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
|
---|---|---|---|---|---|---|---|---|
2024-07-02T15:01:55Z | http://arxiv.org/pdf/2407.02333v1 | http://arxiv.org/abs/2407.02333v1 | Musashi Hinck, Carolin Holtermann, Matthew Lyle Olson, Florian Schneider, Sungduk Yu, Anahita Bhiwandiwalla, Anne Lauscher, Shaoyen Tseng, Vasudev Lal | 2024-07-02T15:01:55Z | cs.CL, cs.CV | We uncover a surprising multilingual bias occurring in a popular class of
multimodal vision-language models (VLMs). Including an image in the query to a
LLaVA-style VLM significantly increases the likelihood of the model returning
an English response, regardless of the language of the query. This paper
investigates the causes of this loss with a two-pronged approach that combines
extensive ablation of the design space with a mechanistic analysis of the
models' internal representations of image and text inputs. Both approaches
indicate that the issue stems in the language modelling component of the LLaVA
model. Statistically, we find that switching the language backbone for a
bilingual language model has the strongest effect on reducing this error.
Mechanistically, we provide compelling evidence that visual inputs are not
mapped to a similar space as text ones, and that intervening on intermediary
attention layers can reduce this bias. Our findings provide important insights
to researchers and engineers seeking to understand the crossover between
multimodal and multilingual spaces, and contribute to the goal of developing
capable and inclusive VLMs for non-English contexts. | cs.CL | Why do LLaVA Vision-Language Models Reply to Images in English? |
2024-07-02T14:58:44Z | http://arxiv.org/pdf/2407.02328v1 | http://arxiv.org/abs/2407.02328v1 | Chaoran Zhang, Lixin Zou, Dan Luo, Min Tang, Xiangyang Luo, Zihao Li, Chenliang Li | 2024-07-02T14:58:44Z | cs.CL | In recent years, Large Language Models (LLMs) have demonstrated remarkable
capabilities across a wide array of text-centric tasks. However, their `large'
scale introduces significant computational and storage challenges, particularly
in managing the key-value states of the transformer, which limits their wider
applicability. Therefore, we propose to adaptively release resources from
caches and rebuild the necessary key-value states. Particularly, we accomplish
this by a lightweight controller module to approximate an ideal top-$K$ sparse
attention. This module retains the tokens with the highest top-$K$ attention
weights and simultaneously rebuilds the discarded but necessary tokens, which
may become essential for future decoding. Comprehensive experiments in natural
language generation and modeling reveal that our method is not only competitive
with full attention in terms of performance but also achieves a significant
throughput improvement of up to 221.8%. The code for replication is available
on the https://github.com/WHUIR/ADORE. | cs.CL | Efficient Sparse Attention needs Adaptive Token Release |
2024-07-02T14:51:20Z | http://arxiv.org/pdf/2407.02320v1 | http://arxiv.org/abs/2407.02320v1 | Chunlan Ma, Yihong Liu, Haotian Ye, Hinrich Schütze | 2024-07-02T14:51:20Z | cs.CL, cs.AI | Decoder-only large language models (LLMs) excel in high-resource languages
across various tasks through few-shot or even zero-shot in-context learning
(ICL). However, their performance often does not transfer well to low-resource
languages, especially those written in non-Latin scripts. Inspired by recent
work that leverages transliteration in encoder-only models, we investigate
whether transliteration is also effective in improving LLMs' performance for
low-resource languages written in non-Latin scripts. To this end, we propose
three prompt templates, where the target-language text is represented in (1)
its original script, (2) Latin script, or (3) both. We apply these methods to
several representative LLMs of different sizes on various tasks including text
classification and sequential labeling. Our findings show that the
effectiveness of transliteration varies by task type and model size. For
instance, all models benefit from transliterations for sequential labeling
(with increases of up to 25%). | cs.CL | Exploring the Role of Transliteration in In-Context Learning for Low-resource Languages Written in Non-Latin Scripts |
2024-07-02T14:50:03Z | http://arxiv.org/pdf/2407.02317v1 | http://arxiv.org/abs/2407.02317v1 | Ivan Vykopal, Simon Ostermann, Marián Šimko | 2024-07-02T14:50:03Z | cs.CL | Cross-lingual knowledge transfer, especially between high- and low-resource
languages, remains a challenge in natural language processing (NLP). This study
offers insights for improving cross-lingual NLP applications through the
combination of parameter-efficient fine-tuning methods. We systematically
explore strategies for enhancing this cross-lingual transfer through the
incorporation of language-specific and task-specific adapters and soft prompts.
We present a detailed investigation of various combinations of these methods,
exploring their efficiency across six languages, focusing on three low-resource
languages, including the to our knowledge first use of soft language prompts.
Our findings demonstrate that in contrast to claims of previous work, a
combination of language and task adapters does not always work best; instead,
combining a soft language prompt with a task adapter outperforms other
configurations in many cases. | cs.CL | Soft Language Prompts for Language Transfer |
2024-07-02T14:44:49Z | http://arxiv.org/pdf/2407.02310v1 | http://arxiv.org/abs/2407.02310v1 | Adrian Rebmann, Fabian David Schmidt, Goran Glavaš, Han van der Aa | 2024-07-02T14:44:49Z | cs.CL | The process mining community has recently recognized the potential of large
language models (LLMs) for tackling various process mining tasks. Initial
studies report the capability of LLMs to support process analysis and even, to
some extent, that they are able to reason about how processes work. This latter
property suggests that LLMs could also be used to tackle process mining tasks
that benefit from an understanding of process behavior. Examples of such tasks
include (semantic) anomaly detection and next activity prediction, which both
involve considerations of the meaning of activities and their inter-relations.
In this paper, we investigate the capabilities of LLMs to tackle such
semantics-aware process mining tasks. Furthermore, whereas most works on the
intersection of LLMs and process mining only focus on testing these models out
of the box, we provide a more principled investigation of the utility of LLMs
for process mining, including their ability to obtain process mining knowledge
post-hoc by means of in-context learning and supervised fine-tuning.
Concretely, we define three process mining tasks that benefit from an
understanding of process semantics and provide extensive benchmarking datasets
for each of them. Our evaluation experiments reveal that (1) LLMs fail to solve
challenging process mining tasks out of the box and when provided only a
handful of in-context examples, (2) but they yield strong performance when
fine-tuned for these tasks, consistently surpassing smaller, encoder-based
language models. | cs.CL | Evaluating the Ability of LLMs to Solve Semantics-Aware Process Mining Tasks |
2024-07-02T14:35:10Z | http://arxiv.org/pdf/2407.02302v1 | http://arxiv.org/abs/2407.02302v1 | Dominik Meier, Jan Philip Wahle, Terry Ruas, Bela Gipp | 2024-07-02T14:35:10Z | cs.CL, I.2.7 | Paraphrases represent a human's intuitive ability to understand expressions
presented in various different ways. Current paraphrase evaluations of language
models primarily use binary approaches, offering limited interpretability of
specific text changes. Atomic paraphrase types (APT) decompose paraphrases into
different linguistic changes and offer a granular view of the flexibility in
linguistic expression (e.g., a shift in syntax or vocabulary used). In this
study, we assess the human preferences towards ChatGPT in generating English
paraphrases with ten APTs and five prompting techniques. We introduce APTY
(Atomic Paraphrase TYpes), a dataset of 500 sentence-level and word-level
annotations by 15 annotators. The dataset also provides a human preference
ranking of paraphrases with different types that can be used to fine-tune
models with RLHF and DPO methods. Our results reveal that ChatGPT can generate
simple APTs, such as additions and deletions, but struggle with complex
structures (e.g., subordination changes). This study contributes to
understanding which aspects of paraphrasing language models have already
succeeded at understanding and what remains elusive. In addition, our curated
datasets can be used to develop language models with specific linguistic
capabilities. | cs.CL | Towards Human Understanding of Paraphrase Types in ChatGPT |
2024-07-02T14:34:36Z | http://arxiv.org/pdf/2407.02301v1 | http://arxiv.org/abs/2407.02301v1 | Ying Nie, Binwei Yan, Tianyu Guo, Hao Liu, Haoyu Wang, Wei He, Binfan Zheng, Weihao Wang, Qiang Li, Weijian Sun, Yunhe Wang, Dacheng Tao | 2024-07-02T14:34:36Z | cs.CL | Large language models (LLMs) have achieved remarkable performance on various
NLP tasks, yet their potential in more challenging and domain-specific task,
such as finance, has not been fully explored. In this paper, we present
CFinBench: a meticulously crafted, the most comprehensive evaluation benchmark
to date, for assessing the financial knowledge of LLMs under Chinese context.
In practice, to better align with the career trajectory of Chinese financial
practitioners, we build a systematic evaluation from 4 first-level categories:
(1) Financial Subject: whether LLMs can memorize the necessary basic knowledge
of financial subjects, such as economics, statistics and auditing. (2)
Financial Qualification: whether LLMs can obtain the needed financial qualified
certifications, such as certified public accountant, securities qualification
and banking qualification. (3) Financial Practice: whether LLMs can fulfill the
practical financial jobs, such as tax consultant, junior accountant and
securities analyst. (4) Financial Law: whether LLMs can meet the requirement of
financial laws and regulations, such as tax law, insurance law and economic
law. CFinBench comprises 99,100 questions spanning 43 second-level categories
with 3 question types: single-choice, multiple-choice and judgment. We conduct
extensive experiments of 50 representative LLMs with various model size on
CFinBench. The results show that GPT4 and some Chinese-oriented models lead the
benchmark, with the highest average accuracy being 60.16%, highlighting the
challenge presented by CFinBench. The dataset and evaluation code are available
at https://cfinbench.github.io/. | cs.CL | CFinBench: A Comprehensive Chinese Financial Benchmark for Large Language Models |
2024-07-02T14:14:59Z | http://arxiv.org/pdf/2407.02284v1 | http://arxiv.org/abs/2407.02284v1 | Arthur Amalvy, Vincent Labatut, Richard Dufour | 2024-07-02T14:14:59Z | cs.CL | Renard (Relationships Extraction from NARrative Documents) is a Python
library that allows users to define custom natural language processing (NLP)
pipelines to extract character networks from narrative texts. Contrary to the
few existing tools, Renard can extract dynamic networks, as well as the more
common static networks. Renard pipelines are modular: users can choose the
implementation of each NLP subtask needed to extract a character network. This
allows users to specialize pipelines to particular types of texts and to study
the impact of each subtask on the extracted network. | cs.CL | Renard: A Modular Pipeline for Extracting Character Networks from Narrative Texts |
2024-07-02T14:02:53Z | http://arxiv.org/pdf/2407.02273v1 | http://arxiv.org/abs/2407.02273v1 | Zhijing Jin, Sydney Levine, Max Kleiman-Weiner, Giorgio Piatti, Jiarui Liu, Fernando Gonzalez Adauto, Francesco Ortu, András Strausz, Mrinmaya Sachan, Rada Mihalcea, Yejin Choi, Bernhard Schölkopf | 2024-07-02T14:02:53Z | cs.CL | As large language models (LLMs) are deployed in more and more real-world
situations, it is crucial to understand their decision-making when faced with
moral dilemmas. Inspired by a large-scale cross-cultural study of human moral
preferences, "The Moral Machine Experiment", we set up the same set of moral
choices for LLMs. We translate 1K vignettes of moral dilemmas, parametrically
varied across key axes, into 100+ languages, and reveal the preferences of LLMs
in each of these languages. We then compare the responses of LLMs to that of
human speakers of those languages, harnessing a dataset of 40 million human
moral judgments. We discover that LLMs are more aligned with human preferences
in languages such as English, Korean, Hungarian, and Chinese, but less aligned
in languages such as Hindi and Somali (in Africa). Moreover, we characterize
the explanations LLMs give for their moral choices and find that fairness is
the most dominant supporting reason behind GPT-4's decisions and utilitarianism
by GPT-3. We also discover "language inequality" (which we define as the
model's different development levels in different languages) in a series of
meta-properties of moral decision making. | cs.CL | Multilingual Trolley Problems for Language Models |
2024-07-02T13:04:04Z | http://arxiv.org/pdf/2407.02243v1 | http://arxiv.org/abs/2407.02243v1 | Yuchen Hu, Chen Chen, Siyin Wang, Eng Siong Chng, Chao Zhang | 2024-07-02T13:04:04Z | cs.CL, cs.SD, eess.AS | In this paper, we propose reverse inference optimization (RIO), a simple and
effective method designed to enhance the robustness of
autoregressive-model-based zero-shot text-to-speech (TTS) systems using
reinforcement learning from human feedback (RLHF). To assess the quality of
speech produced by the TTS system without human annotations, RIO introduces a
novel concept termed as reverse inference based on the Bayesian principle,
which suggests that a high-quality generated speech should be able to be used
as a prompt for subsequent generation using the same TTS model. By leveraging
reverse inference as the standard to select exemplars used in RLHF from the
speech samples generated by the TTS system itself, RIO steers the subsequent
optimization towards a direction of enhancing the TTS robustness. The RIO
framework, comprising sampling, automatic annotating, and learning, obviates
the need for a reward model or pairwise preference data, and significantly
improves the stability of zero-shot TTS performance by reducing the
discrepancies between training and inference conditions. Our experimental
results verify that RIO can effectively improve both subjective and objective
metrics, including mean opinion scores, word error rates, and speaker
similarity. Remarkably, RIO can also diminish the incidence of bad outputs to
nearly zero percent, rivalling the robustness when using ground-truth speech as
the prompt. | cs.CL | Robust Zero-Shot Text-to-Speech Synthesis with Reverse Inference Optimization |
2024-07-02T12:58:35Z | http://arxiv.org/pdf/2407.02235v1 | http://arxiv.org/abs/2407.02235v1 | Cheng-Yi Li, Kao-Jung Chang, Cheng-Fu Yang, Hsin-Yu Wu, Wenting Chen, Hritik Bansal, Ling Chen, Yi-Ping Yang, Yu-Chun Chen, Shih-Pin Chen, Jiing-Feng Lirng, Kai-Wei Chang, Shih-Hwa Chiou | 2024-07-02T12:58:35Z | cs.CL | Multi-modal large language models (MLLMs) have been given free rein to
explore exciting medical applications with a primary focus on radiology report
generation. Nevertheless, the preliminary success in 2D radiology captioning is
incompetent to reflect the real-world diagnostic challenge in the volumetric 3D
anatomy. To mitigate three crucial limitation aspects in the existing
literature, including (1) data complexity, (2) model capacity, and (3)
evaluation metric fidelity, we collected an 18,885 text-scan pairs 3D-BrainCT
dataset and applied clinical visual instruction tuning (CVIT) to train BrainGPT
models to generate radiology-adherent 3D brain CT reports. Statistically, our
BrainGPT scored BLEU-1 = 44.35, BLEU-4 = 20.38, METEOR = 30.13, ROUGE-L = 47.6,
and CIDEr-R = 211.77 during internal testing and demonstrated an accuracy of
0.91 in captioning midline shifts on the external validation CQ500 dataset. By
further inspecting the captioned report, we reported that the traditional
metrics appeared to measure only the surface text similarity and failed to
gauge the information density of the diagnostic purpose. To close this gap, we
proposed a novel Feature-Oriented Radiology Task Evaluation (FORTE) to estimate
the report's clinical relevance (lesion feature and landmarks). Notably, the
BrainGPT model scored an average FORTE F1-score of 0.71 (degree=0.661;
landmark=0.706; feature=0.693; impression=0.779). To demonstrate that BrainGPT
models possess objective readiness to generate human-like radiology reports, we
conducted a Turing test that enrolled 11 physician evaluators, and around 74%
of the BrainGPT-generated captions were indistinguishable from those written by
humans. Our work embodies a holistic framework that showcased the first-hand
experience of curating a 3D brain CT dataset, fine-tuning anatomy-sensible
language models, and proposing robust radiology evaluation metrics. | cs.CL | Towards a Holistic Framework for Multimodal Large Language Models in Three-dimensional Brain CT Report Generation |
2024-07-02T12:57:42Z | http://arxiv.org/pdf/2407.02233v1 | http://arxiv.org/abs/2407.02233v1 | Ian Wu, Sravan Jayanthi, Vijay Viswanathan, Simon Rosenberg, Sina Pakazad, Tongshuang Wu, Graham Neubig | 2024-07-02T12:57:42Z | cs.CL, cs.AI, cs.LG | Multimodal Retrieval Augmented Generation (MMRAG) is a powerful approach to
question-answering over multimodal documents. A key challenge with evaluating
MMRAG is the paucity of high-quality datasets matching the question styles and
modalities of interest. In light of this, we propose SMMQG, a synthetic data
generation framework. SMMQG leverages interplay between a retriever, large
language model (LLM) and large multimodal model (LMM) to generate question and
answer pairs directly from multimodal documents, with the questions conforming
to specified styles and modalities. We use SMMQG to generate an MMRAG dataset
of 1024 questions over Wikipedia documents and evaluate state-of-the-art models
using it, revealing insights into model performance that are attainable only
through style- and modality-specific evaluation data. Next, we measure the
quality of data produced by SMMQG via a human study. We find that the quality
of our synthetic data is on par with the quality of the crowdsourced benchmark
MMQA and that downstream evaluation results using both datasets strongly
concur. | cs.CL | Synthetic Multimodal Question Generation |
2024-07-02T12:21:14Z | http://arxiv.org/pdf/2407.02211v1 | http://arxiv.org/abs/2407.02211v1 | Jiaru Zou, Mengyu Zhou, Tao Li, Shi Han, Dongmei Zhang | 2024-07-02T12:21:14Z | cs.CL, cs.AI, cs.LG | Large language models (LLMs) have played a fundamental role in various
natural language processing tasks with powerful prompt techniques. However, in
real-world applications, there are often similar prompt components for repeated
queries, which causes significant computational burdens during inference.
