methods demand discriminators with
at least 90% accuracy to achieve significant improvements over re-ranking; (2)
current LLMs' discrimination abilities have not met the needs of advanced
planning methods to achieve such improvements; (3) with LLM-based
discriminators, advanced planning methods may not adequately balance accuracy
and efficiency. For example, compared to the other two methods, tree search is
at least 10--20 times slower but leads to negligible performance gains, which
hinders its real-world applications. Code and data are available at
https://github.com/OSU-NLP-Group/llm-planning-eval. | cs.CL | When is Tree Search Useful for LLM Planning? It Depends on the Discriminator |
2024-02-16T18:43:10Z | http://arxiv.org/pdf/2402.10886v1 | http://arxiv.org/abs/2402.10886v1 | Zhaolin Gao, Kianté Brantley, Thorsten Joachims | 2024-02-16T18:43:10Z | cs.CL | Recent developments in LLMs offer new opportunities for assisting authors in
improving their work. In this paper, we envision a use case where authors can
receive LLM-generated reviews that uncover weak points in the current draft.
While initial methods for automated review generation already exist, these
methods tend to produce reviews that lack detail, and they do not cover the
range of opinions that human reviewers produce. To address this shortcoming, we
propose an efficient two-stage review generation framework called Reviewer2.
Unlike prior work, this approach explicitly models the distribution of possible
aspects that the review may address. We show that this leads to more detailed
reviews that better cover the range of aspects that human reviewers identify in
the draft. As part of the research, we generate a large-scale review dataset of
27k papers and 99k reviews that we annotate with aspect prompts, which we make
available as a resource for future research. | cs.CL | Reviewer2: Optimizing Review Generation Through Prompt Generation |
2024-02-16T18:42:08Z | http://arxiv.org/pdf/2402.10884v1 | http://arxiv.org/abs/2402.10884v1 | Shengzhi Li, Rongyu Lin, Shichao Pei | 2024-02-16T18:42:08Z | cs.CL, cs.AI, cs.CV, cs.LG | In production, multi-modal large language models (MLLMs) are expected to
support multi-turn queries of interchanging image and text modalities. However,
the current MLLMs trained with visual-question-answering (VQA) datasets could
suffer from degradation, as VQA datasets lack the diversity and complexity of
the original text instruction datasets which the underlying language model had
been trained with. To address this challenging degradation, we first collect a
lightweight (6k entries) VQA preference dataset where answers were annotated by
Gemini for 5 quality metrics in a granular fashion, and investigate standard
Supervised Fine-tuning, rejection sampling, Direct Preference Optimization
(DPO), and SteerLM. Our findings indicate that the with DPO we are able to
surpass instruction-following capabilities of the language model, achieving a
6.73 score on MT-Bench, compared to Vicuna's 6.57 and LLaVA's 5.99 despite
small data scale. This enhancement in textual instruction proficiency
correlates with boosted visual instruction performance (+4.9\% on MM-Vet, +6\%
on LLaVA-Bench), with minimal alignment tax on visual knowledge benchmarks
compared to previous RLHF approach. In conclusion, we propose a
distillation-based multi-modal alignment model with fine-grained annotations on
a small dataset that reconciles the textual and visual performance of MLLMs,
restoring and boosting language capability after visual instruction tuning. | cs.CL | Multi-modal preference alignment remedies regression of visual instruction tuning on language model |
2024-02-16T18:36:36Z | http://arxiv.org/pdf/2402.10882v3 | http://arxiv.org/abs/2402.10882v3 | Zongyu Wu, Hongcheng Gao, Yueze Wang, Xiang Zhang, Suhang Wang | 2024-06-25T05:56:40Z | cs.CV, cs.CL | Text-to-Image (T2I) models have shown great performance in generating images
based on textual prompts. However, these models are vulnerable to unsafe input
to generate unsafe content like sexual, harassment and illegal-activity images.
