ACL 2024 Papers

Despite recent advances in large language models, building dependable and deployable NLP models typically requires abundant, high-quality training data. However, task-specific data is not available for many use cases, and manually curating task-specific data is labor-intensive. Recent work has studied prompt-driven synthetic data generation using large language models, but these generated datasets tend to lack complexity and diversity. To address these limitations, we introduce a method, _DataTune_, to make better use of existing, publicly available datasets to improve automatic dataset generation. DataTune performs _dataset transformation_, enabling the repurposing of publicly available datasets into a format that is directly aligned with the specific requirements of target tasks. On a diverse set of language-based tasks from the BIG-Bench benchmark, we find that finetuning language models via DataTune improves over a few-shot prompting baseline by 49% and improves over existing methods that use synthetic or retrieved training data by 34%. We find that dataset transformation significantly increases the diversity and difficulty of generated data on many tasks. We release a Python package and open-source repository to make this method accessible to the community (URL will be added upon acceptance).

Clinical text summarization has proven successful in generating concise and coherent summaries. However, these summaries may include unintended text with hallucinations, which can mislead clinicians and patients. Existing methods for mitigating hallucinations can be categorized into task-specific and task-agnostic approaches. Task-specific methods lack versatility for real-world applicability. Meanwhile, task-agnostic methods are not model-agnostic, so they require retraining for different models, resulting in considerable computational costs. To address these challenges, we propose MEDAL, a model-agnostic framework designed to post-process medical hallucinations. MEDAL can seamlessly integrate with any medical summarization model, requiring no additional computational overhead. MEDAL comprises a medical infilling model and a hallucination correction model. The infilling model generates non-factual summaries with common errors to train the correction model. The correction model is incorporated with a self-examination mechanism to activate its cognitive capability. We conduct comprehensive experiments using 11 widely accepted metrics on 7 baseline models across 3 medical text summarization tasks. MEDAL demonstrates superior performance in correcting hallucinations when applied to summaries generated by pre-trained language models and large language models.

Recent studies have illuminated the promising capabilities of large language models (LLMs) in handling long texts. However, their performance in machine translation (MT) of long documents remains underexplored. This paper aims to shed light on how LLMs navigate this complex task, offering a comprehensive evaluation of their capabilities and limitations in long-text MT. First, we collect and construct an instruction-based benchmark dataset, specifically designed for the finetuning and evaluation of LLMs, encompassing multilingual, multi-domain, and document-level parallel data. Second, we conduct a comprehensive comparison between MT and LLM models concerning document-level translation. Our analysis uncovers that LLMs exhibit shortcomings in long-text domains, and their performance diminishes as document size escalates. By exploiting various extrapolation strategies, we enhance the capacity of LLMs to translate longer texts. We release data, code, and models at https://github.com/longyuewangdcu/Document-MT-LLM.

Compositional generalization, representing the model’s ability to generate text with new attribute combinations obtained by recombining single attributes from the training data, is a crucial property for multi-aspect controllable text generation (MCTG) methods. Nonetheless, a comprehensive compositional generalization evaluation benchmark of MCTG is still lacking. We propose CompMCTG, a benchmark encompassing diverse multi-aspect labeled datasets and a crafted three-dimensional evaluation protocol, to holistically evaluate the compositional generalization of MCTG approaches. We observe that existing MCTG works generally confront a noticeable performance drop in compositional testing. To mitigate this issue, we introduce Meta-MCTG, a training framework incorporating meta-learning, where we enable models to learn how to generalize by simulating compositional generalization scenarios in the training phase. We demonstrate the effectiveness of Meta-MCTG through achieving obvious improvement (by at most 3.64%) for compositional testing performance in 94.4%.

While large language models (LLMs) have achieved state-of-the-art performance on a wide range of medical question answering (QA) tasks, they still face challenges with hallucinations and outdated knowledge. Retrieval-augmented generation (RAG) is a promising solution and has been widely adopted. However, a RAG system can involve multiple flexible components, and there is a lack of best practices regarding the optimal RAG setting for various medical purposes. To systematically evaluate such systems, we propose the Medical Information Retrieval-Augmented Generation Evaluation (MIRAGE), a first-of-its-kind benchmark including 7,663 questions from five medical QA datasets. Using MIRAGE, we conducted large-scale experiments with over 1.8 trillion prompt tokens on 41 combinations of different corpora, retrievers, and backbone LLMs through the MedRAG toolkit introduced in this work. Overall, MedRAG improves the accuracy of six different LLMs by up to 18% over chain-of-thought prompting, elevating the performance of GPT-3.5 and Mixtral to GPT-4-level. Our results show that the combination of various medical corpora and retrievers achieves the best performance. In addition, we discovered a log-linear scaling property and the “lost-in-the-middle” effects in medical RAG. We believe our comprehensive evaluations can serve as practical guidelines for implementing RAG systems for medicine.