Existing prompt compression and direct fine-tuning methods aim to tackle these
challenges, yet they frequently struggle to strike an optimal balance between
cost-efficiency and performance effectiveness, especially in complex tasks such
as NL2Code. In this paper, we propose a novel method namely PromptIntern to
internalize the prompt knowledge into model parameters via progressive
fine-tuning. Our method enables LLMs to emulate the human learning process for
a new task, where detailed templates and examples in a prompt are gradually
internalized and phased out progressively as the model grows accustomed to the
task. Extensive experiments demonstrate that our method reduces inference
tokens over 90%, speedups inference by 4.2 times, and saves 88.3% monetary
cost. | cs.CL | PromptIntern: Saving Inference Costs by Internalizing Recurrent Prompt during Large Language Model Fine-tuning |
2024-07-02T12:17:07Z | http://arxiv.org/pdf/2407.02209v1 | http://arxiv.org/abs/2407.02209v1 | Fan Wu, Emily Black, Varun Chandrasekaran | 2024-07-02T12:17:07Z | cs.CL, cs.AI | We introduce {\em generative monoculture}, a behavior observed in large
language models (LLMs) characterized by a significant narrowing of model output
diversity relative to available training data for a given task: for example,
generating only positive book reviews for books with a mixed reception. While
in some cases, generative monoculture enhances performance (e.g., LLMs more
often produce efficient code), the dangers are exacerbated in others (e.g.,
LLMs refuse to share diverse opinions). As LLMs are increasingly used in
high-impact settings such as education and web search, careful maintenance of
LLM output diversity is essential to ensure a variety of facts and perspectives
are preserved over time. We experimentally demonstrate the prevalence of
generative monoculture through analysis of book review and code generation
tasks, and find that simple countermeasures such as altering sampling or
prompting strategies are insufficient to mitigate the behavior. Moreover, our
results suggest that the root causes of generative monoculture are likely
embedded within the LLM's alignment processes, suggesting a need for developing
fine-tuning paradigms that preserve or promote diversity. | cs.CL | Generative Monoculture in Large Language Models |
2024-07-02T12:15:15Z | http://arxiv.org/pdf/2407.02208v1 | http://arxiv.org/abs/2407.02208v1 | Yan Meng, Di Wu, Christof Monz | 2024-07-02T12:15:15Z | cs.CL, cs.AI | The massive amounts of web-mined parallel data contain large amounts of
noise. Semantic misalignment, as the primary source of the noise, poses a
challenge for training machine translation systems. In this paper, we first
study the impact of real-world hard-to-detect misalignment noise by proposing a
process to simulate the realistic misalignment controlled by semantic
similarity. After quantitatively analyzing the impact of simulated misalignment
on machine translation, we show the limited effectiveness of widely used
pre-filters to improve the translation performance, underscoring the necessity
of more fine-grained ways to handle data noise. By observing the increasing
reliability of the model's self-knowledge for distinguishing misaligned and
clean data at the token-level, we propose a self-correction approach which
leverages the model's prediction distribution to revise the training
supervision from the ground-truth data over training time. Through
comprehensive experiments, we show that our self-correction method not only
improves translation performance in the presence of simulated misalignment
noise but also proves effective for real-world noisy web-mined datasets across
eight translation tasks. | cs.CL | How to Learn in a Noisy World? Self-Correcting the Real-World Data Noise on Machine Translation |
2024-07-02T12:06:40Z | http://arxiv.org/pdf/2407.02203v1 | http://arxiv.org/abs/2407.02203v1 | Yusei Ishimizu, Jialong Li, Jinglue Xu, Jinyu Cai, Hitoshi Iba, Kenji Tei | 2024-07-02T12:06:40Z | cs.CL, cs.AI | Rule-based adaptation is a foundational approach to self-adaptation,
characterized by its human readability and rapid response. However, building
high-performance and robust adaptation rules is often a challenge because it
essentially involves searching the optimal design in a complex (variables)
space. In response, this paper attempt to employ large language models (LLMs)
as a optimizer to construct and optimize adaptation rules, leveraging the
common sense and reasoning capabilities inherent in LLMs. Preliminary
experiments conducted in SWIM have validated the effectiveness and limitation
of our method. | cs.CL | Automatic Adaptation Rule Optimization via Large Language Models |
2024-07-02T10:43:49Z | http://arxiv.org/pdf/2407.02147v1 | http://arxiv.org/abs/2407.02147v1 | Hasna Chouikhi, Manel Aloui, Cyrine Ben Hammou, Ghaith Chaabane, Haithem Kchaou, Chehir Dhaouadi | 2024-07-02T10:43:49Z | cs.CL, cs.AI | Large language models (LLMs) have greatly impacted the natural language
processing (NLP) field, particularly for the English language. These models
have demonstrated capabilities in understanding and generating human-like text.
The success of language models largely depends on the availability of
high-quality instruction datasets, which consist of detailed task descriptions
and corresponding responses that are essential for training the models to
accurately address a variety of prompts. However, the availability and quality
of these resources vary by language. While models perform well in English, they
often struggle with languages like Arabic, due to the lack of datasets for
fine-tuning Arabic-specific tasks. To address this issue, we introduce
InstAr-500k, a new Arabic instruction dataset created by generating and
collecting content that covers several domains and instruction types. We then
assess this dataset by fine-tuning two open-source models, Llama-3-8B-Instruct
and Gemma-7B-IT, on several downstream tasks to scale improvements in their
functionality. Based on multiple evaluations, our fine-tuned models achieve
state-of-the-art performance on several Arabic NLP benchmarks. These outcomes
emphasize the effectiveness of our dataset in elevating the capabilities of
language models for Arabic. Our instruction dataset bridges the performance gap
between English and Arabic language models by providing resources that amplify
Arabic NLP development. Building on this foundation, we developed two
state-of-the-art models, LlamAr-8B and GemmAr-7B, which are specifically tuned
to excel at a wide range of Arabic NLP tasks. | cs.CL | LlamAr & GemmAr: Enhancing LLMs Through Arabic Instruction-Tuning |
2024-07-02T10:33:31Z | http://arxiv.org/pdf/2407.02138v1 | http://arxiv.org/abs/2407.02138v1 | Wataru Hashimoto, Hidetaka Kamigaito, Taro Watanabe | 2024-07-02T10:33:31Z | cs.CL, cs.AI, cs.LG | Trustworthy prediction in Deep Neural Networks (DNNs), including Pre-trained
Language Models (PLMs) is important for safety-critical applications in the
real world. However, DNNs often suffer from uncertainty estimation, such as
miscalibration. In particular, approaches that require multiple stochastic
inference can mitigate this problem, but the expensive cost of inference makes
them impractical. In this study, we propose $k$-Nearest Neighbor Uncertainty
Estimation ($k$NN-UE), which is an uncertainty estimation method that uses the
distances from the neighbors and label-existence ratio of neighbors.
Experiments on sentiment analysis, natural language inference, and named entity
recognition show that our proposed method outperforms the baselines or recent
density-based methods in confidence calibration, selective prediction, and
out-of-distribution detection. Moreover, our analyses indicate that introducing
dimension reduction or approximate nearest neighbor search inspired by recent
$k$NN-LM studies reduces the inference overhead without significantly degrading
estimation performance when combined them appropriately. | cs.CL | Efficient Nearest Neighbor based Uncertainty Estimation for Natural Language Processing Tasks |
2024-07-02T10:29:09Z | http://arxiv.org/pdf/2407.02136v1 | http://arxiv.org/abs/2407.02136v1 | Jaap Jumelet, Lisa Bylinina, Willem Zuidema, Jakub Szymanik | 2024-07-02T10:29:09Z | cs.CL | In English and other languages, multiple adjectives in a complex noun phrase
show intricate ordering patterns that have been a target of much linguistic
theory. These patterns offer an opportunity to assess the ability of language
models (LMs) to learn subtle rules of language involving factors that cross the
traditional divisions of syntax, semantics, and pragmatics. We review existing
hypotheses designed to explain Adjective Order Preferences (AOPs) in humans and
develop a setup to study AOPs in LMs: we present a reusable corpus of adjective
pairs and define AOP measures for LMs. With these tools, we study a series of
LMs across intermediate checkpoints during training. We find that all models'
predictions are much closer to human AOPs than predictions generated by factors
identified in theoretical linguistics. At the same time, we demonstrate that
the observed AOPs in LMs are strongly correlated with the frequency of the
adjective pairs in the training data and report limited generalization to
unseen combinations. This highlights the difficulty in establishing the link
between LM performance and linguistic theory. We therefore conclude with a road
map for future studies our results set the stage for, and a discussion of key
questions about the nature of knowledge in LMs and their ability to generalize
beyond the training sets. | cs.CL | Black Big Boxes: Do Language Models Hide a Theory of Adjective Order? |
2024-07-02T10:12:06Z | http://arxiv.org/pdf/2407.02122v1 | http://arxiv.org/abs/2407.02122v1 | Soveatin Kuntur, Anna Wróblewska, Marcin Paprzycki, Maria Ganzha | 2024-07-02T10:12:06Z | cs.CL | This comprehensive survey serves as an indispensable resource for researchers
embarking on the journey of fake news detection. By highlighting the pivotal
role of dataset quality and diversity, it underscores the significance of these
elements in the effectiveness and robustness of detection models. The survey
meticulously outlines the key features of datasets, various labeling systems
employed, and prevalent biases that can impact model performance. Additionally,
it addresses critical ethical issues and best practices, offering a thorough
overview of the current state of available datasets. Our contribution to this
field is further enriched by the provision of GitHub repository, which
consolidates publicly accessible datasets into a single, user-friendly portal.
This repository is designed to facilitate and stimulate further research and
development efforts aimed at combating the pervasive issue of fake news. | cs.CL | Fake News Detection: It's All in the Data! |
2024-07-02T10:09:19Z | http://arxiv.org/pdf/2407.02119v1 | http://arxiv.org/abs/2407.02119v1 | Yifang Chen, Shuohang Wang, Ziyi Yang, Hiteshi Sharma, Nikos Karampatziakis, Donghan Yu, Kevin Jamieson, Simon Shaolei Du, Yelong Shen | 2024-07-02T10:09:19Z | cs.LG, cs.AI, cs.CL | Reinforcement learning with human feedback (RLHF), as a widely adopted
approach in current large language model pipelines, is \textit{bottlenecked by
the size of human preference data}. While traditional methods rely on offline
preference dataset constructions, recent approaches have shifted towards online
settings, where a learner uses a small amount of labeled seed data and a large
pool of unlabeled prompts to iteratively construct new preference data through
self-generated responses and high-quality reward/preference feedback. However,
most current online algorithms still focus on preference labeling during policy
model updating with given feedback oracles, which incurs significant expert
query costs. \textit{We are the first to explore cost-effective proxy reward
oracles construction strategies for further labeling preferences or rewards
with extremely limited labeled data and expert query budgets}. Our approach
introduces two key innovations: (1) on-policy query to avoid OOD and imbalance
issues in seed data, and (2) active learning to select the most informative
data for preference queries. Using these methods, we train a evaluation model
with minimal expert-labeled data, which then effectively labels nine times more
preference pairs for further RLHF training. For instance, our model using
Direct Preference Optimization (DPO) gains around over 1% average improvement
on AlpacaEval2, MMLU-5shot and MMLU-0shot, with only 1.7K query cost. Our
methodology is orthogonal to other direct expert query-based strategies and
therefore might be integrated with them to further reduce query costs. | cs.LG | Cost-Effective Proxy Reward Model Construction with On-Policy and Active Learning |
2024-07-02T10:06:41Z | http://arxiv.org/pdf/2407.02118v1 | http://arxiv.org/abs/2407.02118v1 | Wenzhen Zheng, Wenbo Pan, Xu Xu, Libo Qin, Li Yue, Ming Zhou | 2024-07-02T10:06:41Z | cs.CL | In recent years, Large Language Models (LLMs) have made significant strides
towards Artificial General Intelligence. However, training these models from
scratch requires substantial computational resources and vast amounts of text
data. In this paper, we explore an alternative approach to constructing an LLM
for a new language by continually pretraining (CPT) from existing pretrained
LLMs, instead of using randomly initialized parameters. Based on parallel
experiments on 40 model sizes ranging from 40M to 5B parameters, we find that
1) CPT converges faster and saves significant resources in a scalable manner;
2) CPT adheres to an extended scaling law derived from Hoffmann et al. (2022)
with a joint data-parameter scaling term; 3) The compute-optimal data-parameter
allocation for CPT markedly differs based on our estimated scaling factors; 4)
The effectiveness of transfer at scale is influenced by training duration and
linguistic properties, while robust to data replaying, a method that
effectively mitigates catastrophic forgetting in CPT. We hope our findings
provide deeper insights into the transferability of LLMs at scale for the
research community. | cs.CL | Breaking Language Barriers: Cross-Lingual Continual Pre-Training at Scale |
2024-07-02T09:36:54Z | http://arxiv.org/pdf/2407.02099v1 | http://arxiv.org/abs/2407.02099v1 | Pedro Henrique Luz de Araujo, Benjamin Roth | 2024-07-02T09:36:54Z | cs.CL | One way to personalize and steer generations from large language models (LLM)
is to assign a persona: a role that describes how the user expects the LLM to
behave (e.g., a helpful assistant, a teacher, a woman). This paper investigates
how personas affect diverse aspects of model behavior. We assign to seven LLMs
162 personas from 12 categories spanning variables like gender, sexual
orientation, and occupation. We prompt them to answer questions from five
datasets covering objective (e.g., questions about math and history) and
subjective tasks (e.g., questions about beliefs and values). We also compare
persona's generations to two baseline settings: a control persona setting with
30 paraphrases of "a helpful assistant" to control for models' prompt
sensitivity, and an empty persona setting where no persona is assigned. We find
that for all models and datasets, personas show greater variability than the
control setting and that some measures of persona behavior generalize across
models. | cs.CL | Helpful assistant or fruitful facilitator? Investigating how personas affect language model behavior |
2024-07-02T08:55:41Z | http://arxiv.org/pdf/2407.02067v1 | http://arxiv.org/abs/2407.02067v1 | Anjishnu Mukherjee, Ziwei Zhu, Antonios Anastasopoulos | 2024-07-02T08:55:41Z | cs.CL | In this work, we present a comprehensive three-phase study to examine (1) the
effectiveness of large multimodal models (LMMs) in recognizing cultural
contexts; (2) the accuracy of their representations of diverse cultures; and
(3) their ability to adapt content across cultural boundaries. We first
introduce Dalle Street, a large-scale dataset generated by DALL-E 3 and
validated by humans, containing 9,935 images of 67 countries and 10 concept
classes. We reveal disparities in cultural understanding at the sub-region
level with both open-weight (LLaVA) and closed-source (GPT-4V) models on Dalle
Street and other existing benchmarks. Next, we assess models' deeper culture
understanding by an artifact extraction task and identify over 18,000 artifacts
associated with different countries. Finally, we propose a highly composable
pipeline, CultureAdapt, to adapt images from culture to culture. Our findings
reveal a nuanced picture of the cultural competence of LMMs, highlighting the
need to develop culture-aware systems. Dataset and code are available at
https://github.com/iamshnoo/crossroads | cs.CL | Crossroads of Continents: Automated Artifact Extraction for Cultural Adaptation with Large Multimodal Models |
2024-07-02T08:55:40Z | http://arxiv.org/pdf/2407.02066v1 | http://arxiv.org/abs/2407.02066v1 | Chahat Raj, Anjishnu Mukherjee, Aylin Caliskan, Antonios Anastasopoulos, Ziwei Zhu | 2024-07-02T08:55:40Z | cs.CL | Existing works examining Vision Language Models (VLMs) for social biases
predominantly focus on a limited set of documented bias associations, such as
gender:profession or race:crime. This narrow scope often overlooks a vast range
of unexamined implicit associations, restricting the identification and, hence,
mitigation of such biases. We address this gap by probing VLMs to (1) uncover
hidden, implicit associations across 9 bias dimensions. We systematically
explore diverse input and output modalities and (2) demonstrate how biased
associations vary in their negativity, toxicity, and extremity. Our work (3)
identifies subtle and extreme biases that are typically not recognized by
existing methodologies. We make the Dataset of retrieved associations, (Dora),
publicly available here https://github.com/chahatraj/BiasDora. | cs.CL | BiasDora: Exploring Hidden Biased Associations in Vision-Language Models |
2024-07-02T08:49:43Z | http://arxiv.org/pdf/2407.02062v1 | http://arxiv.org/abs/2407.02062v1 | Wataru Hashimoto, Hidetaka Kamigaito, Taro Watanabe | 2024-07-02T08:49:43Z | cs.CL, cs.AI, cs.LG | This work investigates the impact of data augmentation on confidence
calibration and uncertainty estimation in Named Entity Recognition (NER) tasks.