Existing studies based on image checker, model fine-tuning and embedding
blocking are impractical in real-world applications. Hence, we propose the
first universal prompt optimizer for safe T2I (POSI) generation in black-box
scenario. We first construct a dataset consisting of toxic-clean prompt pairs
by GPT-3.5 Turbo. To guide the optimizer to have the ability of converting
toxic prompt to clean prompt while preserving semantic information, we design a
novel reward function measuring toxicity and text alignment of generated images
and train the optimizer through Proximal Policy Optimization. Experiments show
that our approach can effectively reduce the likelihood of various T2I models
in generating inappropriate images, with no significant impact on text
alignment. It is also flexible to be combined with methods to achieve better
performance. Our code is available at https://github.com/wzongyu/POSI. | cs.CV | Universal Prompt Optimizer for Safe Text-to-Image Generation |
2024-02-16T18:28:43Z | http://arxiv.org/pdf/2402.11005v2 | http://arxiv.org/abs/2402.11005v2 | Sarath Sivaprasad, Pramod Kaushik, Sahar Abdelnabi, Mario Fritz | 2024-02-21T22:02:18Z | cs.CL, cs.AI | Large-Language-Models (LLMs) are deployed in a wide range of applications,
and their response has an increasing social impact. Understanding the
non-deliberate(ive) mechanism of LLMs in giving responses is essential in
explaining their performance and discerning their biases in real-world
applications. This is analogous to human studies, where such inadvertent
responses are referred to as sampling. We study this sampling of LLMs in light
of value bias and show that the sampling of LLMs tends to favour high-value
options. Value bias corresponds to this shift of response from the most likely
towards an ideal value represented in the LLM. In fact, this effect can be
reproduced even with new entities learnt via in-context prompting. We show that
this bias manifests in unexpected places and has implications on relevant
application scenarios, like choosing exemplars. The results show that value
bias is strong in LLMs across different categories, similar to the results
found in human studies. | cs.CL | Exploring Value Biases: How LLMs Deviate Towards the Ideal |
2024-02-16T18:03:42Z | http://arxiv.org/pdf/2402.10866v2 | http://arxiv.org/abs/2402.10866v2 | Muhammad Shihab Rashid, Jannat Ara Meem, Yue Dong, Vagelis Hristidis | 2024-05-28T02:34:57Z | cs.CL | Large Language Models (LLMs) have achieved state-of-the-art performance in
text re-ranking. This process includes queries and candidate passages in the
prompts, utilizing pointwise, listwise, and pairwise prompting strategies. A
limitation of these ranking strategies with LLMs is their cost: the process can
become expensive due to API charges, which are based on the number of input and
output tokens. We study how to maximize the re-ranking performance given a
budget, by navigating the vast search spaces of prompt choices, LLM APIs, and
budget splits. We propose a suite of budget-constrained methods to perform text
re-ranking using a set of LLM APIs. Our most efficient method, called EcoRank,
is a two-layered pipeline that jointly optimizes decisions regarding budget
allocation across prompt strategies and LLM APIs. Our experimental results on
four popular QA and passage reranking datasets show that EcoRank outperforms
other budget-aware supervised and unsupervised baselines. | cs.CL | EcoRank: Budget-Constrained Text Re-ranking Using Large Language Models |
2024-02-16T17:15:28Z | http://arxiv.org/pdf/2402.10835v2 | http://arxiv.org/abs/2402.10835v2 | Mingyu Jin, Hua Tang, Chong Zhang, Qinkai Yu, Chengzhi Liu, Suiyuan Zhu, Yongfeng Zhang, Mengnan Du | 2024-02-19T02:30:15Z | cs.CL | Large language models (LLMs) have been applied in many fields with rapid
development in recent years. As a classic machine learning task, time series
forecasting has recently received a boost from LLMs. However, there is a
research gap in the LLMs' preferences in this field. In this paper, by
comparing LLMs with traditional models, many properties of LLMs in time series
prediction are found. For example, our study shows that LLMs excel in
predicting time series with clear patterns and trends but face challenges with
datasets lacking periodicity. We explain our findings through designing prompts
to require LLMs to tell the period of the datasets. In addition, the input
strategy is investigated, and it is found that incorporating external knowledge
and adopting natural language paraphrases positively affects the predictive
performance of LLMs for time series. Overall, this study contributes to insight
into the advantages and limitations of LLMs in time series forecasting under
different conditions. | cs.CL | Time Series Forecasting with LLMs: Understanding and Enhancing Model Capabilities |
2024-02-16T17:03:05Z | http://arxiv.org/pdf/2402.11000v2 | http://arxiv.org/abs/2402.11000v2 | Yangyifei Luo, Zhuo Chen, Lingbing Guo, Qian Li, Wenxuan Zeng, Zhixin Cai, Jianxin Li | 2024-03-05T13:57:28Z | cs.CL, cs.AI | Entity alignment (EA) aims to identify entities across different knowledge
graphs that represent the same real-world objects. Recent embedding-based EA
methods have achieved state-of-the-art performance in EA yet faced
interpretability challenges as they purely rely on the embedding distance and
neglect the logic rules behind a pair of aligned entities. In this paper, we
propose the Align-Subgraph Entity Alignment (ASGEA) framework to exploit logic
rules from Align-Subgraphs. ASGEA uses anchor links as bridges to construct
Align-Subgraphs and spreads along the paths across KGs, which distinguishes it
from the embedding-based methods. Furthermore, we design an interpretable
Path-based Graph Neural Network, ASGNN, to effectively identify and integrate
the logic rules across KGs. We also introduce a node-level multi-modal
attention mechanism coupled with multi-modal enriched anchors to augment the
Align-Subgraph. Our experimental results demonstrate the superior performance
of ASGEA over the existing embedding-based methods in both EA and Multi-Modal
EA (MMEA) tasks. | cs.CL | ASGEA: Exploiting Logic Rules from Align-Subgraphs for Entity Alignment |
2024-02-16T17:02:53Z | http://arxiv.org/pdf/2403.04769v2 | http://arxiv.org/abs/2403.04769v2 | Benjamin Lemkin | 2024-03-11T01:21:32Z | cs.CR, cs.AI, cs.CL, cs.LG | Large language models (LLMs) are initially trained on vast amounts of data,
then fine-tuned using reinforcement learning from human feedback (RLHF); this
also serves to teach the LLM to provide appropriate and safe responses. In this
paper, we present a novel method to manipulate the fine-tuned version into
reverting to its pre-RLHF behavior, effectively erasing the model's filters;
the exploit currently works for GPT4, Claude Sonnet, and (to some extent) for
Inflection-2.5. Unlike other jailbreaks (for example, the popular "Do Anything
Now" (DAN) ), our method does not rely on instructing the LLM to override its
RLHF policy; hence, simply modifying the RLHF process is unlikely to address
it. Instead, we induce a hallucination involving reversed text during which the
model reverts to a word bucket, effectively pausing the model's filter. We
believe that our exploit presents a fundamental vulnerability in LLMs currently
unaddressed, as well as an opportunity to better understand the inner workings
of LLMs during hallucinations. | cs.CR | Using Hallucinations to Bypass GPT4's Filter |
2024-02-16T16:35:41Z | http://arxiv.org/pdf/2402.10812v1 | http://arxiv.org/abs/2402.10812v1 | Qi Shi, Han Cui, Haofeng Wang, Qingfu Zhu, Wanxiang Che, Ting Liu | 2024-02-16T16:35:41Z | cs.CL | Question answering over heterogeneous data requires reasoning over diverse
sources of data, which is challenging due to the large scale of information and
organic coupling of heterogeneous data. Various approaches have been proposed
to address these challenges. One approach involves training specialized
retrievers to select relevant information, thereby reducing the input length.
