arXiv subjects

Pengfei Liu

Publications and source records attributed to Pengfei Liu.

At least 145 records · Page 8Linked to original sources

arXiv2024

Heavy flavor-asymmetric pseudoscalar mesons on the light front

We extract the leading Fock-state light front wave functions (LF-LFWFs) of heavy flavor-asymmetric pseudoscalar mesons $D$, $B$ and $B_c$ from their Bethe-Salpeter wave functions based on Dyson-Schwinger equations approach, and study their leading twist parton distribution amplitudes, generalized parton distribution functions and transverse momentum dependent parton distributions. The spatial distributions of the quark and antiquark on the transverse plane are given, along with their charge and energy distributions on the light front. We find that in the considered mesons, the heavier quarks carry most longitudinal momentum fraction and yield narrow $x$-distributions, while the lighter quarks play an active role in shaping the transverse distributions within both spatial and momentum space, exhibiting a duality embodying characteristics from both light mesons and heavy quarkonium.

hep-ph↗

arXiv2024

Understanding Reference Policies in Direct Preference Optimization

Direct Preference Optimization (DPO) has become a widely used training method for the instruction fine-tuning of large language models (LLMs). In this work, we explore an under-investigated aspect of DPO - its dependency on the reference model or policy. Such reference policies, typically instantiated as the model to be further fine-tuned, are important since they can impose an upper limit on DPO's effectiveness. Therefore, we address three related research questions in this work. First, we explore the optimal strength of the KL divergence constraint in DPO, which penalizes deviations from the reference policy, and find that DPO is sensitive to this strength. Next, we examine the necessity of the KL-constraint from the reference policies in DPO by providing both theoretical and empirical comparisons between DPO and related learning objectives, demonstrating DPO's superiority in this controlled setting. Additionally, we investigate whether DPO benefits from stronger reference policies, finding that a stronger reference policy can lead to improved performance, but only when it is similar to the model being fine-tuned. Our findings highlight the confounding role of reference policies in DPO and offer insights for best practices, while also identifying open research questions for future studies.

cs.CL↗

arXiv2024

RAGChecker: A Fine-grained Framework for Diagnosing Retrieval-Augmented Generation

Despite Retrieval-Augmented Generation (RAG) showing promising capability in leveraging external knowledge, a comprehensive evaluation of RAG systems is still challenging due to the modular nature of RAG, evaluation of long-form responses and reliability of measurements. In this paper, we propose a fine-grained evaluation framework, RAGChecker, that incorporates a suite of diagnostic metrics for both the retrieval and generation modules. Meta evaluation verifies that RAGChecker has significantly better correlations with human judgments than other evaluation metrics. Using RAGChecker, we evaluate 8 RAG systems and conduct an in-depth analysis of their performance, revealing insightful patterns and trade-offs in the design choices of RAG architectures. The metrics of RAGChecker can guide researchers and practitioners in developing more effective RAG systems. This work has been open sourced at https://github.com/amazon-science/RAGChecker.

cs.CL↗

arXiv2024

SAFETY-J: Evaluating Safety with Critique

The deployment of Large Language Models (LLMs) in content generation raises significant safety concerns, particularly regarding the transparency and interpretability of content evaluations. Current methods, primarily focused on binary safety classifications, lack mechanisms for detailed critique, limiting their utility for model improvement and user trust. To address these limitations, we introduce SAFETY-J, a bilingual generative safety evaluator for English and Chinese with critique-based judgment. SAFETY-J utilizes a robust training dataset that includes diverse dialogues and augmented query-response pairs to assess safety across various scenarios comprehensively. We establish an automated meta-evaluation benchmark that objectively assesses the quality of critiques with minimal human intervention, facilitating scalable and continuous improvement. Additionally, SAFETY-J employs an iterative preference learning technique to dynamically refine safety assessments based on meta-evaluations and critiques. Our evaluations demonstrate that SAFETY-J provides more nuanced and accurate safety evaluations, thereby enhancing both critique quality and predictive reliability in complex content scenarios. To facilitate further research and application, we open-source SAFETY-J's training protocols, datasets, and code at https://github.com/GAIR-NLP/Safety-J.