For the future advance of NER in safety-critical fields like healthcare and
finance, it is essential to achieve accurate predictions with calibrated
confidence when applying Deep Neural Networks (DNNs), including Pre-trained
Language Models (PLMs), as a real-world application. However, DNNs are prone to
miscalibration, which limits their applicability. Moreover, existing methods
for calibration and uncertainty estimation are computational expensive. Our
investigation in NER found that data augmentation improves calibration and
uncertainty in cross-genre and cross-lingual setting, especially in-domain
setting. Furthermore, we showed that the calibration for NER tends to be more
effective when the perplexity of the sentences generated by data augmentation
is lower, and that increasing the size of the augmentation further improves
calibration and uncertainty. | cs.CL | Are Data Augmentation Methods in Named Entity Recognition Applicable for Uncertainty Estimation? |
2024-07-02T08:38:31Z | http://arxiv.org/pdf/2407.02056v1 | http://arxiv.org/abs/2407.02056v1 | Xinglin Wang, Yiwei Li, Shaoxiong Feng, Peiwen Yuan, Boyuan Pan, Heda Wang, Yao Hu, Kan Li | 2024-07-02T08:38:31Z | cs.CL, cs.AI | Self-consistency (SC), leveraging multiple samples from LLMs, shows
significant gains on various reasoning tasks but struggles with free-form
generation due to the difficulty of aggregating answers. Its variants, UCS and
USC, rely on sample selection or voting mechanisms to improve output quality.
These methods, however, face limitations due to their inability to fully
utilize the nuanced consensus knowledge present within multiple candidate
samples, often resulting in suboptimal outputs. We propose Fine-Grained
Self-Consistency (FSC) to addresses these limitations by extracting and
integrating segment-level commonalities from candidate samples, enhancing the
performance of LLMs both in open-ended and reasoning tasks. Based on this, we
present two additional strategies: candidate filtering, which enhances overall
quality by identifying highly similar candidate sets, and merging, which
reduces input token requirements by combining similar samples. The
effectiveness of FSC is demonstrated through extensive experiments on various
tasks, including summarization, code generation, and mathematical reasoning,
using GPT-3.5-turbo and GPT-4. The results indicate significant improvements
over baseline methods, showcasing the potential of FSC to optimize output
quality by effectively synthesizing fine-grained consensus knowledge from
multiple samples. | cs.CL | Integrate the Essence and Eliminate the Dross: Fine-Grained Self-Consistency for Free-Form Language Generation |
2024-07-02T08:23:38Z | http://arxiv.org/pdf/2407.02049v1 | http://arxiv.org/abs/2407.02049v1 | Ruiqi Li, Zhiqing Hong, Yongqi Wang, Lichao Zhang, Rongjie Huang, Siqi Zheng, Zhou Zhao | 2024-07-02T08:23:38Z | eess.AS, cs.CL, cs.SD | Text-to-song (TTSong) is a music generation task that synthesizes accompanied
singing voices. Current TTSong methods, inherited from singing voice synthesis
(SVS), require melody-related information that can sometimes be impractical,
such as music scores or MIDI sequences. We present MelodyLM, the first TTSong
model that generates high-quality song pieces with fully text-controlled
melodies, achieving minimal user requirements and maximum control flexibility.
MelodyLM explicitly models MIDI as the intermediate melody-related feature and
sequentially generates vocal tracks in a language model manner, conditioned on
textual and vocal prompts. The accompaniment music is subsequently synthesized
by a latent diffusion model with hybrid conditioning for temporal alignment.
With minimal requirements, users only need to input lyrics and a reference
voice to synthesize a song sample. For full control, just input textual prompts
or even directly input MIDI. Experimental results indicate that MelodyLM
achieves superior performance in terms of both objective and subjective
metrics. Audio samples are available at https://melodylm666.github.io. | eess.AS | Accompanied Singing Voice Synthesis with Fully Text-controlled Melody |
2024-07-02T08:17:00Z | http://arxiv.org/pdf/2407.02043v1 | http://arxiv.org/abs/2407.02043v1 | Yang Xu, Yunlong Feng, Honglin Mu, Yutai Hou, Yitong Li, Xinghao Wang, Wanjun Zhong, Zhongyang Li, Dandan Tu, Qingfu Zhu, Min Zhang, Wanxiang Che | 2024-07-02T08:17:00Z | cs.CL | Through reading the documentation in the context, tool-using language models
can dynamically extend their capability using external tools. The cost is that
we have to input lengthy documentation every time the model needs to use the
tool, occupying the input window as well as slowing down the decoding process.
Given the progress in general-purpose compression, soft context compression
is a suitable approach to alleviate the problem. However, when compressing tool
documentation, existing methods suffer from the weaknesses of key information
loss (specifically, tool/parameter name errors) and difficulty in adjusting the
length of compressed sequences based on documentation lengths.
To address these problems, we propose two strategies for compressing tool
documentation into concise and precise summary sequences for tool-using
language models. 1) Selective compression strategy mitigates key information
loss by deliberately retaining key information as raw text tokens. 2) Block
compression strategy involves dividing tool documentation into short chunks and
then employing a fixed-length compression model to achieve variable-length
compression. This strategy facilitates the flexible adjustment of the
compression ratio.
Results on API-Bank and APIBench show that our approach reaches a performance
comparable to the upper-bound baseline under up to 16x compression ratio. | cs.CL | Concise and Precise Context Compression for Tool-Using Language Models |
2024-07-02T08:16:43Z | http://arxiv.org/pdf/2407.02042v1 | http://arxiv.org/abs/2407.02042v1 | Ruihan Jin, Ruibo Fu, Zhengqi Wen, Shuai Zhang, Yukun Liu, Jianhua Tao | 2024-07-02T08:16:43Z | cs.CL, cs.AI | Fake news becomes a growing threat to information security and public opinion
with the rapid sprawl of media manipulation. Therefore, fake news detection
attracts widespread attention from academic community. Traditional fake news
detection models demonstrate remarkable performance on authenticity binary
classification but their ability to reason detailed faked traces based on the
news content remains under-explored. Furthermore, due to the lack of external
knowledge, the performance of existing methods on fact-related news is
questionable, leaving their practical implementation unclear. In this paper, we
propose a new multi-media research topic, namely manipulation reasoning.
Manipulation reasoning aims to reason manipulations based on news content. To
support the research, we introduce a benchmark for fake news detection and
manipulation reasoning, referred to as Human-centric and Fact-related Fake News
(HFFN). The benchmark highlights the centrality of human and the high factual
relevance, with detailed manual annotations. HFFN encompasses four realistic
domains with fake news samples generated through three manipulation approaches.
Moreover, a Multi-modal news Detection and Reasoning langUage Model (M-DRUM) is
presented not only to judge on the authenticity of multi-modal news, but also
raise analytical reasoning about potential manipulations. On the feature
extraction level, a cross-attention mechanism is employed to extract
fine-grained fusion features from multi-modal inputs. On the reasoning level, a
large vision-language model (LVLM) serves as the backbone to facilitate
fact-related reasoning. A two-stage training framework is deployed to better
activate the capacity of identification and reasoning. Comprehensive
experiments demonstrate that our model outperforms state-of-the-art (SOTA) fake
news detection models and powerful LVLMs like GPT-4 and LLaVA. | cs.CL | Fake News Detection and Manipulation Reasoning via Large Vision-Language Models |
2024-07-02T08:11:18Z | http://arxiv.org/pdf/2407.02039v1 | http://arxiv.org/abs/2407.02039v1 | Christopher Barrie, Elli Palaiologou, Petter Törnberg | 2024-07-02T08:11:18Z | cs.CL | Researchers are increasingly using language models (LMs) for text annotation.
These approaches rely only on a prompt telling the model to return a given
output according to a set of instructions. The reproducibility of LM outputs
may nonetheless be vulnerable to small changes in the prompt design. This calls
into question the replicability of classification routines. To tackle this
problem, researchers have typically tested a variety of semantically similar
prompts to determine what we call "prompt stability." These approaches remain
ad-hoc and task specific. In this article, we propose a general framework for
diagnosing prompt stability by adapting traditional approaches to intra- and
inter-coder reliability scoring. We call the resulting metric the Prompt
Stability Score (PSS) and provide a Python package PromptStability for its
estimation. Using six different datasets and twelve outcomes, we classify >150k
rows of data to: a) diagnose when prompt stability is low; and b) demonstrate
the functionality of the package. We conclude by providing best practice
recommendations for applied researchers. | cs.CL | Prompt Stability Scoring for Text Annotation with Large Language Models |
2024-07-02T07:58:46Z | http://arxiv.org/pdf/2407.02030v1 | http://arxiv.org/abs/2407.02030v1 | Chahat Raj, Anjishnu Mukherjee, Aylin Caliskan, Antonios Anastasopoulos, Ziwei Zhu | 2024-07-02T07:58:46Z | cs.CL | Large Language Models (LLMs) perpetuate social biases, reflecting prejudices
in their training data and reinforcing societal stereotypes and inequalities.
Our work explores the potential of the Contact Hypothesis, a concept from
social psychology for debiasing LLMs. We simulate various forms of social
contact through LLM prompting to measure their influence on the model's biases,
mirroring how intergroup interactions can reduce prejudices in social contexts.
We create a dataset of 108,000 prompts following a principled approach
replicating social contact to measure biases in three LLMs (LLaMA 2, Tulu, and
NousHermes) across 13 social bias dimensions. We propose a unique debiasing
technique, Social Contact Debiasing (SCD), that instruction-tunes these models
with unbiased responses to prompts. Our research demonstrates that LLM
responses exhibit social biases when subject to contact probing, but more
importantly, these biases can be significantly reduced by up to 40% in 1 epoch
of instruction tuning LLaMA 2 following our SCD strategy. Our code and data are
available at https://github.com/chahatraj/breakingbias. | cs.CL | Breaking Bias, Building Bridges: Evaluation and Mitigation of Social Biases in LLMs via Contact Hypothesis |
2024-07-02T07:52:30Z | http://arxiv.org/pdf/2407.02028v1 | http://arxiv.org/abs/2407.02028v1 | Xiang Li, Haoran Tang, Siyu Chen, Ziwei Wang, Ryan Chen, Marcin Abram | 2024-07-02T07:52:30Z | cs.CL, cs.AI, cs.IR, cs.LG | We measure the performance of in-context learning as a function of task
novelty and difficulty for open and closed questions. For that purpose, we
created a novel benchmark consisting of hard scientific questions, each paired
with a context of various relevancy. We show that counter-intuitively, a
context that is more aligned with the topic does not always help more than a
less relevant context. This effect is especially visible for open questions and
questions of high difficulty or novelty. This result reveals a fundamental
difference between the treatment of close-form and open-form questions by
large-language models and shows a need for a more robust evaluation of
in-context learning on the variety of different types of questions. It also
poses a new question of how to optimally select a context for large language
models, especially in the context of Retrieval Augmented Generation (RAG)
systems. Our results suggest that the answer to this question can be highly
application-dependent and might be contingent on factors including the format
of the question, the perceived difficulty level of the questions, and the
novelty or popularity of the information we seek. | cs.CL | Why does in-context learning fail sometimes? Evaluating in-context learning on open and closed questions |
2024-07-02T07:22:57Z | http://arxiv.org/pdf/2407.02005v1 | http://arxiv.org/abs/2407.02005v1 | Hengchao Shang, Zongyao Li, Jiaxin Guo, Shaojun Li, Zhiqiang Rao, Yuanchang Luo, Daimeng Wei, Hao Yang | 2024-07-02T07:22:57Z | cs.CL, cs.SD, eess.AS | Abstractive Speech Summarization (SSum) aims to generate human-like text
summaries from spoken content. It encounters difficulties in handling long
speech input and capturing the intricate cross-modal mapping between long
speech inputs and short text summaries. Research on large language models
(LLMs) and multimodal information fusion has provided new insights for
addressing these challenges. In this paper, we propose an end-to-end SSum model
that utilizes Q-Former as a connector for the audio-text modality and employs
LLMs to generate text summaries directly from speech features. We adopt a
multi-stage training approach that includes LLM based ASR and Text
Summarization (TSum) tasks as auxiliary tasks. ASR tasks are used to align
feature spaces and enhance the LLM's ability to handle longer speech. Then, we
utilize a curriculum learning strategy to facilitate the model's transition
from TSum to SSum. Finally, our model achieves competitive performance on the
How-2 dataset. | cs.CL | An End-to-End Speech Summarization Using Large Language Model |
2024-07-02T07:13:35Z | http://arxiv.org/pdf/2407.02543v1 | http://arxiv.org/abs/2407.02543v1 | Varun Krishna, Sriram Ganapathy | 2024-07-02T07:13:35Z | cs.CL, cs.AI, cs.LG, cs.SD, eess.AS | The popular frameworks for self-supervised learning of speech representations
have largely focused on frame-level masked prediction of speech regions. While
this has shown promising downstream task performance for speech recognition and
related tasks, this has largely ignored factors of speech that are encoded at
coarser level, like characteristics of the speaker or channel that remain
consistent through-out a speech utterance. In this work, we propose a framework
for Learning Disentangled Self Supervised (termed as Learn2Diss)
representations of speech, which consists of frame-level and an utterance-level
encoder modules. The two encoders are initially learned independently, where
the frame-level model is largely inspired by existing self supervision
techniques, thereby learning pseudo-phonemic representations, while the
utterance-level encoder is inspired by constrastive learning of pooled
embeddings, thereby learning pseudo-speaker representations. The joint learning
of these two modules consists of disentangling the two encoders using a mutual
information based criterion. With several downstream evaluation experiments, we
show that the proposed Learn2Diss achieves state-of-the-art results on a
variety of tasks, with the frame-level encoder representations improving
semantic tasks, while the utterance-level representations improve non-semantic
tasks. | cs.CL | Towards the Next Frontier in Speech Representation Learning Using Disentanglement |
2024-07-02T07:07:59Z | http://arxiv.org/pdf/2407.01994v1 | http://arxiv.org/abs/2407.01994v1 | Ananjan Nandi, Navdeep Kaur, Parag Singla, Mausam | 2024-07-02T07:07:59Z | cs.AI, cs.CL, cs.IR | High-quality and high-coverage rule sets are imperative to the success of
Neuro-Symbolic Knowledge Graph Completion (NS-KGC) models, because they form
the basis of all symbolic inferences. Recent literature builds neural models
for generating rule sets, however, preliminary experiments show that they
struggle with maintaining high coverage. In this work, we suggest three simple
augmentations to existing rule sets: (1) transforming rules to their abductive
forms, (2) generating equivalent rules that use inverse forms of constituent
relations and (3) random walks that propose new rules. Finally, we prune
potentially low quality rules. Experiments over four datasets and five
ruleset-baseline settings suggest that these simple augmentations consistently
improve results, and obtain up to 7.1 pt MRR and 8.5 pt Hits@1 gains over using
rules without augmentations. | cs.AI | Simple Augmentations of Logical Rules for Neuro-Symbolic Knowledge Graph Completion |
2024-07-02T07:06:53Z | http://arxiv.org/pdf/2407.01992v1 | http://arxiv.org/abs/2407.01992v1 | Nishant Balepur, Rachel Rudinger | 2024-07-02T07:06:53Z | cs.CL | Recent work shows that large language models (LLMs) can answer
multiple-choice questions using only the choices, but does this mean that MCQA
leaderboard rankings of LLMs are largely influenced by abilities in
choices-only settings? To answer this, we use a contrast set that probes if
LLMs over-rely on choices-only shortcuts in MCQA. While previous works build
contrast sets via expensive human annotations or model-generated data which can
be biased, we employ graph mining to extract contrast sets from existing MCQA
datasets. We use our method on UnifiedQA, a group of six commonsense reasoning
datasets with high choices-only accuracy, to build an 820-question contrast
set. After validating our contrast set, we test 12 LLMs, finding that these
models do not exhibit reliance on choice-only shortcuts when given both the
question and choices. Thus, despite the susceptibility~of MCQA to high
choices-only accuracy, we argue that LLMs are not obtaining high ranks on MCQA
leaderboards just due to their ability to exploit choices-only shortcuts. | cs.CL | Is Your Large Language Model Knowledgeable or a Choices-Only Cheater? |
2024-07-02T06:29:05Z | http://arxiv.org/pdf/2407.01976v1 | http://arxiv.org/abs/2407.01976v1 | Jinghui Lu, Haiyang Yu, Yanjie Wang, Yongjie Ye, Jingqun Tang, Ziwei Yang, Binghong Wu, Qi Liu, Hao Feng, Han Wang, Hao Liu, Can Huang | 2024-07-02T06:29:05Z | cs.CL, cs.AI, cs.MM | Recently, many studies have demonstrated that exclusively incorporating
OCR-derived text and spatial layouts with large language models (LLMs) can be
highly effective for document understanding tasks. However, existing methods
that integrate spatial layouts with text have limitations, such as producing
overly long text sequences or failing to fully leverage the autoregressive
traits of LLMs. In this work, we introduce Interleaving Layout and Text in a
Large Language Model (LayTextLLM)} for document understanding. In particular,
LayTextLLM projects each bounding box to a single embedding and interleaves it
with text, efficiently avoiding long sequence issues while leveraging
autoregressive traits of LLMs. LayTextLLM not only streamlines the interaction
of layout and textual data but also shows enhanced performance in Key
Information Extraction (KIE) and Visual Question Answering (VQA). Comprehensive
benchmark evaluations reveal significant improvements, with a 27.0% increase on
KIE tasks and 24.1% on VQA tasks compared to previous state-of-the-art document
understanding MLLMs, as well as a 15.5% improvement over other SOTA OCR-based
LLMs on KIE tasks. | cs.CL | A Bounding Box is Worth One Token: Interleaving Layout and Text in a Large Language Model for Document Understanding |
2024-07-02T05:50:16Z | http://arxiv.org/pdf/2407.01965v1 | http://arxiv.org/abs/2407.01965v1 | Yilong Lai, Jialong Wu, Congzhi Zhang, Haowen Sun, Deyu Zhou | 2024-07-02T05:50:16Z | cs.CL, cs.IR | Conversational Query Reformulation (CQR) has significantly advanced in
addressing the challenges of conversational search, particularly those stemming
from the latent user intent and the need for historical context. Recent works
aimed to boost the performance of CRQ through alignment. However, they are
designed for one specific retrieval system, which potentially results in poor
generalization. To overcome this limitation, we present a novel framework
AdaCQR. By aligning reformulation models with both term-based and