Traditional applications of natural language processing (NLP) in healthcare have predominantly focused on patient-centered services, enhancing patient interactions and care delivery, such as through medical dialogue systems. However, the potential of NLP to benefit inexperienced doctors, particularly in areas such as communicative medical coaching, remains largely unexplored. We introduce “ChatCoach”, a human-AI cooperative framework designed to assist medical learners in practicing their communication skills during patient consultations. ChatCoach differentiates itself from conventional dialogue systems by offering a simulated environment where medical learners can practice dialogues with a patient agent, while a coach agent provides immediate, structured feedback. This is facilitated by our proposed Generalized Chain-of-Thought (GCoT) approach, which fosters the generation of structured feedback and enhances the utilization of external knowledge sources. Additionally, we have developed a dataset specifically for evaluating Large Language Models (LLMs) within the ChatCoach framework on communicative medical coaching tasks. Our empirical results validate the effectiveness of ChatCoach.

In light of recent breakthroughs in large language models (LLMs) that have revolutionized natural language processing (NLP), there is an urgent need for new benchmarks to keep pace with the fast development of LLMs. In this paper, we propose CFLUE, the Chinese Financial Language Understanding Evaluation benchmark, designed to assess the capability of LLMs across various dimensions. Specifically, CFLUE provides datasets tailored for both knowledge assessment and application assessment. In knowledge assessment, it consists of 38K+ multiple-choice questions with associated solution explanations. These questions serve dual purposes: answer prediction and question reasoning. In application assessment, CFLUE features 16K+ test instances across distinct groups of NLP tasks such as text classification, machine translation, relation extraction, reading comprehension, and text generation. Upon CFLUE, we conduct a thorough evaluation of representative LLMs. The results reveal that only Qwen-72B, GPT-4, and GPT-4-turbo achieve an accuracy exceeding 60% in answer prediction for knowledge assessment, suggesting that there is still substantial room for improvement in current LLMs. In application assessment, while GPT-4 and GPT-4-turbo rank as the top two performers on average, their significant advantage over open-source LLMs is noticeably diminished, given that Qwen-72B achieves the best performance in 2 out of 5 tasks. The datasets and scripts associated with CFLUE are openly accessible at https://github.com/aliyun/cflue.

In recent years, substantial advancements have been made in the development of large language models, achieving remarkable performance across diverse tasks.To evaluate the knowledge ability of language models, previous studies have proposed lots of benchmarks based on question-answering pairs.We argue that it is not reliable and comprehensive to evaluate language models with a fixed question or limited paraphrases as the query, since language models are sensitive to prompt.Therefore, we introduce a novel concept named knowledge boundary to encompass both prompt-agnostic and prompt-sensitive knowledge within language models.Knowledge boundary avoids prompt sensitivity in language model evaluations, rendering them more dependable and robust.To explore the knowledge boundary for a given model, we propose projected gradient descent method with semantic constraints, a new algorithm designed to identify the optimal prompt for each piece of knowledge.Experiments demonstrate a superior performance of our algorithm in computing the knowledge boundary compared to existing methods.Furthermore, we evaluate the ability of multiple language models in several domains with knowledge boundary.

In the era of data-driven decision-making, the complexity of data analysis necessitates advanced expertise and tools of data science, presenting significant challenges even for specialists. Large Language Models (LLMs) have emerged as promising aids as data science agents, assisting humans in data analysis and processing. Yet their practical efficacy remains constrained by the varied demands of real-world applications and complicated analytical process. In this paper, we introduce DSEval – a novel evaluation paradigm, as well as a series of innovative benchmarks tailored for assessing the performance of these agents throughout the entire data science lifecycle. Incorporating a novel bootstrapped annotation method, we streamline dataset preparation, improve the evaluation coverage, and expand benchmarking comprehensiveness. Our findings uncover prevalent obstacles and provide critical insights to inform future advancements in the field.