cs.CL↗

arXiv2024

Data Contamination Report from the 2024 CONDA Shared Task

The 1st Workshop on Data Contamination (CONDA 2024) focuses on all relevant aspects of data contamination in natural language processing, where data contamination is understood as situations where evaluation data is included in pre-training corpora used to train large scale models, compromising evaluation results. The workshop fostered a shared task to collect evidence on data contamination in current available datasets and models. The goal of the shared task and associated database is to assist the community in understanding the extent of the problem and to assist researchers in avoiding reporting evaluation results on known contaminated resources. The shared task provides a structured, centralized public database for the collection of contamination evidence, open to contributions from the community via GitHub pool requests. This first compilation paper is based on 566 reported entries over 91 contaminated sources from a total of 23 contributors. The details of the individual contamination events are available in the platform. The platform continues to be online, open to contributions from the community.

cs.CL↗

arXiv2024

On Learning to Summarize with Large Language Models as References

Recent studies have found that summaries generated by large language models (LLMs) are favored by human annotators over the original reference summaries in commonly used summarization datasets. Therefore, we study an LLM-as-reference learning setting for smaller text summarization models to investigate whether their performance can be substantially improved. To this end, we use LLMs as both oracle summary generators for standard supervised fine-tuning and oracle summary evaluators for efficient contrastive learning that leverages the LLMs' supervision signals. We conduct comprehensive experiments with source news articles and find that (1) summarization models trained under the LLM-as-reference setting achieve significant performance improvement in both LLM and human evaluations; (2) contrastive learning outperforms standard supervised fine-tuning under both low and high resource settings. Our experimental results also enable a meta-analysis of LLMs' summary evaluation capacities under a challenging setting, showing that LLMs are not well-aligned with human evaluators. Particularly, our expert human evaluation reveals remaining nuanced performance gaps between LLMs and our fine-tuned models, which LLMs fail to capture. Thus, we call for further studies into both the potential and challenges of using LLMs in summarization model development.

cs.CL↗

arXiv2024

Halu-J: Critique-Based Hallucination Judge

Large language models (LLMs) frequently generate non-factual content, known as hallucinations. Existing retrieval-augmented-based hallucination detection approaches typically address this by framing it as a classification task, evaluating hallucinations based on their consistency with retrieved evidence. However, this approach usually lacks detailed explanations for these evaluations and does not assess the reliability of these explanations. Furthermore, deficiencies in retrieval systems can lead to irrelevant or partially relevant evidence retrieval, impairing the detection process. Moreover, while real-world hallucination detection requires analyzing multiple pieces of evidence, current systems usually treat all evidence uniformly without considering its relevance to the content. To address these challenges, we introduce Halu-J, a critique-based hallucination judge with 7 billion parameters. Halu-J enhances hallucination detection by selecting pertinent evidence and providing detailed critiques. Our experiments indicate that Halu-J outperforms GPT-4o in multiple-evidence hallucination detection and matches its capability in critique generation and evidence selection. We also introduce ME-FEVER, a new dataset designed for multiple-evidence hallucination detection. Our code and dataset can be found in https://github.com/GAIR-NLP/factool .

cs.CL↗

arXiv2024

Benchmarking Generation and Evaluation Capabilities of Large Language Models for Instruction Controllable Summarization