semantic-based retrieval systems, AdaCQR enhances the generalizability of
information-seeking queries across diverse retrieval environments through a
dual-phase training strategy. We also developed two effective approaches for
acquiring superior labels and diverse input candidates, boosting the efficiency
and robustness of the framework. Experimental evaluations on the TopiOCQA and
QReCC datasets demonstrate that AdaCQR significantly outperforms existing
methods, offering both quantitative and qualitative improvements in
conversational query reformulation. | cs.CL | AdaCQR: Enhancing Query Reformulation for Conversational Search via Sparse and Dense Retrieval Alignment |
2024-07-02T05:43:15Z | http://arxiv.org/pdf/2407.01964v2 | http://arxiv.org/abs/2407.01964v2 | Chenlong Deng, Kelong Mao, Yuyao Zhang, Zhicheng Dou | 2024-07-03T02:25:23Z | cs.CL | Legal judgment prediction is essential for enhancing judicial efficiency. In
this work, we identify that existing large language models (LLMs) underperform
in this domain due to challenges in understanding case complexities and
distinguishing between similar charges. To adapt LLMs for effective legal
judgment prediction, we introduce the Ask-Discriminate-Predict (ADAPT)
reasoning framework inspired by human judicial reasoning. ADAPT involves
decomposing case facts, discriminating among potential charges, and predicting
the final judgment. We further enhance LLMs through fine-tuning with multi-task
synthetic trajectories to improve legal judgment prediction accuracy and
efficiency under our ADAPT framework. Extensive experiments conducted on two
widely-used datasets demonstrate the superior performance of our framework in
legal judgment prediction, particularly when dealing with complex and confusing
charges. | cs.CL | Enabling Discriminative Reasoning in LLMs for Legal Judgment Prediction |
2024-07-02T05:14:15Z | http://arxiv.org/pdf/2407.01955v1 | http://arxiv.org/abs/2407.01955v1 | Parsa Kavehzadeh, Mohammadreza Pourreza, Mojtaba Valipour, Tinashu Zhu, Haoli Bai, Ali Ghodsi, Boxing Chen, Mehdi Rezagholizadeh | 2024-07-02T05:14:15Z | cs.CL | Deployment of autoregressive large language models (LLMs) is costly, and as
these models increase in size, the associated costs will become even more
considerable. Consequently, different methods have been proposed to accelerate
the token generation process and reduce costs. Speculative decoding (SD) is
among the most promising approaches to speed up the LLM decoding process by
verifying multiple tokens in parallel and using an auxiliary smaller draft
model to generate the possible tokens. In SD, usually, one draft model is used
to serve a specific target model; however, in practice, LLMs are diverse, and
we might need to deal with many target models or more than one target model
simultaneously. In this scenario, it is not clear which draft model should be
used for which target model, and searching among different draft models or
training customized draft models can further increase deployment costs. In this
paper, we first introduce a novel multi-target scenario for the deployment of
draft models for faster inference. Then, we present a novel, more efficient
sorted speculative decoding mechanism that outperforms regular baselines in
multi-target settings. We evaluated our method on Spec-Bench in different
settings, including base models such as Vicuna 7B, 13B, and LLama Chat 70B. Our
results suggest that our draft models perform better than baselines for
multiple target models at the same time. | cs.CL | S2D: Sorted Speculative Decoding For More Efficient Deployment of Nested Large Language Models |
2024-07-02T04:39:19Z | http://arxiv.org/pdf/2407.01948v1 | http://arxiv.org/abs/2407.01948v1 | Pablo Messina, René Vidal, Denis Parra, Álvaro Soto, Vladimir Araujo | 2024-07-02T04:39:19Z | cs.CL, cs.AI, cs.LG | Advancing representation learning in specialized fields like medicine remains
challenging due to the scarcity of expert annotations for text and images. To
tackle this issue, we present a novel two-stage framework designed to extract
high-quality factual statements from free-text radiology reports in order to
improve the representations of text encoders and, consequently, their
performance on various downstream tasks. In the first stage, we propose a
\textit{Fact Extractor} that leverages large language models (LLMs) to identify
factual statements from well-curated domain-specific datasets. In the second
stage, we introduce a \textit{Fact Encoder} (CXRFE) based on a BERT model
fine-tuned with objective functions designed to improve its representations
using the extracted factual data. Our framework also includes a new
embedding-based metric (CXRFEScore) for evaluating chest X-ray text generation
systems, leveraging both stages of our approach. Extensive evaluations show
that our fact extractor and encoder outperform current state-of-the-art methods
in tasks such as sentence ranking, natural language inference, and label
extraction from radiology reports. Additionally, our metric proves to be more
robust and effective than existing metrics commonly used in the radiology
report generation literature. The code of this project is available at
\url{https://github.com/PabloMessina/CXR-Fact-Encoder}. | cs.CL | Extracting and Encoding: Leveraging Large Language Models and Medical Knowledge to Enhance Radiological Text Representation |
2024-07-02T04:23:54Z | http://arxiv.org/pdf/2407.01942v1 | http://arxiv.org/abs/2407.01942v1 | Khyathi Raghavi Chandu, Linjie Li, Anas Awadalla, Ximing Lu, Jae Sung Park, Jack Hessel, Lijuan Wang, Yejin Choi | 2024-07-02T04:23:54Z | cs.AI, cs.CL, cs.CV | The ability to acknowledge the inevitable uncertainty in their knowledge and
reasoning is a prerequisite for AI systems to be truly truthful and reliable.
In this paper, we present a taxonomy of uncertainty specific to vision-language
AI systems, distinguishing between epistemic uncertainty (arising from a lack
of information) and aleatoric uncertainty (due to inherent unpredictability),
and further explore finer categories within. Based on this taxonomy, we
synthesize a benchmark dataset, CertainlyUncertain, featuring 178K visual
question answering (VQA) samples as contrastive pairs. This is achieved by 1)
inpainting images to make previously answerable questions into unanswerable
ones; and 2) using image captions to prompt large language models for both
answerable and unanswerable questions. Additionally, we introduce a new metric
confidence-weighted accuracy, that is well correlated with both accuracy and
calibration error, to address the shortcomings of existing metrics. | cs.AI | Certainly Uncertain: A Benchmark and Metric for Multimodal Epistemic and Aleatoric Awareness |
2024-07-02T04:11:52Z | http://arxiv.org/pdf/2407.01937v1 | http://arxiv.org/abs/2407.01937v1 | Linzhuang Sun, Hao Liang, Jingxuan Wei, Linkun Sun, Bihui Yu, Bin Cui, Wentao Zhang | 2024-07-02T04:11:52Z | cs.CL | In recent years, with the rapid advancements in large language models (LLMs),
achieving excellent empathetic response capability has become a crucial
prerequisite. Consequently, managing and understanding large-scale video
datasets has gained increasing importance. However, empathetic data are
typically trained without any quality selection, leading to inefficient data
usage and wasted computational resources. Additionally, using raw data can
result in low performance in empathetic dialogues. In this work, we present
Efficient-Empathy, a sensibility and rationality score-based data selection
algorithm that automatically selects sensibility and rationality data while
discarding low-quality data. With only the sensibility data (59% of the full
dataset), our trained sensibility model efficiently achieves state-of-the-art
(SoTA) performance. Furthermore, with multiple data selection hyperparameters,
the sensibility model demonstrates SoTA performance, showcasing the robustness
of our method. By integrating sensibility and rationality data with a MoE
structure, we achieve even higher performance, demonstrating the effectiveness
of our Efficient-Empathy algorithm. | cs.CL | Efficient-Empathy: Towards Efficient and Effective Selection of Empathy Data |
2024-07-02T03:45:55Z | http://arxiv.org/pdf/2407.01929v1 | http://arxiv.org/abs/2407.01929v1 | Shengqi Zhu, Jeffrey M. Rzeszotarski | 2024-07-02T03:45:55Z | cs.CL, cs.AI | The term Language Models (LMs), as a time-specific collection of models of
interest, is constantly reinvented, with its referents updated much like the
$\textit{Ship of Theseus}$ replaces its parts but remains the same ship in
essence. In this paper, we investigate this $\textit{Ship of Language Models}$
problem, wherein scientific evolution takes the form of continuous, implicit
retrofits of key existing terms. We seek to initiate a novel perspective of
scientific progress, in addition to the more well-studied emergence of new
terms. To this end, we construct the data infrastructure based on recent NLP
publications. Then, we perform a series of text-based analyses toward a
detailed, quantitative understanding of the use of Language Models as a term of
art. Our work highlights how systems and theories influence each other in
scientific discourse, and we call for attention to the transformation of this
Ship that we all are contributing to. | cs.CL | What We Talk About When We Talk About LMs: Implicit Paradigm Shifts and the Ship of Language Models |
2024-07-02T03:34:16Z | http://arxiv.org/pdf/2407.01920v1 | http://arxiv.org/abs/2407.01920v1 | Bozhong Tian, Xiaozhuan Liang, Siyuan Cheng, Qingbin Liu, Mengru Wang, Dianbo Sui, Xi Chen, Huajun Chen, Ningyu Zhang | 2024-07-02T03:34:16Z | cs.CL, cs.AI, cs.CV, cs.LG, cs.MM | Large Language Models (LLMs) trained on extensive corpora inevitably retain
sensitive data, such as personal privacy information and copyrighted material.
Recent advancements in knowledge unlearning involve updating LLM parameters to
erase specific knowledge. However, current unlearning paradigms are mired in
vague forgetting boundaries, often erasing knowledge indiscriminately. In this
work, we introduce KnowUnDo, a benchmark containing copyrighted content and
user privacy domains to evaluate if the unlearning process inadvertently erases
essential knowledge. Our findings indicate that existing unlearning methods
often suffer from excessive unlearning. To address this, we propose a simple
yet effective method, MemFlex, which utilizes gradient information to precisely
target and unlearn sensitive parameters. Experimental results show that MemFlex
is superior to existing methods in both precise knowledge unlearning and
general knowledge retaining of LLMs. Code and dataset will be released at
https://github.com/zjunlp/KnowUnDo. | cs.CL | To Forget or Not? Towards Practical Knowledge Unlearning for Large Language Models |
2024-07-02T03:22:41Z | http://arxiv.org/pdf/2407.01911v1 | http://arxiv.org/abs/2407.01911v1 | Yu-Kuan Fu, Cheng-Kuang Lee, Hsiu-Hsuan Wang, Hung-yi Lee | 2024-07-02T03:22:41Z | cs.CL, cs.HC, cs.SD, eess.AS | Recent efforts in Spoken Dialogue Modeling aim to synthesize spoken dialogue
without the need for direct transcription, thereby preserving the wealth of
non-textual information inherent in speech. However, this approach faces a
challenge when speakers talk simultaneously, requiring stereo dialogue data
with speakers recorded on separate channels, a notably scarce resource. To
address this, we have developed an innovative pipeline capable of transforming
single-channel dialogue data into pseudo-stereo data. This expanded our
training dataset from a mere 2,000 to an impressive 17,600 hours, significantly
enriching the diversity and quality of the training examples available. The
inclusion of this pseudo-stereo data has proven to be effective in improving
the performance of spoken dialogue language models. Additionally, we explored
the use of discrete units of different speech foundation models for spoken
dialogue generation. | cs.CL | Investigating the Effects of Large-Scale Pseudo-Stereo Data and Different Speech Foundation Model on Dialogue Generative Spoken Language Model |
2024-07-02T03:16:47Z | http://arxiv.org/pdf/2407.01909v1 | http://arxiv.org/abs/2407.01909v1 | Zhiyuan Tang, Dong Wang, Shen Huang, Shidong Shang | 2024-07-02T03:16:47Z | cs.CL, cs.SD, eess.AS | Recent studies have demonstrated the efficacy of large language models (LLMs)
in error correction for automatic speech recognition (ASR). However, much of
the research focuses on the English language. This paper redirects the
attention to Chinese. Firstly, we construct a specialized benchmark dataset
aimed at error correction for Chinese ASR with 724K hypotheses-transcription
pairs, named the Chinese Hypotheses Paradise dataset (ChineseHP), which
contains a wide range of scenarios and presents significant challenges.
Subsequently, we conduct a preliminary evaluation using the dataset for both
direct-prompting and fine-tuning pre-trained LLMs. Furthermore, we propose a
straightforward method of Pinyin regularization for prompts, which involves the
transcription of Pinyin directly from text hypotheses. The experimental results
reveal that Pinyin regularization consistently enhances the error-correcting
ability of LLMs when compared with those without regularization. The dataset is
available on the website. | cs.CL | Pinyin Regularization in Error Correction for Chinese Speech Recognition with Large Language Models |
2024-07-02T03:11:13Z | http://arxiv.org/pdf/2407.01906v1 | http://arxiv.org/abs/2407.01906v1 | Zihan Wang, Deli Chen, Damai Dai, Runxin Xu, Zhuoshu Li, Y. Wu | 2024-07-02T03:11:13Z | cs.CL, cs.AI, cs.LG | Parameter-efficient fine-tuning (PEFT) is crucial for customizing Large
Language Models (LLMs) with constrained resources. Although there have been
various PEFT methods for dense-architecture LLMs, PEFT for sparse-architecture
LLMs is still underexplored. In this work, we study the PEFT method for LLMs
with the Mixture-of-Experts (MoE) architecture and the contents of this work
are mainly threefold: (1) We investigate the dispersion degree of the activated
experts in customized tasks, and found that the routing distribution for a
specific task tends to be highly concentrated, while the distribution of
activated experts varies significantly across different tasks. (2) We propose
Expert-Specialized Fine-Tuning, or ESFT, which tunes the experts most relevant
to downstream tasks while freezing the other experts and modules; experimental
results demonstrate that our method not only improves the tuning efficiency,
but also matches or even surpasses the performance of full-parameter
fine-tuning. (3) We further analyze the impact of the MoE architecture on
expert-specialized fine-tuning. We find that MoE models with finer-grained
experts are more advantageous in selecting the combination of experts that are
most relevant to downstream tasks, thereby enhancing both the training
efficiency and effectiveness. | cs.CL | Let the Expert Stick to His Last: Expert-Specialized Fine-Tuning for Sparse Architectural Large Language Models |
2024-07-02T02:58:29Z | http://arxiv.org/pdf/2407.01902v1 | http://arxiv.org/abs/2407.01902v1 | Yan Yang, Zeguan Xiao, Xin Lu, Hongru Wang, Hailiang Huang, Guanhua Chen, Yun Chen | 2024-07-02T02:58:29Z | cs.CR, cs.AI, cs.CL | The widespread applications of large language models (LLMs) have brought
about concerns regarding their potential misuse. Although aligned with human
preference data before release, LLMs remain vulnerable to various malicious
attacks. In this paper, we adopt a red-teaming strategy to enhance LLM safety
and introduce SoP, a simple yet effective framework to design jailbreak prompts
automatically. Inspired by the social facilitation concept, SoP generates and
optimizes multiple jailbreak characters to bypass the guardrails of the target
LLM. Different from previous work which relies on proprietary LLMs or seed
jailbreak templates crafted by human expertise, SoP can generate and optimize
the jailbreak prompt in a cold-start scenario using open-sourced LLMs without
any seed jailbreak templates. Experimental results show that SoP achieves
attack success rates of 88% and 60% in bypassing the safety alignment of
GPT-3.5-1106 and GPT-4, respectively. Furthermore, we extensively evaluate the
transferability of the generated templates across different LLMs and held-out
malicious requests, while also exploring defense strategies against the
jailbreak attack designed by SoP. Code is available at
https://github.com/Yang-Yan-Yang-Yan/SoP. | cs.CR | SoP: Unlock the Power of Social Facilitation for Automatic Jailbreak Attack |
2024-07-02T02:50:15Z | http://arxiv.org/pdf/2407.01899v1 | http://arxiv.org/abs/2407.01899v1 | Xiulin Yang, Jonas Groschwitz, Alexander Koller, Johan Bos | 2024-07-02T02:50:15Z | cs.CL | Discourse Representation Theory (DRT) distinguishes itself from other
semantic representation frameworks by its ability to model complex semantic and
discourse phenomena through structural nesting and variable binding. While
seq2seq models hold the state of the art on DRT parsing, their accuracy
degrades with the complexity of the sentence, and they sometimes struggle to
produce well-formed DRT representations. We introduce the AMS parser, a
compositional, neurosymbolic semantic parser for DRT. It rests on a novel
mechanism for predicting quantifier scope. We show that the AMS parser reliably
produces well-formed outputs and performs well on DRT parsing, especially on
complex sentences. | cs.CL | Scope-enhanced Compositional Semantic Parsing for DRT |
2024-07-02T02:42:29Z | http://arxiv.org/pdf/2407.01897v1 | http://arxiv.org/abs/2407.01897v1 | Pengpeng Li, Tingmin Li, Jingyuan Wang, Boyuan Wang, Yang Yang | 2024-07-02T02:42:29Z | cs.CL | In this paper, we propose a method for document summarization using auxiliary
information. This approach effectively summarizes descriptions related to
specific images, tables, and appendices within lengthy texts. Our experiments
demonstrate that leveraging high-quality OCR data and initially extracted
information from the original text enables efficient summarization of the
content related to described objects. Based on these findings, we enhanced
popular text generation model models by incorporating additional auxiliary
branches to improve summarization performance. Our method achieved top scores
of 4.33 and 4.66 in the long caption and short caption tracks, respectively, of
the 2024 SciCAP competition, ranking highest in both categories. | cs.CL | Proposal Report for the 2nd SciCAP Competition 2024 |
2024-07-02T02:39:33Z | http://arxiv.org/pdf/2407.01896v1 | http://arxiv.org/abs/2407.01896v1 | Tianyu Cui, Shiyu Ma, Ziang Chen, Tong Xiao, Shimin Tao, Yilun Liu, Shenglin Zhang, Duoming Lin, Changchang Liu, Yuzhe Cai, Weibin Meng, Yongqian Sun, Dan Pei | 2024-07-02T02:39:33Z | cs.CL, cs.IR | Log analysis is crucial for ensuring the orderly and stable operation of
information systems, particularly in the field of Artificial Intelligence for
IT Operations (AIOps). Large Language Models (LLMs) have demonstrated
significant potential in natural language processing tasks. In the AIOps
domain, they excel in tasks such as anomaly detection, root cause analysis of
faults, operations and maintenance script generation, and alert information
summarization. However, the performance of current LLMs in log analysis tasks
remains inadequately validated. To address this gap, we introduce LogEval, a
comprehensive benchmark suite designed to evaluate the capabilities of LLMs in
various log analysis tasks for the first time. This benchmark covers tasks such
as log parsing, log anomaly detection, log fault diagnosis, and log
summarization. LogEval evaluates each task using 4,000 publicly available log
data entries and employs 15 different prompts for each task to ensure a
thorough and fair assessment. By rigorously evaluating leading LLMs, we
demonstrate the impact of various LLM technologies on log analysis performance,
focusing on aspects such as self-consistency and few-shot contextual learning.