Large Language Models (LLMs) have recently been shown to be effective as automatic evaluators with simple prompting and in-context learning. In this work, we assemble 16 LLMs encompassing four different size ranges and evaluate their output responses by preference ranking from the other LLMs as evaluators, such as System Star is better than System Square. We then evaluate the quality of ranking outputs introducing the Cognitive Bias Benchmark for LLMs as Evaluators (CoBBLer), a benchmark to measure six different cognitive biases in LLM evaluation outputs, such as the Egocentric bias where a model prefers to rank its own outputs highly in evaluation. We find that LLMs are biased text quality evaluators, exhibiting strong indications on our bias benchmark (40% of comparisons made by all models) within each of their evaluations that question their robustness as evaluators. Furthermore, we examine the correlation between human and machine preferences and calculate the average Rank-Biased Overlap (RBO) score to be 44%, indicating that machine preferences are misaligned with humans. According to our findings, LLMs may still be unable to be utilized for automatic annotation aligned with human preferences.

We introduce CHARM, the first benchmark for comprehensively and in-depth evaluating the commonsense reasoning ability of large language models (LLMs) in Chinese, which covers both globally known and Chinese-specific commonsense. We evaluated 7 English and 12 Chinese-oriented LLMs on CHARM, employing 5 representative prompt strategies for improving LLMs’ reasoning ability, such as Chain-of-Thought. Our findings indicated that the LLM’s language orientation and the task’s domain influence the effectiveness of the prompt strategy, which enriches previous research findings. We built closely-interconnected reasoning and memorization tasks, and found that some LLMs struggle with memorizing Chinese commonsense, affecting their reasoning ability, while others show differences in reasoning despite similar memorization performance. We also evaluated the LLMs’ memorization-independent reasoning abilities and analyzed the typical errors. Our study precisely identified the LLMs’ strengths and weaknesses, providing the clear direction for optimization. It can also serve as a reference for studies in other fields. We will release CHARM at https://github.com/opendatalab/CHARM.

Open Information Extraction (OIE) is a field of natural language processing that aims to present textual information in a format that allows it to be organized, analyzed and reflected upon. Numerous OIE systems are developed, claiming ever-increasing performance, marking the need for objective benchmarks. BenchIE is the latest reference we know of. Despite being very well thought out, we noticed a number of issues we believe are limiting. Therefore, we propose BenchIE^FL, a new OIE benchmark which fully enforces the principles of BenchIE while containing fewer errors, omissions and shortcomings when candidate facts are matched towards reference ones. BenchIE^FL allows insightful conclusions to be drawn on the actual performance of OIE extractors.

Large Language Models (LLMs) driven by In-Context Learning (ICL) have significantly improved the performance of text-to-SQL. Previous methods generally employ a two-stage reasoning framework, namely 1) schema linking and 2) logical synthesis, making the framework not only effective but also interpretable. Despite these advancements, the inherent bad nature of the generalization of LLMs often results in hallucinations, which limits the full potential of LLMs. In this work, we first identify and categorize the common types of hallucinations at each stage in text-to-SQL. We then introduce a novel strategy, Task Alignment (TA), designed to mitigate hallucinations at each stage. TA encourages LLMs to take advantage of experiences from similar tasks rather than starting the tasks from scratch. This can help LLMs reduce the burden of generalization, thereby mitigating hallucinations effectively. We further propose TA-SQL, a text-to-SQL framework based on this strategy. The experimental results and comprehensive analysis demonstrate the effectiveness and robustness of our framework. Specifically, it enhances the performance of the GPT-4 baseline by 21.23% relatively on BIRD dev and it yields significant improvements across six models and four mainstream, complex text-to-SQL benchmarks.

Large language models (LLMs) have demonstrated strong reasoning capabilities.Nevertheless, they still suffer from factual errors when tackling knowledge-intensive tasks.Retrieval-augmented reasoning represents a promising approach.However, significant challenges still persist, including inaccurate and insufficient retrieval for complex questions, as well as difficulty in integrating multi-source knowledge.To address this, we propose Beam Aggregation Reasoning (BeamAggR), a reasoning framework for knowledge-intensive multi-hop QA.BeamAggR explores and prioritizes promising answers at each hop of question.Concretely, we parse the complex questions into trees, which include atom and composite questions, followed by bottom-up reasoning.For atomic questions, the LLM conducts reasoning on multi-source knowledge to get answer candidates.For composite questions, the LLM combines beam candidates, explores multiple reasoning paths through probabilistic aggregation, and prioritizes the most promising trajectory.Extensive experiments on four open-domain multi-hop reasoning datasets show that our method significantly outperforms SOTA methods by 8.5%.Furthermore, our analysis reveals that BeamAggR elicits better knowledge collaboration and answer aggregation.