While large language models (LLMs) can already achieve strong performance on standard generic summarization benchmarks, their performance on more complex summarization task settings is less studied. Therefore, we benchmark LLMs on instruction controllable text summarization, where the model input consists of both a source article and a natural language requirement for desired summary characteristics. To this end, we curate an evaluation-only dataset for this task setting and conduct human evaluations of five LLM-based systems to assess their instruction-following capabilities in controllable summarization. We then benchmark LLM-based automatic evaluation for this task with 4 different evaluation protocols and 11 LLMs, resulting in 40 evaluation methods. Our study reveals that instruction controllable text summarization remains a challenging task for LLMs, since (1) all LLMs evaluated still make factual and other types of errors in their summaries; (2) no LLM-based evaluation methods can achieve a strong alignment with human annotators when judging the quality of candidate summaries; (3) different LLMs show large performance gaps in summary generation and evaluation capabilities. We make our collected benchmark InstruSum publicly available to facilitate future research in this direction.

cs.CL↗

arXiv2024

BeHonest: Benchmarking Honesty in Large Language Models

Previous works on Large Language Models (LLMs) have mainly focused on evaluating their helpfulness or harmlessness. However, honesty, another crucial alignment criterion, has received relatively less attention. Dishonest behaviors in LLMs, such as spreading misinformation and defrauding users, present severe risks that intensify as these models approach superintelligent levels. Enhancing honesty in LLMs addresses critical limitations and helps uncover latent capabilities that are not readily expressed. This underscores the urgent need for reliable methods and benchmarks to effectively ensure and evaluate the honesty of LLMs. In this paper, we introduce BeHonest, a pioneering benchmark specifically designed to assess honesty in LLMs comprehensively. BeHonest evaluates three essential aspects of honesty: awareness of knowledge boundaries, avoidance of deceit, and consistency in responses. Building on this foundation, we designed 10 scenarios to evaluate and analyze 9 popular LLMs on the market, including both closed-source and open-source models from different model families with varied model sizes. Our findings indicate that there is still significant room for improvement in the honesty of LLMs. We encourage the AI community to prioritize honesty alignment in these models, which can harness their full potential to benefit society while preventing them from causing harm through deception or inconsistency. Our benchmark and code can be found at: \url{https://github.com/GAIR-NLP/BeHonest}.

cs.CL↗

arXiv2024

ANOLE: An Open, Autoregressive, Native Large Multimodal Models for Interleaved Image-Text Generation

Previous open-source large multimodal models (LMMs) have faced several limitations: (1) they often lack native integration, requiring adapters to align visual representations with pre-trained large language models (LLMs); (2) many are restricted to single-modal generation; (3) while some support multimodal generation, they rely on separate diffusion models for visual modeling and generation. To mitigate these limitations, we present Anole, an open, autoregressive, native large multimodal model for interleaved image-text generation. We build Anole from Meta AI's Chameleon, adopting an innovative fine-tuning strategy that is both data-efficient and parameter-efficient. Anole demonstrates high-quality, coherent multimodal generation capabilities. We have open-sourced our model, training framework, and instruction tuning data.

cs.CL↗

arXiv2024

Progress or Regress? Self-Improvement Reversal in Post-training

Self-improvement through post-training methods such as iterative preference learning has been acclaimed for enhancing the problem-solving capabilities (e.g., mathematical reasoning) of Large Language Models (LLMs) without human intervention. However, as exploration deepens, it becomes crucial to assess whether these improvements genuinely signify progress in solving more challenging problems or if they could lead to unintended regressions. To address this, we propose a comprehensive evaluative framework that goes beyond the superficial pass@1 metric to scrutinize the underlying enhancements of post-training paradigms for self-improvement. Through rigorous experimentation and analysis across diverse problem-solving tasks, the empirical results point out the phenomenon of \emph{self-improvement reversal}, where models showing improved performance across benchmarks will paradoxically exhibit declines in broader, essential capabilities, like output diversity and out-of-distribution (OOD) generalization. These findings indicate that current self-improvement practices through post-training are inadequate for equipping models to tackle more complex problems. Furthermore, they underscore the necessity of our critical evaluation metrics in discerning the \emph{progress or regress} dichotomy for self-improving LLMs.