We also discuss findings related to model quantification, Chinese-English
question-answering evaluation, and prompt engineering. These findings provide
insights into the strengths and weaknesses of LLMs in multilingual environments
and the effectiveness of different prompt strategies. Various evaluation
methods are employed for different tasks to accurately measure the performance
of LLMs in log analysis, ensuring a comprehensive assessment. The insights
gained from LogEvals evaluation reveal the strengths and limitations of LLMs in
log analysis tasks, providing valuable guidance for researchers and
practitioners. | cs.CL | LogEval: A Comprehensive Benchmark Suite for Large Language Models In Log Analysis |
2024-07-02T02:27:46Z | http://arxiv.org/pdf/2407.01892v1 | http://arxiv.org/abs/2407.01892v1 | Zhisheng Tang, Mayank Kejriwal | 2024-07-02T02:27:46Z | cs.AI, cs.CL | Spatial reasoning, an important faculty of human cognition with many
practical applications, is one of the core commonsense skills that is not
purely language-based and, for satisfying (as opposed to optimal) solutions,
requires some minimum degree of planning. Existing benchmarks of Commonsense
Spatial Reasoning (CSR) tend to evaluate how Large Language Models (LLMs)
interpret text-based spatial descriptions rather than directly evaluate a plan
produced by the LLM in response to a spatial reasoning scenario. In this paper,
we construct a large-scale benchmark called $\textbf{GRASP}$, which consists of
16,000 grid-based environments where the agent is tasked with an energy
collection problem. These environments include 100 grid instances instantiated
using each of the 160 different grid settings, involving five different energy
distributions, two modes of agent starting position, and two distinct obstacle
configurations, as well as three kinds of agent constraints. Using GRASP, we
compare classic baseline approaches, such as random walk and greedy search
methods, with advanced LLMs like GPT-3.5-Turbo and GPT-4o. The experimental
results indicate that even these advanced LLMs struggle to consistently achieve
satisfactory solutions. | cs.AI | GRASP: A Grid-Based Benchmark for Evaluating Commonsense Spatial Reasoning |
2024-07-02T02:18:14Z | http://arxiv.org/pdf/2407.01887v1 | http://arxiv.org/abs/2407.01887v1 | Fanzeng Xia, Hao Liu, Yisong Yue, Tongxin Li | 2024-07-02T02:18:14Z | cs.LG, cs.AI, cs.CL | In-context decision-making is an important capability of artificial general
intelligence, which Large Language Models (LLMs) have effectively demonstrated
in various scenarios. However, LLMs often face challenges when dealing with
numerical contexts, and limited attention has been paid to evaluating their
performance through preference feedback generated by the environment. This
paper investigates the performance of LLMs as decision-makers in the context of
Dueling Bandits (DB). We first evaluate the performance of LLMs by comparing
GPT-3.5-Turbo, GPT-4, and GPT-4-Turbo against established DB algorithms. Our
results reveal that LLMs, particularly GPT-4 Turbo, quickly identify the
Condorcet winner, thus outperforming existing state-of-the-art algorithms in
terms of weak regret. Nevertheless, LLMs struggle to converge even when
explicitly prompted to do so, and are sensitive to prompt variations. To
overcome these issues, we introduce an LLM-augmented algorithm, IF-Enhanced
LLM, which takes advantage of both in-context decision-making capabilities of
LLMs and theoretical guarantees inherited from classic DB algorithms. The
design of such an algorithm sheds light on how to enhance trustworthiness for
LLMs used in decision-making tasks where performance robustness matters. We
show that IF-Enhanced LLM has theoretical guarantees on both weak and strong
regret. Our experimental results validate that IF-Enhanced LLM is robust even
with noisy and adversarial prompts. | cs.LG | Beyond Numeric Awards: In-Context Dueling Bandits with LLM Agents |
2024-07-02T02:14:42Z | http://arxiv.org/pdf/2407.01885v1 | http://arxiv.org/abs/2407.01885v1 | Chuanpeng Yang, Wang Lu, Yao Zhu, Yidong Wang, Qian Chen, Chenlong Gao, Bingjie Yan, Yiqiang Chen | 2024-07-02T02:14:42Z | cs.CL, cs.AI | Large Language Models (LLMs) have showcased exceptional capabilities in
various domains, attracting significant interest from both academia and
industry. Despite their impressive performance, the substantial size and
computational demands of LLMs pose considerable challenges for practical
deployment, particularly in environments with limited resources. The endeavor
to compress language models while maintaining their accuracy has become a focal
point of research. Among the various methods, knowledge distillation has
emerged as an effective technique to enhance inference speed without greatly
compromising performance. This paper presents a thorough survey from three
aspects: method, evaluation, and application, exploring knowledge distillation
techniques tailored specifically for LLMs. Specifically, we divide the methods
into white-box KD and black-box KD to better illustrate their differences.
Furthermore, we also explored the evaluation tasks and distillation effects
between different distillation methods, and proposed directions for future
research. Through in-depth understanding of the latest advancements and
practical applications, this survey provides valuable resources for
researchers, paving the way for sustained progress in this field. | cs.CL | Survey on Knowledge Distillation for Large Language Models: Methods, Evaluation, and Application |
2024-07-02T01:37:56Z | http://arxiv.org/pdf/2407.01878v1 | http://arxiv.org/abs/2407.01878v1 | Sayan Ghosh, Tejas Srinivasan, Swabha Swayamdipta | 2024-07-02T01:37:56Z | cs.CL | Human evaluation of generated language through pairwise preference judgments
is pervasive. However, under common scenarios, such as when generations from a
model pair are very similar, or when stochastic decoding results in large
variations in generations, it results in inconsistent preference ratings. We
address these challenges by introducing a meta-evaluation measure,
separability, which estimates how suitable a test instance is for pairwise
preference evaluation. For a candidate test instance, separability samples
multiple generations from a pair of models, and measures how distinguishable
the two sets of generations are. Our experiments show that instances with high
separability values yield more consistent preference ratings from both human-
and auto-raters. Further, the distribution of separability allows insights into
which test benchmarks are more valuable for comparing models. Finally, we
incorporate separability into ELO ratings, accounting for how suitable each
test instance might be for reliably ranking LLMs. Overall, separability has
implications for consistent, efficient and robust preference evaluation of LLMs
with both human- and auto-raters. | cs.CL | Compare without Despair: Reliable Preference Evaluation with Generation Separability |
2024-07-02T01:17:01Z | http://arxiv.org/pdf/2407.01873v1 | http://arxiv.org/abs/2407.01873v1 | Christopher Michael Ormerod, Alexander Kwako | 2024-07-02T01:17:01Z | cs.LG, cs.AI, cs.CL | Current research on generative language models (GLMs) for automated text
scoring (ATS) has focused almost exclusively on querying proprietary models via
Application Programming Interfaces (APIs). Yet such practices raise issues
around transparency and security, and these methods offer little in the way of
efficiency or customizability. With the recent proliferation of smaller,
open-source models, there is the option to explore GLMs with computers equipped
with modest, consumer-grade hardware, that is, for the "GPU poor." In this
study, we analyze the performance and efficiency of open-source, small-scale
GLMs for ATS. Results show that GLMs can be fine-tuned to achieve adequate,
though not state-of-the-art, performance. In addition to ATS, we take small
steps towards analyzing models' capacity for generating feedback by prompting
GLMs to explain their scores. Model-generated feedback shows promise, but
requires more rigorous evaluation focused on targeted use cases. | cs.LG | Automated Text Scoring in the Age of Generative AI for the GPU-poor |
2024-07-02T00:24:01Z | http://arxiv.org/pdf/2407.01863v1 | http://arxiv.org/abs/2407.01863v1 | Qiucheng Wu, Handong Zhao, Michael Saxon, Trung Bui, William Yang Wang, Yang Zhang, Shiyu Chang | 2024-07-02T00:24:01Z | cs.CL | Vision language models (VLMs) are an exciting emerging class of language
models (LMs) that have merged classic LM capabilities with those of image
processing systems. However, the ways that these capabilities combine are not
always intuitive and warrant direct investigation. One understudied capability
in VLMs is visual spatial planning -- the ability to comprehend the spatial
arrangements of objects and devise action plans to achieve desired outcomes in
visual scenes. In our study, we introduce VSP, a benchmark that 1) evaluates
the spatial planning capability in these models in general, and 2) breaks down
the visual planning task into finer-grained sub-tasks, including perception and
reasoning, and measure the LMs capabilities in these sub-tasks. Our evaluation
shows that both open-source and private VLMs fail to generate effective plans
for even simple spatial planning tasks. Evaluations on the fine-grained
analytical tasks further reveal fundamental deficiencies in the models' visual
perception and bottlenecks in reasoning abilities, explaining their worse
performance in the general spatial planning tasks. Our work illuminates future
directions for improving VLMs' abilities in spatial planning. Our benchmark is
publicly available at
https://github.com/UCSB-NLP-Chang/Visual-Spatial-Planning. | cs.CL | VSP: Assessing the dual challenges of perception and reasoning in spatial planning tasks for VLMs |
2024-07-01T23:47:09Z | http://arxiv.org/pdf/2407.01853v1 | http://arxiv.org/abs/2407.01853v1 | Sathish Reddy Indurthi, Wenxuan Zhou, Shamil Chollampatt, Ravi Agrawal, Kaiqiang Song, Lingxiao Zhao, Chenguang Zhu | 2024-07-01T23:47:09Z | cs.CL, cs.AI, cs.LG | Advancements in Large Language Models (LLMs) have significantly enhanced
instruction-following capabilities. However, most Instruction Fine-Tuning (IFT)
datasets are predominantly in English, limiting model performance in other
languages. Traditional methods for creating multilingual IFT datasets such as
translating existing English IFT datasets or converting existing NLP datasets
into IFT datasets by templating, struggle to capture linguistic nuances and
ensure prompt (instruction) diversity. To address this issue, we propose a
novel method for collecting multilingual IFT datasets that preserves linguistic
naturalness and ensures prompt diversity. This approach leverages
English-focused LLMs, monolingual corpora, and a scoring function to create
high-quality, diversified IFT datasets in multiple languages. Experiments
demonstrate that LLMs finetuned using these IFT datasets show notable
improvements in both generative and discriminative tasks, indicating enhanced
language comprehension by LLMs in non-English contexts. Specifically, on the
multilingual summarization task, LLMs using our IFT dataset achieved 17.57% and
15.23% improvements over LLMs fine-tuned with translation-based and
template-based datasets, respectively. | cs.CL | Improving Multilingual Instruction Finetuning via Linguistically Natural and Diverse Datasets |
2024-07-01T23:25:30Z | http://arxiv.org/pdf/2407.01850v1 | http://arxiv.org/abs/2407.01850v1 | Jingyan Zhou, Kun Li, Junan Li, Jiawen Kang, Minda Hu, Xixin Wu, Helen Meng | 2024-07-01T23:25:30Z | cs.CL | Existing efforts in safeguarding LLMs are limited in actively exposing the
vulnerabilities of the target LLM and readily adapting to newly emerging safety
risks. To address this, we present Purple-teaming LLMs with Adversarial
Defender training (PAD), a pipeline designed to safeguard LLMs by novelly
incorporating the red-teaming (attack) and blue-teaming (safety training)
techniques. In PAD, we automatically collect conversational data that cover the
vulnerabilities of an LLM around specific safety risks in a self-play manner,
where the attacker aims to elicit unsafe responses and the defender generates
safe responses to these attacks. We then update both modules in a generative
adversarial network style by training the attacker to elicit more unsafe
responses and updating the defender to identify them and explain the unsafe
reason. Experimental results demonstrate that PAD significantly outperforms
existing baselines in both finding effective attacks and establishing a robust
safe guardrail. Furthermore, our findings indicate that PAD excels in striking
a balance between safety and overall model quality. We also reveal key
challenges in safeguarding LLMs, including defending multi-turn attacks and the
need for more delicate strategies to identify specific risks. | cs.CL | Purple-teaming LLMs with Adversarial Defender Training |
2024-07-01T22:17:17Z | http://arxiv.org/pdf/2407.01834v1 | http://arxiv.org/abs/2407.01834v1 | Valentin Barriere, Sebastian Cifuentes | 2024-07-01T22:17:17Z | cs.CL | In this paper, we apply a method to quantify biases associated with named
entities from various countries. We create counterfactual examples with small
perturbations on target-domain data instead of relying on templates or specific
datasets for bias detection. On widely used classifiers for subjectivity
analysis, including sentiment, emotion, hate speech, and offensive text using
Twitter data, our results demonstrate positive biases related to the language
spoken in a country across all classifiers studied. Notably, the presence of
certain country names in a sentence can strongly influence predictions, up to a
23\% change in hate speech detection and up to a 60\% change in the prediction
of negative emotions such as anger. We hypothesize that these biases stem from
the training data of pre-trained language models (PLMs) and find correlations
between affect predictions and PLMs likelihood in English and unknown languages
like Basque and Maori, revealing distinct patterns with exacerbate
correlations. Further, we followed these correlations in-between counterfactual
examples from a same sentence to remove the syntactical component, uncovering
interesting results suggesting the impact of the pre-training data was more
important for English-speaking-country names. Our anonymized code is
[https://anonymous.4open.science/r/biases_ppl-576B/README.md](available here). | cs.CL | A Study of Nationality Bias in Names and Perplexity using Off-the-Shelf Affect-related Tweet Classifiers |
2024-07-01T21:46:30Z | http://arxiv.org/pdf/2407.01824v1 | http://arxiv.org/abs/2407.01824v1 | Mehdi Arjmand, Farnaz Nouraei, Ian Steenstra, Timothy Bickmore | 2024-07-01T21:46:30Z | cs.HC, cs.CL, cs.RO | We introduce the concept of "empathic grounding" in conversational agents as
an extension of Clark's conceptualization of grounding in conversation in which
the grounding criterion includes listener empathy for the speaker's affective
state. Empathic grounding is generally required whenever the speaker's emotions
are foregrounded and can make the grounding process more efficient and reliable
by communicating both propositional and affective understanding. Both speaker
expressions of affect and listener empathic grounding can be multimodal,
including facial expressions and other nonverbal displays. Thus, models of
empathic grounding for embodied agents should be multimodal to facilitate
natural and efficient communication. We describe a multimodal model that takes
as input user speech and facial expression to generate multimodal grounding
moves for a listening agent using a large language model. We also describe a
testbed to evaluate approaches to empathic grounding, in which a humanoid robot
interviews a user about a past episode of pain and then has the user rate their
perception of the robot's empathy. We compare our proposed model to one that
only generates non-affective grounding cues in a between-subjects experiment.