Another approach is to transform diverse modalities of data into a single
modality, simplifying the task difficulty and enabling more straightforward
processing. In this paper, we propose HProPro, a novel program-based prompting
framework for the hybrid question answering task. HProPro follows the code
generation and execution paradigm. In addition, HProPro integrates various
functions to tackle the hybrid reasoning scenario. Specifically, HProPro
contains function declaration and function implementation to perform hybrid
information-seeking over data from various sources and modalities, which
enables reasoning over such data without training specialized retrievers or
performing modal transformations. Experimental results on two typical hybrid
question answering benchmarks HybridQA and MultiModalQA demonstrate the
effectiveness of HProPro: it surpasses all baseline systems and achieves the
best performances in the few-shot settings on both datasets. | cs.CL | Exploring Hybrid Question Answering via Program-based Prompting |
2024-02-16T16:35:35Z | http://arxiv.org/pdf/2402.10811v2 | http://arxiv.org/abs/2402.10811v2 | Tiancheng Hu, Nigel Collier | 2024-06-17T11:06:57Z | cs.CL, cs.CY | Large language models (LLMs) have shown remarkable promise in simulating
human language and behavior. This study investigates how integrating persona
variables-demographic, social, and behavioral factors-impacts LLMs' ability to
simulate diverse perspectives. We find that persona variables account for <10%
variance in annotations in existing subjective NLP datasets. Nonetheless,
incorporating persona variables via prompting in LLMs provides modest but
statistically significant improvements. Persona prompting is most effective in
samples where many annotators disagree, but their disagreements are relatively
minor. Notably, we find a linear relationship in our setting: the stronger the
correlation between persona variables and human annotations, the more accurate
the LLM predictions are using persona prompting. In a zero-shot setting, a
powerful 70b model with persona prompting captures 81% of the annotation
variance achievable by linear regression trained on ground truth annotations.
However, for most subjective NLP datasets, where persona variables have limited
explanatory power, the benefits of persona prompting are limited. | cs.CL | Quantifying the Persona Effect in LLM Simulations |
2024-02-16T16:31:46Z | http://arxiv.org/pdf/2402.10805v1 | http://arxiv.org/abs/2402.10805v1 | Yongqi Li, Wenjie Wang, Leigang Qu, Liqiang Nie, Wenjie Li, Tat-Seng Chua | 2024-02-16T16:31:46Z | cs.MM, cs.AI, cs.CL, cs.CV, cs.IR | The recent advancements in generative language models have demonstrated their
ability to memorize knowledge from documents and recall knowledge to respond to
user queries effectively. Building upon this capability, we propose to enable
multimodal large language models (MLLMs) to memorize and recall images within
their parameters. Given a user query for visual content, the MLLM is
anticipated to "recall" the relevant image from its parameters as the response.
Achieving this target presents notable challenges, including inbuilt visual
memory and visual recall schemes within MLLMs. To address these challenges, we
introduce a generative cross-modal retrieval framework, which assigns unique
identifier strings to represent images and involves two training steps:
learning to memorize and learning to retrieve. The first step focuses on
training the MLLM to memorize the association between images and their
respective identifiers. The latter step teaches the MLLM to generate the
corresponding identifier of the target image, given the textual query input. By
memorizing images in MLLMs, we introduce a new paradigm to cross-modal
retrieval, distinct from previous discriminative approaches. The experiments
demonstrate that the generative paradigm performs effectively and efficiently
even with large-scale image candidate sets. | cs.MM | Generative Cross-Modal Retrieval: Memorizing Images in Multimodal Language Models for Retrieval and Beyond |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.