cs.CL↗

arXiv2024

FRoG: Evaluating Fuzzy Reasoning of Generalized Quantifiers in Large Language Models

Fuzzy reasoning is vital due to the frequent use of imprecise information in daily contexts. However, the ability of current large language models (LLMs) to handle such reasoning remains largely uncharted. In this paper, we introduce a new benchmark, FRoG, for fuzzy reasoning, featuring real-world mathematical word problems that incorporate generalized quantifiers. Our experimental findings reveal that fuzzy reasoning continues to pose significant challenges for LLMs. Moreover, we find that existing methods designed to enhance reasoning do not consistently improve performance in tasks involving fuzzy logic. Additionally, our results show an inverse scaling effect in the performance of LLMs on FRoG. Interestingly, we also demonstrate that strong mathematical reasoning skills are not necessarily indicative of success on our benchmark.

cs.AI↗

arXiv2024

Impact of Domain Knowledge and Multi-Modality on Intelligent Molecular Property Prediction: A Systematic Survey

The precise prediction of molecular properties is essential for advancements in drug development, particularly in virtual screening and compound optimization. The recent introduction of numerous deep learning-based methods has shown remarkable potential in enhancing molecular property prediction (MPP), especially improving accuracy and insights into molecular structures. Yet, two critical questions arise: does the integration of domain knowledge augment the accuracy of molecular property prediction and does employing multi-modal data fusion yield more precise results than unique data source methods? To explore these matters, we comprehensively review and quantitatively analyze recent deep learning methods based on various benchmarks. We discover that integrating molecular information significantly improves molecular property prediction (MPP) for both regression and classification tasks. Specifically, regression improvements, measured by reductions in root mean square error (RMSE), are up to 4.0%, while classification enhancements, measured by the area under the receiver operating characteristic curve (ROC-AUC), are up to 1.7%. We also discover that enriching 2D graphs with 1D SMILES boosts multi-modal learning performance for regression tasks by up to 9.1%, and augmenting 2D graphs with 3D information increases performance for classification tasks by up to 13.2%, with both enhancements measured using ROC-AUC. The two consolidated insights offer crucial guidance for future advancements in drug discovery.

cs.LG↗

arXiv2024

Electric-field control of the perpendicular magnetization switching in ferroelectric/ferrimagnet heterostructures

Electric field control of the magnetic state in ferrimagnets holds great promise for developing spintronic devices due to low power consumption. Here, we demonstrate a non-volatile reversal of perpendicular net magnetization in a ferrimagnet by manipulating the electric-field driven polarization within the Pb (Zr0.2Ti0.8) O3 (PZT)/CoGd heterostructure. Electron energy loss spectra and X-ray absorption spectrum directly verify that the oxygen ion migration at the PZT/CoGd interface associated with reversing the polarization causes the enhanced/reduced oxidation in CoGd. Ab initio calculations further substantiate that the migrated oxygen ions can modulate the relative magnetization of Co/Gd sublattices, facilitating perpendicular net magnetization switching. Our findings offer an approach to effectively control ferrimagnetic net magnetization, holding significant implications for ferrimagnetic spintronic applications.

cond-mat.mtrl-sci↗

arXiv2024

OlympicArena Medal Ranks: Who Is the Most Intelligent AI So Far?

In this report, we pose the following question: Who is the most intelligent AI model to date, as measured by the OlympicArena (an Olympic-level, multi-discipline, multi-modal benchmark for superintelligent AI)? We specifically focus on the most recently released models: Claude-3.5-Sonnet, Gemini-1.5-Pro, and GPT-4o. For the first time, we propose using an Olympic medal Table approach to rank AI models based on their comprehensive performance across various disciplines. Empirical results reveal: (1) Claude-3.5-Sonnet shows highly competitive overall performance over GPT-4o, even surpassing GPT-4o on a few subjects (i.e., Physics, Chemistry, and Biology). (2) Gemini-1.5-Pro and GPT-4V are ranked consecutively just behind GPT-4o and Claude-3.5-Sonnet, but with a clear performance gap between them. (3) The performance of AI models from the open-source community significantly lags behind these proprietary models. (4) The performance of these models on this benchmark has been less than satisfactory, indicating that we still have a long way to go before achieving superintelligence. We remain committed to continuously tracking and evaluating the performance of the latest powerful models on this benchmark (available at https://github.com/GAIR-NLP/OlympicArena).