Findings demonstrate that empathic grounding increases user perceptions of
empathy, understanding, emotional intelligence, and trust. Our work highlights
the role of emotion awareness and multimodality in generating appropriate
grounding moves for conversational agents. | cs.HC | Empathic Grounding: Explorations using Multimodal Interaction and Large Language Models with Conversational Agents |
2024-07-01T21:34:51Z | http://arxiv.org/pdf/2407.01817v1 | http://arxiv.org/abs/2407.01817v1 | Pranav Narayanan Venkit, Christopher Graziul, Miranda Ardith Goodman, Samantha Nicole Kenny, Shomir Wilson | 2024-07-01T21:34:51Z | cs.CL, cs.CY, cs.HC | Radios are essential for the operations of modern police departments, and
they function as both a collaborative communication technology and a
sociotechnical system. However, little prior research has examined their usage
or their connections to individual privacy and the role of race in policing,
two growing topics of concern in the US. As a case study, we examine the
Chicago Police Department's (CPD's) use of broadcast police communications
(BPC) to coordinate the activity of law enforcement officers (LEOs) in the
city. From a recently assembled archive of 80,775 hours of BPC associated with
CPD operations, we analyze text transcripts of radio transmissions broadcast
9:00 AM to 5:00 PM on August 10th, 2018 in one majority Black, one majority
white, and one majority Hispanic area of the city (24 hours of audio) to
explore three research questions: (1) Do BPC reflect reported racial
disparities in policing? (2) How and when is gender, race/ethnicity, and age
mentioned in BPC? (3) To what extent do BPC include sensitive information, and
who is put at most risk by this practice? (4) To what extent can large language
models (LLMs) heighten this risk? We explore the vocabulary and speech acts
used by police in BPC, comparing mentions of personal characteristics to local
demographics, the personal information shared over BPC, and the privacy
concerns that it poses. Analysis indicates (a) policing professionals in the
city of Chicago exhibit disproportionate attention to Black members of the
public regardless of context, (b) sociodemographic characteristics like gender,
race/ethnicity, and age are primarily mentioned in BPC about event information,
and (c) disproportionate attention introduces disproportionate privacy risks
for Black members of the public. | cs.CL | Race and Privacy in Broadcast Police Communications |
2024-07-01T20:47:47Z | http://arxiv.org/pdf/2407.01796v1 | http://arxiv.org/abs/2407.01796v1 | Sirui Xia, Xintao Wang, Jiaqing Liang, Yifei Zhang, Weikang Zhou, Jiaji Deng, Fei Yu, Yanghua Xiao | 2024-07-01T20:47:47Z | cs.CL | Retrieval-Augmented Generation (RAG) has been widely adopted to enhance Large
Language Models (LLMs) in knowledge-intensive tasks. Recently, Attributed Text
Generation (ATG) has attracted growing attention, which provides citations to
support the model's responses in RAG, so as to enhance the credibility of
LLM-generated content and facilitate verification. Prior methods mainly adopt
coarse-grained attributions, linking to passage-level references or providing
paragraph-level citations. However, these methods still fall short in
verifiability and require certain time costs for fact checking. This paper
proposes a fine-grained ATG method called ReClaim(Refer & Claim), which
alternates the generation of references and answers step by step. Unlike
traditional coarse-grained attribution, ReClaim allows the model to add
sentence-level fine-grained citations to each answer sentence in long-form
question-answering tasks. Our experiments encompass various training and
inference methods and multiple LLMs, verifying the effectiveness of our
approach. | cs.CL | Ground Every Sentence: Improving Retrieval-Augmented LLMs with Interleaved Reference-Claim Generation |
2024-07-01T20:25:20Z | http://arxiv.org/pdf/2407.01784v1 | http://arxiv.org/abs/2407.01784v1 | Kota Shamanth Ramanath Nayak, Leila Kosseim | 2024-07-01T20:25:20Z | cs.CL, cs.AI, cs.LG | This paper describes our approach to hierarchical multi-label detection of
persuasion techniques in meme texts. Our model, developed as a part of the
recent SemEval task, is based on fine-tuning individual language models (BERT,
XLM-RoBERTa, and mBERT) and leveraging a mean-based ensemble model in addition
to dataset augmentation through paraphrase generation from ChatGPT. The scope
of the study encompasses enhancing model performance through innovative
training techniques and data augmentation strategies. The problem addressed is
the effective identification and classification of multiple persuasive
techniques in meme texts, a task complicated by the diversity and complexity of
such content. The objective of the paper is to improve detection accuracy by
refining model training methods and examining the impact of balanced versus
unbalanced training datasets. Novelty in the results and discussion lies in the
finding that training with paraphrases enhances model performance, yet a
balanced training set proves more advantageous than a larger unbalanced one.
Additionally, the analysis reveals the potential pitfalls of indiscriminate
incorporation of paraphrases from diverse distributions, which can introduce
substantial noise. Results with the SemEval 2024 data confirm these insights,
demonstrating improved model efficacy with the proposed methods. | cs.CL | Analyzing Persuasive Strategies in Meme Texts: A Fusion of Language Models with Paraphrase Enrichment |
2024-07-01T18:58:22Z | http://arxiv.org/pdf/2407.01725v1 | http://arxiv.org/abs/2407.01725v1 | Bodhisattwa Prasad Majumder, Harshit Surana, Dhruv Agarwal, Bhavana Dalvi Mishra, Abhijeetsingh Meena, Aryan Prakhar, Tirth Vora, Tushar Khot, Ashish Sabharwal, Peter Clark | 2024-07-01T18:58:22Z | cs.CL, cs.AI, cs.LG | Can the rapid advances in code generation, function calling, and data
analysis using large language models (LLMs) help automate the search and
verification of hypotheses purely from a set of provided datasets? To evaluate
this question, we present DiscoveryBench, the first comprehensive benchmark
that formalizes the multi-step process of data-driven discovery. The benchmark
is designed to systematically assess current model capabilities in discovery
tasks and provide a useful resource for improving them. Our benchmark contains
264 tasks collected across 6 diverse domains, such as sociology and
engineering, by manually deriving discovery workflows from published papers to
approximate the real-world challenges faced by researchers, where each task is
defined by a dataset, its metadata, and a discovery goal in natural language.
We additionally provide 903 synthetic tasks to conduct controlled evaluations
across task complexity. Furthermore, our structured formalism of data-driven
discovery enables a facet-based evaluation that provides useful insights into
different failure modes. We evaluate several popular LLM-based reasoning
frameworks using both open and closed LLMs as baselines on DiscoveryBench and
find that even the best system scores only 25%. Our benchmark, thus,
illustrates the challenges in autonomous data-driven discovery and serves as a
valuable resource for the community to make progress. | cs.CL | DiscoveryBench: Towards Data-Driven Discovery with Large Language Models |
2024-07-01T18:08:17Z | http://arxiv.org/pdf/2407.01697v1 | http://arxiv.org/abs/2407.01697v1 | Salvatore Greco, Ke Zhou, Licia Capra, Tania Cerquitelli, Daniele Quercia | 2024-07-01T18:08:17Z | cs.CL, cs.AI, cs.HC | AI regulations are expected to prohibit machine learning models from using
sensitive attributes during training. However, the latest Natural Language
Processing (NLP) classifiers, which rely on deep learning, operate as black-box
systems, complicating the detection and remediation of such misuse. Traditional
bias mitigation methods in NLP aim for comparable performance across different
groups based on attributes like gender or race but fail to address the
underlying issue of reliance on protected attributes. To partly fix that, we
introduce NLPGuard, a framework for mitigating the reliance on protected
attributes in NLP classifiers. NLPGuard takes an unlabeled dataset, an existing
NLP classifier, and its training data as input, producing a modified training
dataset that significantly reduces dependence on protected attributes without
compromising accuracy. NLPGuard is applied to three classification tasks:
identifying toxic language, sentiment analysis, and occupation classification.
Our evaluation shows that current NLP classifiers heavily depend on protected
attributes, with up to $23\%$ of the most predictive words associated with
these attributes. However, NLPGuard effectively reduces this reliance by up to
$79\%$, while slightly improving accuracy. | cs.CL | NLPGuard: A Framework for Mitigating the Use of Protected Attributes by NLP Classifiers |
2024-07-01T18:01:07Z | http://arxiv.org/pdf/2407.01687v1 | http://arxiv.org/abs/2407.01687v1 | Akshara Prabhakar, Thomas L. Griffiths, R. Thomas McCoy | 2024-07-01T18:01:07Z | cs.CL, cs.AI | Chain-of-Thought (CoT) prompting has been shown to enhance the multi-step
reasoning capabilities of Large Language Models (LLMs). However, debates
persist about whether LLMs exhibit abstract generalization or rely on shallow
heuristics when given CoT prompts. To understand the factors influencing CoT
reasoning we provide a detailed case study of the symbolic reasoning task of
decoding shift ciphers, where letters are shifted forward some number of steps
in the alphabet. GPT-4 achieves zero accuracy on most shift ciphers with
standard prompting, but with CoT its accuracy improves to an average of 32%. By
focusing on a single relatively simple task, we are able to identify three
factors that systematically affect CoT performance: the probability of the
task's expected output (probability), what the model has implicitly learned
during pre-training (memorization), and the number of intermediate operations
involved in reasoning (noisy reasoning). We show that these factors can
drastically influence the task accuracy; e.g., varying the output's probability
of occurrence can shift accuracy from 26% to 70%. We also demonstrate that it
is essential for the model to explicitly produce intermediate steps as output
that can be conditioned on to increase the probability of the correct answer.
Our experiments indicate that as long as the model does so, the validity of the
demonstrations in the prompt does not matter. Overall, we conclude that CoT
prompting performance reflects both memorization and a probabilistic version of
genuine reasoning. | cs.CL | Deciphering the Factors Influencing the Efficacy of Chain-of-Thought: Probability, Memorization, and Noisy Reasoning |
2024-07-01T17:59:47Z | http://arxiv.org/pdf/2407.01527v1 | http://arxiv.org/abs/2407.01527v1 | Jiayi Yuan, Hongyi Liu, Shaochen, Zhong, Yu-Neng Chuang, Songchen Li, Guanchu Wang, Duy Le, Hongye Jin, Vipin Chaudhary, Zhaozhuo Xu, Zirui Liu, Xia Hu | 2024-07-01T17:59:47Z | cs.CL | Long context capability is a crucial competency for large language models
(LLMs) as it mitigates the human struggle to digest long-form texts. This
capability enables complex task-solving scenarios such as book summarization,
code assistance, and many more tasks that are traditionally manpower-intensive.
However, transformer-based LLMs face significant challenges with long context
input due to the growing size of the KV cache and the intrinsic complexity of
attending to extended inputs; where multiple schools of efficiency-driven
approaches -- such as KV cache quantization, token dropping, prompt
compression, linear-time sequence models, and hybrid architectures -- have been
proposed to produce efficient yet long context-capable models. Despite these
advancements, no existing work has comprehensively benchmarked these methods in
a reasonably aligned environment. In this work, we fill this gap by providing a
taxonomy of current methods and evaluating 10+ state-of-the-art approaches
across seven categories of long context tasks. Our work reveals numerous
previously unknown phenomena and offers insights -- as well as a friendly
workbench -- for the future development of long context-capable LLMs. The
source code will be available at https://github.com/henryzhongsc/longctx_bench | cs.CL | KV Cache Compression, But What Must We Give in Return? A Comprehensive Benchmark of Long Context Capable Approaches |
2024-07-01T17:59:35Z | http://arxiv.org/pdf/2407.01525v2 | http://arxiv.org/abs/2407.01525v2 | Chenming Zhu, Tai Wang, Wenwei Zhang, Kai Chen, Xihui Liu | 2024-07-02T07:37:56Z | cs.CV, cs.AI, cs.CL | Although great progress has been made in 3D visual grounding, current models
still rely on explicit textual descriptions for grounding and lack the ability
to reason human intentions from implicit instructions. We propose a new task
called 3D reasoning grounding and introduce a new benchmark ScanReason which
provides over 10K question-answer-location pairs from five reasoning types that
require the synerization of reasoning and grounding. We further design our
approach, ReGround3D, composed of the visual-centric reasoning module empowered
by Multi-modal Large Language Model (MLLM) and the 3D grounding module to
obtain accurate object locations by looking back to the enhanced geometry and
fine-grained details from the 3D scenes. A chain-of-grounding mechanism is
proposed to further boost the performance with interleaved reasoning and
grounding steps during inference. Extensive experiments on the proposed
benchmark validate the effectiveness of our proposed approach. | cs.CV | Empowering 3D Visual Grounding with Reasoning Capabilities |
2024-07-01T17:59:26Z | http://arxiv.org/pdf/2407.01523v1 | http://arxiv.org/abs/2407.01523v1 | Yubo Ma, Yuhang Zang, Liangyu Chen, Meiqi Chen, Yizhu Jiao, Xinze Li, Xinyuan Lu, Ziyu Liu, Yan Ma, Xiaoyi Dong, Pan Zhang, Liangming Pan, Yu-Gang Jiang, Jiaqi Wang, Yixin Cao, Aixin Sun | 2024-07-01T17:59:26Z | cs.CV, cs.CL | Understanding documents with rich layouts and multi-modal components is a
long-standing and practical task. Recent Large Vision-Language Models (LVLMs)
have made remarkable strides in various tasks, particularly in single-page
document understanding (DU). However, their abilities on long-context DU remain
an open problem. This work presents MMLongBench-Doc, a long-context,
multi-modal benchmark comprising 1,062 expert-annotated questions. Distinct
from previous datasets, it is constructed upon 130 lengthy PDF-formatted
documents with an average of 49.4 pages and 20,971 textual tokens. Towards
comprehensive evaluation, answers to these questions rely on pieces of evidence
from (1) different sources (text, image, chart, table, and layout structure)
and (2) various locations (i.e. page number). Moreover, 33.2% of the questions
are cross-page questions requiring evidence across multiple pages. 22.8% of the
questions are designed to be unanswerable for detecting potential
hallucinations. Experiments on 14 LVLMs demonstrate that long-context DU
greatly challenges current models. Notably, the best-performing model, GPT-4o,
achieves an F1 score of only 42.7%, while the second-best, GPT-4V, scores
31.4%. Furthermore, 12 LVLMs (all except GPT-4o and GPT-4V) even present worse
performance than their LLM counterparts which are fed with lossy-parsed OCR
documents. These results validate the necessity of future research toward more
capable long-context LVLMs. Project Page:
https://mayubo2333.github.io/MMLongBench-Doc | cs.CV | MMLongBench-Doc: Benchmarking Long-context Document Understanding with Visualizations |
2024-07-01T17:53:35Z | http://arxiv.org/pdf/2407.01509v1 | http://arxiv.org/abs/2407.01509v1 | Yusu Qian, Hanrong Ye, Jean-Philippe Fauconnier, Peter Grasch, Yinfei Yang, Zhe Gan | 2024-07-01T17:53:35Z | cs.CV, cs.CL | We introduce MIA-Bench, a new benchmark designed to evaluate multimodal large
language models (MLLMs) on their ability to strictly adhere to complex
instructions. Our benchmark comprises a diverse set of 400 image-prompt pairs,
each crafted to challenge the models' compliance with layered instructions in
generating accurate responses that satisfy specific requested patterns.
Evaluation results from a wide array of state-of-the-art MLLMs reveal
significant variations in performance, highlighting areas for improvement in
instruction fidelity. Additionally, we create extra training data and explore
supervised fine-tuning to enhance the models' ability to strictly follow
instructions without compromising performance on other tasks. We hope this
benchmark not only serves as a tool for measuring MLLM adherence to
instructions, but also guides future developments in MLLM training methods. | cs.CV | MIA-Bench: Towards Better Instruction Following Evaluation of Multimodal LLMs |
2024-07-01T17:52:05Z | http://arxiv.org/pdf/2407.01505v1 | http://arxiv.org/abs/2407.01505v1 | Dongping Chen, Jiawen Shi, Yao Wan, Pan Zhou, Neil Zhenqiang Gong, Lichao Sun | 2024-07-01T17:52:05Z | cs.CL, cs.AI | While Large Language Models (LLMs) have achieved remarkable success across
various applications, they also raise concerns regarding self-cognition. In
this paper, we perform a pioneering study to explore self-cognition in LLMs.