An important requirement for the reliable deployment of pre-trained large language models (LLMs) is the well-calibrated quantification of the uncertainty in their outputs. While the likelihood of predicting the next token is a practical surrogate of the data uncertainty learned during training, model uncertainty is challenging to estimate, i.e., due to lack of knowledge acquired during training. Prior efforts to quantify uncertainty of neural networks require specific architectures or (re-)training strategies, which are impractical to apply to LLMs with several billion parameters, or for black-box models where the architecture and parameters are not available. In this paper, we propose Bayesian Prompts Ensembles (BayesPE), a novel approach to effectively obtain well-calibrated uncertainty for the output of pre-trained LLMs. BayesPE computes output probabilities through a weighted ensemble of different, but semantically equivalent, task instruction prompts. The relative weights of the different prompts in the ensemble are estimated through approximate Bayesian variational inference over a small labeled validation set. We demonstrate that BayesPE approximates a Bayesian input layer for the LLM, providing a lower bound on the expected model error. In our extensive experiments, we show that BayesPE achieves significantly superior uncertainty calibration compared to several baselines over a range of natural language classification tasks, both in zero- and few-shot settings.

Significant progress has been made in automatic text evaluation with the introduction of large language models (LLMs) as evaluators. However, current sample-wise evaluation paradigm suffers from the following issues: (1) Sensitive to prompt design; (2) Poor resistance to noise; (3) Inferior ensemble performance with static reference. Inspired by the fact that humans treat both criterion definition and inter sample comparison as references for evaluation, we propose BatchEval, a paradigm that conducts batch-wise evaluation iteratively to alleviate the above problems. We explore variants under this paradigm and confirm the optimal settings are two stage procedure with heterogeneous batch composition strategy and decimal scoring format. Comprehensive experiments across 3 LLMs on 4 text evaluation tasks demonstrate that BatchEval outperforms state-of-the-art methods by 10.5% on Pearson correlations with only 64% API cost on average. Further analyses have been conducted to verify the robustness, generalization, and working mechanism of BatchEval.

In this paper, by treating in-context learning (ICL) as a meta-optimization process, we explain why LLMs are sensitive to the order of ICL examples. This understanding leads us to the development of Batch-ICL, an effective, efficient, and order-agnostic inference algorithm for ICL. Differing from the standard N-shot learning approach, Batch-ICL employs N separate 1-shot forward computations and aggregates the resulting meta-gradients. These aggregated meta-gradients are then applied to the forward computation of a zero-shot query to generate the final prediction. This batch processing approach renders the LLM agnostic to the order of ICL examples. Through extensive experiments and analysis, we demonstrate that Batch-ICL consistently outperforms most permutations of ICL examples. In some cases, it even exceeds the performance of the best order for standard ICL, all while reducing the computational resources required. Furthermore, we develop a novel variant of Batch-ICL featuring multiple “epochs” of meta-optimization. This variant implicitly explores permutations of ICL examples, further enhancing ICL performance.

Aligned Large Language Models (LLMs) showcase remarkable versatility, capable of handling diverse real-world tasks. Meanwhile, aligned LLMs are also expected to exhibit speciality, excelling in specific applications. However, fine-tuning with extra data, a common practice to gain speciality, often leads to catastrophic forgetting (CF) of previously acquired versatility, hindering the model’s performance across diverse tasks. In response to this challenge, we propose CoFiTune, a coarse to fine framework in an attempt to strike the balance between speciality and versatility. At the coarse-grained level, an empirical tree-search algorithm is utilized to pinpoint and update specific modules that are crucial for speciality, while keeping other parameters frozen; at the fine-grained level, a soft-masking mechanism regulates the update to the LLMs, mitigating the CF issue without harming speciality. In an overall evaluation of both speciality and versatility, CoFiTune consistently outperforms baseline methods across diverse tasks and model scales. Compared to the full-parameter SFT, CoFiTune leads to about 14% versatility improvement and marginal speciality loss on a 13B model. Lastly, based on further analysis, we provide a speculative insight into the information forwarding process in LLMs, which helps explain the effectiveness of the proposed method. The code is available at https://github.com/rattlesnakey/CoFiTune.

Data plays a fundamental role in the training of Large Language Models (LLMs). While attention has been paid to the collection and composition of datasets, determining the data sampling strategy in training remains an open question. Most LLMs are trained with a simple strategy, random sampling. However, this sampling strategy ignores the unbalanced nature of training data distribution, which can be sub-optimal. In this paper, we propose ClusterClip Sampling to balance the text distribution of training data for better model training. Specifically, ClusterClip Sampling utilizes data clustering to reflect the data distribution of the training set and balances the common samples and rare samples during training based on the cluster results. A repetition clip operation is introduced to mitigate the overfitting issue led by samples from certain clusters. Extensive experiments validate the effectiveness of ClusterClip Sampling, which outperforms random sampling and other cluster-based sampling variants under various training datasets and large language models.