cs.CL↗

arXiv2024

Accurate polymorphous description of the paramagnetic phases in MnBi$_{2}$Te$_{4}$

Temperature-driven phase transition is a long-standing frontier in material science, among which the most common phenomenon is the transition from a low-temperature magnetic-ordered phase to a high-temperature paramagnetic phase. A paramount question is if such a paramagnetic phase of the 'correlated solids' can be well described by single-particle band theory to facilitate the experimental observations. In this work, we investigate the electronic properties of the paramagnetic phase by the static density functional theory via two different approaches, namely monomorphous description and polymorphous description. In the conventional monomorphous description, the local spin moments are naively forced to be zero. By contrast, the polymorphous description based on a large enough supercell with disordered distributed local moments is able to count in the effects of distinct local environments, providing a more reliable paramagnetic electronic structure to simulate realistic materials. From a comparison of total energies, symmetries, and band structures, we demonstrate the necessity for a proper treatment of paramagnetic phases, taking MnBi$_{2}$Te$_{4}$ as an example. Our work provides a theoretical perspective on the evolution of electronic structures through magnetic order-disorder phase transitions in emergent topological magnets.

cond-mat.mtrl-sci↗

arXiv2024

MedBench: A Comprehensive, Standardized, and Reliable Benchmarking System for Evaluating Chinese Medical Large Language Models

Ensuring the general efficacy and goodness for human beings from medical large language models (LLM) before real-world deployment is crucial. However, a widely accepted and accessible evaluation process for medical LLM, especially in the Chinese context, remains to be established. In this work, we introduce "MedBench", a comprehensive, standardized, and reliable benchmarking system for Chinese medical LLM. First, MedBench assembles the currently largest evaluation dataset (300,901 questions) to cover 43 clinical specialties and performs multi-facet evaluation on medical LLM. Second, MedBench provides a standardized and fully automatic cloud-based evaluation infrastructure, with physical separations for question and ground truth. Third, MedBench implements dynamic evaluation mechanisms to prevent shortcut learning and answer remembering. Applying MedBench to popular general and medical LLMs, we observe unbiased, reproducible evaluation results largely aligning with medical professionals' perspectives. This study establishes a significant foundation for preparing the practical applications of Chinese medical LLMs. MedBench is publicly accessible at https://medbench.opencompass.org.cn.

cs.CL↗

arXiv2024

How Far Are LLMs from Believable AI? A Benchmark for Evaluating the Believability of Human Behavior Simulation

In recent years, AI has demonstrated remarkable capabilities in simulating human behaviors, particularly those implemented with large language models (LLMs). However, due to the lack of systematic evaluation of LLMs' simulated behaviors, the believability of LLMs among humans remains ambiguous, i.e., it is unclear which behaviors of LLMs are convincingly human-like and which need further improvements. In this work, we design SimulateBench to evaluate the believability of LLMs when simulating human behaviors. In specific, we evaluate the believability of LLMs based on two critical dimensions: 1) consistency: the extent to which LLMs can behave consistently with the given information of a human to simulate; and 2) robustness: the ability of LLMs' simulated behaviors to remain robust when faced with perturbations. SimulateBench includes 65 character profiles and a total of 8,400 questions to examine LLMs' simulated behaviors. Based on SimulateBench, we evaluate the performances of 10 widely used LLMs when simulating characters. The experimental results reveal that current LLMs struggle to align their behaviors with assigned characters and are vulnerable to perturbations in certain factors.

cs.CL↗