Specifically, we first construct a pool of self-cognition instruction prompts
to evaluate where an LLM exhibits self-cognition and four well-designed
principles to quantify LLMs' self-cognition. Our study reveals that 4 of the 48
models on Chatbot Arena--specifically Command R, Claude3-Opus,
Llama-3-70b-Instruct, and Reka-core--demonstrate some level of detectable
self-cognition. We observe a positive correlation between model size, training
data quality, and self-cognition level. Additionally, we also explore the
utility and trustworthiness of LLM in the self-cognition state, revealing that
the self-cognition state enhances some specific tasks such as creative writing
and exaggeration. We believe that our work can serve as an inspiration for
further research to study the self-cognition in LLMs. | cs.CL | Self-Cognition in Large Language Models: An Exploratory Study |
2024-07-01T17:31:03Z | http://arxiv.org/pdf/2407.01492v1 | http://arxiv.org/abs/2407.01492v1 | Qian Liu, Xiaosen Zheng, Niklas Muennighoff, Guangtao Zeng, Longxu Dou, Tianyu Pang, Jing Jiang, Min Lin | 2024-07-01T17:31:03Z | cs.CL, cs.AI | The data mixture for large language model pre-training significantly impacts
performance, yet how to determine an effective mixture remains unclear. We
propose RegMix to automatically identify a high-performing data mixture by
formulating it as a regression task. RegMix involves training a set of small
models with diverse data mixtures and fitting a regression model to predict
their performance given their respective mixtures. With the fitted regression
model, we simulate the top-ranked mixture and use it to train a large-scale
model with orders of magnitude more compute. To empirically validate RegMix, we
train 512 models with 1M parameters for 1B tokens of different mixtures to fit
the regression model and find the optimal mixture. Using this mixture we train
a 1B parameter model for 25B tokens (i.e. 1000x larger and 25x longer) which we
find performs best among 64 candidate 1B parameter models with other mixtures.
Further, our method demonstrates superior performance compared to human
selection and achieves results that match or surpass DoReMi, while utilizing
only 10% of the compute budget. Our experiments also show that (1) Data
mixtures significantly impact performance with single-task performance
variations of up to 14.6%; (2) Web corpora rather than data perceived as
high-quality like Wikipedia have the strongest positive correlation with
downstream performance; (3) Domains interact in complex ways often
contradicting common sense, thus automatic approaches like RegMix are needed;
(4) Data mixture effects transcend scaling laws, and our approach captures the
complexity by considering all domains together. Our code is available at
https://github.com/sail-sg/regmix. | cs.CL | RegMix: Data Mixture as Regression for Language Model Pre-training |
2024-07-01T17:28:59Z | http://arxiv.org/pdf/2407.01491v1 | http://arxiv.org/abs/2407.01491v1 | Siwei Li, Yifan Yang, Yifei Shen, Fangyun Wei, Zongqing Lu, Lili Qiu, Yuqing Yang | 2024-07-01T17:28:59Z | cs.CL, cs.CV | Efficient fine-tuning plays a fundamental role in modern large models, with
low-rank adaptation emerging as a particularly promising approach. However, the
existing variants of LoRA are hampered by limited expressiveness, a tendency to
overfit, and sensitivity to hyperparameter settings. This paper presents LoRA
Slow Cascade Learning (LoRASC), an innovative technique designed to enhance
LoRA's expressiveness and generalization capabilities while preserving its
training efficiency. Our approach augments expressiveness through a cascaded
learning strategy that enables a mixture-of-low-rank adaptation, thereby
increasing the model's ability to capture complex patterns. Additionally, we
introduce a slow-fast update mechanism and cascading noisy tuning to bolster
generalization. The extensive experiments on various language and vision
datasets, as well as robustness benchmarks, demonstrate that the proposed
method not only significantly outperforms existing baselines, but also
mitigates overfitting, enhances model stability, and improves OOD robustness.
Code will be release in https://github.com/microsoft/LoRASC very soon. | cs.CL | Expressive and Generalizable Low-rank Adaptation for Large Models via Slow Cascaded Learning |
2024-07-01T17:26:21Z | http://arxiv.org/pdf/2407.01490v1 | http://arxiv.org/abs/2407.01490v1 | Luísa Shimabucoro, Sebastian Ruder, Julia Kreutzer, Marzieh Fadaee, Sara Hooker | 2024-07-01T17:26:21Z | cs.CL, cs.AI, cs.LG | The widespread adoption of synthetic data raises new questions about how
models generating the data can influence other large language models (LLMs) via
distilled data. To start, our work exhaustively characterizes the impact of
passive inheritance of model properties by systematically studying the
consequences of synthetic data integration. We provide one of the most
comprehensive studies to-date of how the source of synthetic data shapes
models' internal biases, calibration and generations' textual attributes and
preferences. We find that models are surprisingly sensitive towards certain
attributes even when the synthetic data prompts appear "neutral". which invites
the question whether this sensitivity can be exploited for good.
Our findings invite the question can we explicitly steer the models towards
the properties we want at test time by exploiting the data generation process?
This would have historically been considered infeasible due to the cost of
collecting data with a specific characteristic or objective in mind. However,
improvement in the quality of synthetic data, as well as a shift towards
general-purpose models designed to follow a diverse way of instructions, means
this question is timely. We propose active inheritance as a term to describe
intentionally constraining synthetic data according to a non-differentiable
objective. We demonstrate how active inheritance can steer the generation
profiles of models towards desirable non-differentiable attributes, e.g. high
lexical diversity or low toxicity. | cs.CL | LLM See, LLM Do: Guiding Data Generation to Target Non-Differentiable Objectives |
2024-07-01T17:24:45Z | http://arxiv.org/pdf/2407.01489v1 | http://arxiv.org/abs/2407.01489v1 | Chunqiu Steven Xia, Yinlin Deng, Soren Dunn, Lingming Zhang | 2024-07-01T17:24:45Z | cs.SE, cs.AI, cs.CL, cs.LG | Recent advancements in large language models (LLMs) have significantly
advanced the automation of software development tasks, including code
synthesis, program repair, and test generation. More recently, researchers and
industry practitioners have developed various autonomous LLM agents to perform
end-to-end software development tasks. These agents are equipped with the
ability to use tools, run commands, observe feedback from the environment, and
plan for future actions. However, the complexity of these agent-based
approaches, together with the limited abilities of current LLMs, raises the
following question: Do we really have to employ complex autonomous software
agents? To attempt to answer this question, we build Agentless -- an agentless
approach to automatically solve software development problems. Compared to the
verbose and complex setup of agent-based approaches, Agentless employs a
simplistic two-phase process of localization followed by repair, without
letting the LLM decide future actions or operate with complex tools. Our
results on the popular SWE-bench Lite benchmark show that surprisingly the
simplistic Agentless is able to achieve both the highest performance (27.33%)
and lowest cost (\$0.34) compared with all existing open-source software
agents! Furthermore, we manually classified the problems in SWE-bench Lite and
found problems with exact ground truth patch or insufficient/misleading issue
descriptions. As such, we construct SWE-bench Lite-S by excluding such
problematic issues to perform more rigorous evaluation and comparison. Our work
highlights the current overlooked potential of a simple, interpretable
technique in autonomous software development. We hope Agentless will help reset
the baseline, starting point, and horizon for autonomous software agents, and
inspire future work along this crucial direction. | cs.SE | Agentless: Demystifying LLM-based Software Engineering Agents |
2024-07-01T17:07:55Z | http://arxiv.org/pdf/2407.01476v1 | http://arxiv.org/abs/2407.01476v1 | Jing Yu Koh, Stephen McAleer, Daniel Fried, Ruslan Salakhutdinov | 2024-07-01T17:07:55Z | cs.AI, cs.CL, cs.LG | Autonomous agents powered by language models (LMs) have demonstrated promise
in their ability to perform decision-making tasks such as web automation.
However, a key limitation remains: LMs, primarily optimized for natural
language understanding and generation, struggle with multi-step reasoning,
planning, and using environmental feedback when attempting to solve realistic
computer tasks. Towards addressing this, we propose an inference-time search
algorithm for LM agents to explicitly perform exploration and multi-step
planning in interactive web environments. Our approach is a form of best-first
tree search that operates within the actual environment space, and is
complementary with most existing state-of-the-art agents. It is the first tree
search algorithm for LM agents that shows effectiveness on realistic web tasks.
On the challenging VisualWebArena benchmark, applying our search algorithm on
top of a GPT-4o agent yields a 39.7% relative increase in success rate compared
to the same baseline without search, setting a state-of-the-art success rate of
26.4%. On WebArena, search also yields a 28.0% relative improvement over a
baseline agent, setting a competitive success rate of 19.2%. Our experiments
highlight the effectiveness of search for web agents, and we demonstrate that
performance scales with increased test-time compute. We conduct a thorough
analysis of our results to highlight improvements from search, limitations, and
promising directions for future work. Our code and models are publicly released
at https://jykoh.com/search-agents. | cs.AI | Tree Search for Language Model Agents |
2024-07-01T17:01:54Z | http://arxiv.org/pdf/2407.01470v1 | http://arxiv.org/abs/2407.01470v1 | Tzu-Han Lin, Chen-An Li, Hung-yi Lee, Yun-Nung Chen | 2024-07-01T17:01:54Z | cs.CL | Reinforcement learning from human feedback (RLHF) is a popular strategy for
aligning large language models (LLMs) with desired behaviors. Reward modeling
is a crucial step in RLHF. However, collecting paired preference data for
training reward models is often costly and time-consuming, especially for
domain-specific preferences requiring expert annotation. To address this
challenge, we propose the \textbf{Do}main knowled\textbf{ge} merged
\textbf{R}eward \textbf{M}odel (DogeRM), a novel framework that integrates
domain-specific knowledge into a general reward model by model merging. The
experiments demonstrate that DogeRM enhances performance across different
benchmarks and provide a detailed analysis showcasing the effects of model
merging, showing the great potential of facilitating model alignment. | cs.CL | DogeRM: Equipping Reward Models with Domain Knowledge through Model Merging |
2024-07-01T16:56:50Z | http://arxiv.org/pdf/2407.01463v1 | http://arxiv.org/abs/2407.01463v1 | Nadezhda Chirkova, David Rau, Hervé Déjean, Thibault Formal, Stéphane Clinchant, Vassilina Nikoulina | 2024-07-01T16:56:50Z | cs.CL, cs.AI | Retrieval-augmented generation (RAG) has recently emerged as a promising
solution for incorporating up-to-date or domain-specific knowledge into large
language models (LLMs) and improving LLM factuality, but is predominantly
studied in English-only settings. In this work, we consider RAG in the
multilingual setting (mRAG), i.e. with user queries and the datastore in 13
languages, and investigate which components and with which adjustments are
needed to build a well-performing mRAG pipeline, that can be used as a strong
baseline in future works. Our findings highlight that despite the availability
of high-quality off-the-shelf multilingual retrievers and generators,
task-specific prompt engineering is needed to enable generation in user
languages. Moreover, current evaluation metrics need adjustments for
multilingual setting, to account for variations in spelling named entities. The
main limitations to be addressed in future works include frequent
code-switching in non-Latin alphabet languages, occasional fluency errors,
wrong reading of the provided documents, or irrelevant retrieval. We release
the code for the resulting mRAG baseline pipeline at
https://github.com/naver/bergen. | cs.CL | Retrieval-augmented generation in multilingual settings |
2024-07-01T16:55:28Z | http://arxiv.org/pdf/2407.01461v1 | http://arxiv.org/abs/2407.01461v1 | Zisu Huang, Xiaohua Wang, Feiran Zhang, Zhibo Xu, Cenyuan Zhang, Xiaoqing Zheng, Xuanjing Huang | 2024-07-01T16:55:28Z | cs.CL | The capacity of large language models (LLMs) to generate honest, harmless,
and helpful responses heavily relies on the quality of user prompts. However,
these prompts often tend to be brief and vague, thereby significantly limiting
the full potential of LLMs. Moreover, harmful prompts can be meticulously
crafted and manipulated by adversaries to jailbreak LLMs, inducing them to
produce potentially toxic content. To enhance the capabilities of LLMs while
maintaining strong robustness against harmful jailbreak inputs, this study
proposes a transferable and pluggable framework that refines user prompts
before they are input into LLMs. This strategy improves the quality of the
queries, empowering LLMs to generate more truthful, benign and useful
responses. Specifically, a lightweight query refinement model is introduced and
trained using a specially designed reinforcement learning approach that
incorporates multiple objectives to enhance particular capabilities of LLMs.
Extensive experiments demonstrate that the refinement model not only improves
the quality of responses but also strengthens their robustness against
jailbreak attacks. Code is available at:
https://github.com/Huangzisu/query-refinement . | cs.CL | Enhancing the Capability and Robustness of Large Language Models through Reinforcement Learning-Driven Query Refinement |
2024-07-01T16:50:49Z | http://arxiv.org/pdf/2407.01455v1 | http://arxiv.org/abs/2407.01455v1 | Guiyang Hou, Wenqi Zhang, Yongliang Shen, Linjuan Wu, Weiming Lu | 2024-07-01T16:50:49Z | cs.CL | Theory of Mind (ToM)-the cognitive ability to reason about mental states of
ourselves and others, is the foundation of social interaction. Although ToM
comes naturally to humans, it poses a significant challenge to even the most
advanced Large Language Models (LLMs). Due to the complex logical chains in ToM
reasoning, especially in higher-order ToM questions, simply utilizing reasoning
methods like Chain of Thought (CoT) will not improve the ToM capabilities of
LLMs. We present TimeToM, which constructs a temporal space and uses it as the
foundation to improve the ToM capabilities of LLMs in multiple scenarios.
Specifically, within the temporal space, we construct Temporal Belief State
Chain (TBSC) for each character and inspired by the cognition perspective of
the social world model, we divide TBSC into self-world beliefs and social world
beliefs, aligning with first-order ToM (first-order beliefs) and higher-order
ToM (higher-order beliefs) questions, respectively. Moreover, we design a novel
tool-belief solver that, by considering belief communication between characters
in temporal space, can transform a character's higher-order beliefs into
another character's first-order beliefs under belief communication period.
Experimental results indicate that TimeToM can dramatically improve the
reasoning performance of LLMs on ToM questions while taking a big step towards
coherent and robust ToM reasoning. | cs.CL | TimeToM: Temporal Space is the Key to Unlocking the Door of Large Language Models' Theory-of-Mind |
2024-07-01T16:32:16Z | http://arxiv.org/pdf/2407.01437v1 | http://arxiv.org/abs/2407.01437v1 | Subhajit Chaudhury, Soham Dan, Payel Das, Georgios Kollias, Elliot Nelson | 2024-07-01T16:32:16Z | cs.CL, cs.AI, cs.LG | In this paper, we demonstrate the benefits of using memory augmented Large
Language Model (LLM) architecture in improving the recall abilities of facts
from a potentially long context. As a case study we test LARIMAR, a recently
proposed LLM architecture which augments a LLM decoder with an external
associative memory, on several long-context recall tasks, including passkey and
needle-in-the-haystack tests. We demonstrate that the external memory can be
adapted at test time to handle contexts much longer than those seen during
training, while keeping readouts from the memory recognizable to the trained
decoder and without increasing GPU memory footprint. Compared to alternative
architectures for long-context recall tasks with models of a comparable
parameter count, LARIMAR is able to maintain strong performance without any
task-specific training. | cs.CL | Needle in the Haystack for Memory Based Large Language Models |
2024-07-01T16:14:25Z | http://arxiv.org/pdf/2407.01424v1 | http://arxiv.org/abs/2407.01424v1 | Yiping Sun | 2024-07-01T16:14:25Z | cs.CL, cs.IR | Relation classification, a crucial component of relation extraction, involves
identifying connections between two entities. Previous studies have
predominantly focused on integrating the attention mechanism into relation
classification at a global scale, overlooking the importance of the local
context. To address this gap, this paper introduces a novel global-local
attention mechanism for relation classification, which enhances global
attention with a localized focus. Additionally, we propose innovative hard and
soft localization mechanisms to identify potential keywords for local
attention. By incorporating both hard and soft localization strategies, our
approach offers a more nuanced and comprehensive understanding of the
contextual cues that contribute to effective relation classification. Our
experimental results on the SemEval-2010 Task 8 dataset highlight the superior
performance of our method compared to previous attention-based approaches in
relation classification. | cs.CL | A Global-Local Attention Mechanism for Relation Classification |
2024-07-01T16:00:53Z | http://arxiv.org/pdf/2407.01411v2 | http://arxiv.org/abs/2407.01411v2 | Jesus-German Ortiz-Barajas, Helena Gomez-Adorno, Thamar Solorio | 2024-07-02T06:21:41Z | cs.CL | We present HyperLoader, a simple approach that combines different
parameter-efficient fine-tuning methods in a multi-task setting. To achieve
this goal, our model uses a hypernetwork to generate the weights of these
modules based on the task, the transformer layer, and its position within this
layer. Our method combines the benefits of multi-task learning by capturing the
structure of all tasks while reducing the task interference problem by
encapsulating the task-specific knowledge in the generated weights and the
benefits of combining different parameter-efficient methods to outperform
full-fine tuning. We provide empirical evidence that HyperLoader outperforms
previous approaches in most datasets and obtains the best average performance
across tasks in high-resource and low-resource scenarios. | cs.CL | HyperLoader: Integrating Hypernetwork-Based LoRA and Adapter Layers into Multi-Task Transformers for Sequence Labelling |
2024-07-01T15:59:17Z | http://arxiv.org/pdf/2407.01409v1 | http://arxiv.org/abs/2407.01409v1 | Jacopo D'Abramo, Andrea Zugarini, Paolo Torroni | 2024-07-01T15:59:17Z | cs.CL, cs.AI | Large language models present opportunities for innovative Question Answering
over Knowledge Graphs (KGQA). However, they are not inherently designed for
query generation. To bridge this gap, solutions have been proposed that rely on
fine-tuning or ad-hoc architectures, achieving good results but limited
out-of-domain distribution generalization. In this study, we introduce a novel
approach called Dynamic Few-Shot Learning (DFSL). DFSL integrates the
efficiency of in-context learning and semantic similarity and provides a
generally applicable solution for KGQA with state-of-the-art performance. We
run an extensive evaluation across multiple benchmark datasets and architecture
configurations. | cs.CL | Dynamic Few-Shot Learning for Knowledge Graph Question Answering |
2024-07-01T15:56:24Z | http://arxiv.org/pdf/2407.01406v1 | http://arxiv.org/abs/2407.01406v1 | Daniil Gurgurov, Mareike Hartmann, Simon Ostermann | 2024-07-01T15:56:24Z | cs.CL, cs.AI | This paper explores the integration of graph knowledge from linguistic
ontologies into multilingual Large Language Models (LLMs) using adapters to
improve performance for low-resource languages (LRLs) in sentiment analysis
(SA) and named entity recognition (NER). Building upon successful
parameter-efficient fine-tuning techniques, such as K-ADAPTER and MAD-X, we
propose a similar approach for incorporating knowledge from multilingual
graphs, connecting concepts in various languages with each other through
linguistic relationships, into multilingual LLMs for LRLs. Specifically, we
focus on eight LRLs -- Maltese, Bulgarian, Indonesian, Nepali, Javanese,
Uyghur, Tibetan, and Sinhala -- and employ language-specific adapters
fine-tuned on data extracted from the language-specific section of ConceptNet,
aiming to enable knowledge transfer across the languages covered by the
knowledge graph. We compare various fine-tuning objectives, including standard
Masked Language Modeling (MLM), MLM with full-word masking, and MLM with
targeted masking, to analyse their effectiveness in learning and integrating
the extracted graph data. Through empirical evaluation on language-specific
tasks, we assess how structured graph knowledge affects the performance of
multilingual LLMs for LRLs in SA and NER, providing insights into the potential
benefits of adapting language models for low-resource scenarios. | cs.CL | Adapting Multilingual LLMs to Low-Resource Languages with Knowledge Graphs via Adapters |
2024-07-01T15:53:29Z | http://arxiv.org/pdf/2407.01403v1 | http://arxiv.org/abs/2407.01403v1 | Vitaly Bulgakov | 2024-07-01T15:53:29Z | cs.IR, cs.AI, cs.CL, cs.LG | In this paper, we focus on methods to reduce the size and improve the quality
of the prompt context required for question-answering systems. Attempts to
increase the number of retrieved chunked documents and thereby enlarge the
context related to the query can significantly complicate the processing and
decrease the performance of a Large Language Model (LLM) when generating
responses to queries. It is well known that a large set of documents retrieved
from a database in response to a query may contain irrelevant information,
which often leads to hallucinations in the resulting answers. Our goal is to
select the most semantically relevant documents, treating the discarded ones as
outliers. We propose and evaluate several methods for identifying outliers by
creating features that utilize the distances of embedding vectors, retrieved
from the vector database, to both the centroid and the query vectors. The
methods were evaluated by comparing the similarities of the retrieved LLM
responses to ground-truth answers obtained using the OpenAI GPT-4o model. It
was found that the greatest improvements were achieved with increasing
complexity of the questions and answers. | cs.IR | Optimization of Retrieval-Augmented Generation Context with Outlier Detection |
2024-07-01T15:46:45Z | http://arxiv.org/pdf/2407.01394v1 | http://arxiv.org/abs/2407.01394v1 | Pooya Fayyazsanavi, Antonios Anastasopoulos, Jana Košecká | 2024-07-01T15:46:45Z | cs.CV, cs.CL, cs.LG | Sign language translation from video to spoken text presents unique
challenges owing to the distinct grammar, expression nuances, and high
variation of visual appearance across different speakers and contexts. The
intermediate gloss annotations of videos aim to guide the translation process.
In our work, we focus on {\em Gloss2Text} translation stage and propose several
advances by leveraging pre-trained large language models (LLMs), data
augmentation, and novel label-smoothing loss function exploiting gloss
translation ambiguities improving significantly the performance of
state-of-the-art approaches. Through extensive experiments and ablation studies
on the PHOENIX Weather 2014T dataset, our approach surpasses state-of-the-art
performance in {\em Gloss2Text} translation, indicating its efficacy in
addressing sign language translation and suggesting promising avenues for
future research and development. | cs.CV | Gloss2Text: Sign Language Gloss translation using LLMs and Semantically Aware Label Smoothing |
2024-07-01T15:45:21Z | http://arxiv.org/pdf/2407.01393v1 | http://arxiv.org/abs/2407.01393v1 | Daniel Dzienisiewicz, Filip Graliński, Piotr Jabłoński, Marek Kubis, Paweł Skórzewski, Piotr Wierzchoń | 2024-07-01T15:45:21Z | cs.CL | This paper presents the POLygraph dataset, a unique resource for fake news
detection in Polish. The dataset, created by an interdisciplinary team, is
composed of two parts: the "fake-or-not" dataset with 11,360 pairs of news
articles (identified by their URLs) and corresponding labels, and the
"fake-they-say" dataset with 5,082 news articles (identified by their URLs) and
tweets commenting on them. Unlike existing datasets, POLygraph encompasses a
variety of approaches from source literature, providing a comprehensive
resource for fake news detection. The data was collected through manual
annotation by expert and non-expert annotators. The project also developed a
software tool that uses advanced machine learning techniques to analyze the
data and determine content authenticity. The tool and dataset are expected to
benefit various entities, from public sector institutions to publishers and
fact-checking organizations. Further dataset exploration will foster fake news
detection and potentially stimulate the implementation of similar models in
other languages. The paper focuses on the creation and composition of the
dataset, so it does not include a detailed evaluation of the software tool for
content authenticity analysis, which is planned at a later stage of the
project. | cs.CL | POLygraph: Polish Fake News Dataset |
2024-07-01T15:34:17Z | http://arxiv.org/pdf/2407.01384v1 | http://arxiv.org/abs/2407.01384v1 | Yi-Sheng Hsu, Nils Feldhus, Sherzod Hakimov | 2024-07-01T15:34:17Z | cs.CL | Free-text rationales justify model decisions in natural language and thus
become likable and accessible among approaches to explanation across many
tasks. However, their effectiveness can be hindered by misinterpretation and
hallucination. As a perturbation test, we investigate how large language models
(LLMs) perform the task of natural language explanation (NLE) under the effects
of readability level control, i.e., being prompted for a rationale targeting a
specific expertise level, such as sixth grade or college. We find that
explanations are adaptable to such instruction, but the requested readability
is often misaligned with the measured text complexity according to traditional
readability metrics. Furthermore, the quality assessment shows that LLMs'
ratings of rationales across text complexity exhibit a similar pattern of
preference as observed in natural language generation (NLG). Finally, our human
evaluation suggests a generally satisfactory impression on rationales at all
readability levels, with high-school-level readability being most commonly
perceived and favored. | cs.CL | Free-text Rationale Generation under Readability Level Control |
2024-07-01T15:29:45Z | http://arxiv.org/pdf/2407.01376v1 | http://arxiv.org/abs/2407.01376v1 | Dmitrii Volkov | 2024-07-01T15:29:45Z | cs.LG, cs.AI, cs.CL, cs.CR | We show that extensive LLM safety fine-tuning is easily subverted when an
attacker has access to model weights. We evaluate three state-of-the-art
fine-tuning methods-QLoRA, ReFT, and Ortho-and show how algorithmic advances
enable constant jailbreaking performance with cuts in FLOPs and optimisation
power. We strip safety fine-tuning from Llama 3 8B in one minute and Llama 3
70B in 30 minutes on a single GPU, and sketch ways to reduce this further. | cs.LG | Badllama 3: removing safety finetuning from Llama 3 in minutes |
2024-07-01T15:26:03Z | http://arxiv.org/pdf/2407.01374v1 | http://arxiv.org/abs/2407.01374v1 | Mohan Raj Chanthran, Lay-Ki Soon, Huey Fang Ong, Bhawani Selvaretnam | 2024-07-01T15:26:03Z | cs.CL | Malaysian English is a low resource creole language, where it carries the
elements of Malay, Chinese, and Tamil languages, in addition to Standard
English. Named Entity Recognition (NER) models underperform when capturing
entities from Malaysian English text due to its distinctive morphosyntactic
adaptations, semantic features and code-switching (mixing English and Malay).
Considering these gaps, we introduce MENmBERT and MENBERT, a pre-trained
language model with contextual understanding, specifically tailored for
Malaysian English. We have fine-tuned MENmBERT and MENBERT using manually
annotated entities and relations from the Malaysian English News Article (MEN)
Dataset. This fine-tuning process allows the PLM to learn representations that
capture the nuances of Malaysian English relevant for NER and RE tasks.
MENmBERT achieved a 1.52\% and 26.27\% improvement on NER and RE tasks
respectively compared to the bert-base-multilingual-cased model. Although the
overall performance of NER does not have a significant improvement, our further
analysis shows that there is a significant improvement when evaluated by the 12
entity labels. These findings suggest that pre-training language models on
language-specific and geographically-focused corpora can be a promising
approach for improving NER performance in low-resource settings. The dataset
and code published in this paper provide valuable resources for NLP research
work focusing on Malaysian English. | cs.CL | Bridging the Gap: Transfer Learning from English PLMs to Malaysian English |
2024-07-01T15:23:42Z | http://arxiv.org/pdf/2407.01370v1 | http://arxiv.org/abs/2407.01370v1 | Philippe Laban, Alexander R. Fabbri, Caiming Xiong, Chien-Sheng Wu | 2024-07-01T15:23:42Z | cs.CL | LLMs and RAG systems are now capable of handling millions of input tokens or
more. However, evaluating the output quality of such systems on long-context
tasks remains challenging, as tasks like Needle-in-a-Haystack lack complexity.
In this work, we argue that summarization can play a central role in such
evaluation. We design a procedure to synthesize Haystacks of documents,
ensuring that specific \textit{insights} repeat across documents. The "Summary
of a Haystack" (SummHay) task then requires a system to process the Haystack
and generate, given a query, a summary that identifies the relevant insights
and precisely cites the source documents. Since we have precise knowledge of
what insights should appear in a haystack summary and what documents should be
cited, we implement a highly reproducible automatic evaluation that can score
summaries on two aspects - Coverage and Citation. We generate Haystacks in two
domains (conversation, news), and perform a large-scale evaluation of 10 LLMs
and corresponding 50 RAG systems. Our findings indicate that SummHay is an open
challenge for current systems, as even systems provided with an Oracle signal
of document relevance lag our estimate of human performance (56\%) by 10+
points on a Joint Score. Without a retriever, long-context LLMs like GPT-4o and
Claude 3 Opus score below 20% on SummHay. We show SummHay can also be used to
study enterprise RAG systems and position bias in long-context models. We hope
future systems can equal and surpass human performance on SummHay. | cs.CL | Summary of a Haystack: A Challenge to Long-Context LLMs and RAG Systems |
2024-07-01T15:15:24Z | http://arxiv.org/pdf/2407.01360v1 | http://arxiv.org/abs/2407.01360v1 | Abrar Abir, Kemal Oflazer | 2024-07-01T15:15:24Z | cs.CL | This paper investigates the optimization of propaganda technique detection in
Arabic text, including tweets \& news paragraphs, from ArAIEval shared task 1.
Our approach involves fine-tuning the AraBERT v2 model with a neural network
classifier for sequence tagging. Experimental results show relying on the first
token of the word for technique prediction produces the best performance. In
addition, incorporating genre information as a feature further enhances the
model's performance. Our system achieved a score of 25.41, placing us 4$^{th}$
on the leaderboard. Subsequent post-submission improvements further raised our
score to 26.68. | cs.CL | Nullpointer at ArAIEval Shared Task: Arabic Propagandist Technique Detection with Token-to-Word Mapping in Sequence Tagging |
2024-07-01T15:11:37Z | http://arxiv.org/pdf/2407.01358v1 | http://arxiv.org/abs/2407.01358v1 | Xiaolin Xing, Zhiwei He, Haoyu Xu, Xing Wang, Rui Wang, Yu Hong | 2024-07-01T15:11:37Z | cs.CL | This paper investigates the cross-lingual inconsistencies observed in Large
Language Models (LLMs), such as ChatGPT, Llama, and Baichuan, which have shown
exceptional performance in various Natural Language Processing (NLP) tasks.
Despite their successes, these models often exhibit significant inconsistencies
when processing the same concepts across different languages. This study
focuses on three primary questions: the existence of cross-lingual
inconsistencies in LLMs, the specific aspects in which these inconsistencies
manifest, and the correlation between cross-lingual consistency and
multilingual capabilities of LLMs.To address these questions, we propose an
innovative evaluation method for Cross-lingual Semantic Consistency (xSC) using
the LaBSE model. We further introduce metrics for Cross-lingual Accuracy
Consistency (xAC) and Cross-lingual Timeliness Consistency (xTC) to
comprehensively assess the models' performance regarding semantic, accuracy,
and timeliness inconsistencies. By harmonizing these metrics, we provide a
holistic measurement of LLMs' cross-lingual consistency. Our findings aim to
enhance the understanding and improvement of multilingual capabilities and
interpretability in LLMs, contributing to the development of more robust and
reliable multilingual language models. | cs.CL | Evaluating Knowledge-based Cross-lingual Inconsistency in Large Language Models |
2024-07-01T14:41:59Z | http://arxiv.org/pdf/2407.01334v2 | http://arxiv.org/abs/2407.01334v2 | Re'em Harel, Yair Elboher, Yuval Pinter | 2024-07-03T16:31:52Z | cs.CL, cs.CR | Using language models as a remote service entails sending private information
to an untrusted provider. In addition, potential eavesdroppers can intercept
the messages, thereby exposing the information. In this work, we explore the
prospects of avoiding such data exposure at the level of text manipulation. We
focus on text classification models, examining various token mapping and
contextualized manipulation functions in order to see whether classifier
accuracy may be maintained while keeping the original text unrecoverable. We
find that although some token mapping functions are easy and straightforward to
implement, they heavily influence performance on the downstream task, and via a
sophisticated attacker can be reconstructed. In comparison, the contextualized
manipulation provides an improvement in performance. | cs.CL | Protecting Privacy in Classifiers by Token Manipulation |
2024-07-01T14:26:48Z | http://arxiv.org/pdf/2407.01320v1 | http://arxiv.org/abs/2407.01320v1 | Haobo Song, Hao Zhao, Soumajit Majumder, Tao Lin | 2024-07-01T14:26:48Z | cs.LG, cs.AI, cs.CL | Fine-tuning large pre-trained foundation models, such as the 175B GPT-3, has
attracted more attention for downstream tasks recently. While
parameter-efficient fine-tuning methods have been proposed and proven effective
without retraining all model parameters, their performance is limited by the
capacity of incremental modules, especially under constrained parameter
budgets. \\ To overcome this challenge, we propose CapaBoost, a simple yet
effective strategy that enhances model capacity by leveraging low-rank updates
through parallel weight modules in target layers. By applying static random
masks to the shared weight matrix, CapaBoost constructs a diverse set of weight
matrices, effectively increasing the rank of incremental weights without adding
parameters. Notably, our approach can be seamlessly integrated into various
existing parameter-efficient fine-tuning methods. We extensively validate the
efficacy of CapaBoost through experiments on diverse downstream tasks,
including natural language understanding, question answering, and image
classification. Our results demonstrate significant improvements over
baselines, without incurring additional computation or storage costs. Our code
is available at \url{https://github.com/LINs-lab/CapaBoost}. | cs.LG | Increasing Model Capacity for Free: A Simple Strategy for Parameter Efficient Fine-tuning |
2024-07-01T14:20:54Z | http://arxiv.org/pdf/2407.01315v1 | http://arxiv.org/abs/2407.01315v1 | Ahmed Njifenjou, Virgile Sucal, Bassam Jabaian, Fabrice Lefèvre | 2024-07-01T14:20:54Z | cs.CL | In this paper we propose a study of linguistic portability strategies of
large pre-trained language models (PLMs) used for open-domain dialogue systems
in a high-resource language for this task. In particular the target
low-resource language (L_T) will be simulated with French, as it lacks of
task-specific resources and allows our human evaluation, when the source
language (L_S) is English. For obvious reasons, recent works using such models
for open-domain dialogue are mostly developed in English. Yet building specific
PLMs for each possible target language supposes collecting new datasets and is
costly. For this reason, trying to leverage all existing resources (PLMs and
data) in both L_S and L_T , we wish to assess the performance achievable in L_T
with different approaches. The first two approaches evaluate the usage of
Neural Machine Translation (NMT) at different levels: TrainOnTarget where a L_S
dataset is translated before fine-tuning in L_T and TestOnSource where a L_S
model is coupled with NMT modules during inference. Then, the advent of BLOOM
[2], the world first open-access multilingual large PLM, allow researchers to
develop new approaches aiming to leverage not only the model's full
accessibility but also its multilingualism and translation abilities. In this
context the task is learned in L_S first and adapted to L_T using the MAD-X
Adapter architecture [16]. In the two sets of experiments models are evaluated
in spoken dialogue conditions with human and the strategies can be compared in
terms of perceived interaction quality. | cs.CL | Language Portability Strategies for Open-domain Dialogue with Pre-trained Language Models from High to Low Resource Languages |