ACL 2024 Papers

Large language models (LLMs) have shown impressive success in various applications. However, these models are often not well aligned with human intents, which calls for additional treatments on them; that is, the alignment problem. To make LLMs better follow user instructions, existing alignment methods primarily focus on further training them. However, the extra training of LLMs is usually expensive in terms of GPU computing; even worse, some LLMs are not accessible for user-demanded training, such as GPTs. In this work, we take a different perspective—Black-Box Prompt Optimization (BPO)—to perform alignments. The idea is to optimize user prompts to suit LLMs’ input understanding, so as to best realize users’ intents without updating LLMs’ parameters. BPO leverages human preferences to optimize prompts, thus making it superior to LLM (e.g., ChatGPT) as a prompt engineer. Moreover, BPO is model-agnostic, and the empirical results demonstrate that the BPO-aligned ChatGPT yields a 22% increase in the win rate against its original version and 10% for GPT-4. Notably, the BPO-aligned LLMs can outperform the same models aligned by PPO and DPO, and it also brings additional performance gains when combining BPO with PPO or DPO. Code and datasets are released at https://github.com/thu-coai/BPO.

Answering questions within business and finance requires reasoning, precision, and a wide-breadth of technical knowledge. Together, these requirements make this domain difficult for large language models (LLMs). We introduce BizBench, a benchmark for evaluating models’ ability to reason about realistic financial problems. BizBench comprises eight quantitative reasoning tasks, focusing on question-answering (QA) over financial data via program synthesis. We include three financially-themed code-generation tasks from newly collected and augmented QA data. Additionally, we isolate the reasoning capabilities required for financial QA: reading comprehension of financial text and tables for extracting intermediate values, and understanding financial concepts and formulas needed to calculate complex solutions. Collectively, these tasks evaluate a model’s financial background knowledge, ability to parse financial documents, and capacity to solve problems with code. We conduct an in-depth evaluation of open-source and commercial LLMs, comparing and contrasting the behavior of code-focused and language-focused models. We demonstrate that the current bottleneck in performance is due to LLMs’ limited business and financial understanding, highlighting the value of a challenging benchmark for quantitative reasoning within this domain.

The upscaling of Large Language Models (LLMs) has yielded impressive advances in natural language processing, yet it also poses significant deployment challenges. Weight quantization has emerged as a widely embraced solution to reduce memory and computational demands. This paper introduces BitDistiller, a framework that synergizes Quantization-Aware Training (QAT) with Knowledge Distillation (KD) to boost the performance of LLMs at ultra-low precisions (sub-4-bit). Specifically, BitDistiller first incorporates a tailored asymmetric quantization and clipping technique to maximally preserve the fidelity of quantized weights, and then proposes a novel Confidence-Aware Kullback-Leibler Divergence (CAKLD) objective, which is employed in a self-distillation manner to enable faster convergence and superior model performance. Empirical evaluations demonstrate that BitDistiller significantly surpasses existing methods in both 3-bit and 2-bit configurations on general language understanding and complex reasoning benchmarks. Notably, BitDistiller is shown to be more cost-effective, demanding fewer data and training resources. The code is available at https://github.com/DD-DuDa/BitDistiller.

Recent research trends in computational biology have increasingly focused on integrating text and bio-entity modeling, especially in the context of molecules and proteins. However, previous efforts like BioT5 faced challenges in generalizing across diverse tasks and lacked a nuanced understanding of molecular structures, particularly in their textual representations (e.g., IUPAC). This paper introduces BioT5+, an extension of the BioT5 framework, tailored to enhance biological research and drug discovery. BioT5+ incorporates several novel features: integration of IUPAC names for molecular understanding, inclusion of extensive bio-text and molecule data from sources like bioRxiv and PubChem, the multi-task instruction tuning for generality across tasks, and a numerical tokenization technique for improved processing of numerical data. These enhancements allow BioT5+ to bridge the gap between molecular representations and their textual descriptions, providing a more holistic understanding of biological entities, and largely improving the grounded reasoning of bio-text and bio-sequences. The model is pre-trained and fine-tuned with a large number of experiments, including 3 types of problems (classification, regression, generation), 15 kinds of tasks, and 21 total benchmark datasets, demonstrating the remarkable performance and state-of-the-art results in most cases. BioT5+ stands out for its ability to capture intricate relationships in biological data, thereby contributing significantly to bioinformatics and computational biology. Our code is available at https://github.com/QizhiPei/BioT5.

Large Language Models (LLMs) have demonstrated remarkable versatility in recent years, offering potential applications across specialized domains such as healthcare and medicine. Despite the availability of various open-source LLMs tailored for health contexts, adapting general-purpose LLMs to the medical domain presents significant challenges.In this paper, we introduce BioMistral, an open-source LLM tailored for the biomedical domain, utilizing Mistral as its foundation model and further pre-trained on PubMed Central. We conduct a comprehensive evaluation of BioMistral on a benchmark comprising 10 established medical question-answering (QA) tasks in English. We also explore lightweight models obtained through quantization and model merging approaches. Our results demonstrate BioMistral’s superior performance compared to existing open-source medical models and its competitive edge against proprietary counterparts. Finally, to address the limited availability of data beyond English and to assess the multilingual generalization of medical LLMs, we automatically translated and evaluated this benchmark into 7 other languages. This marks the first large-scale multilingual evaluation of LLMs in the medical domain. Datasets, multilingual evaluation benchmarks, scripts, and all the models obtained during our experiments are freely released.

Real world deployments of word alignment are almost certain to cover both high and low resource languages. However, the state-of-the-art for this task recommends a different model class depending on the availability of gold alignment training data for a particular language pair. We propose BinaryAlign, a novel word alignment technique based on binary sequence labeling that outperforms existing approaches in both scenarios, offering a unifying approach to the task. Additionally, we vary the specific choice of multilingual foundation model, perform stratified error analysis over alignment error type, and explore the performance of BinaryAlign on non-English language pairs. We make our source code publicly available.

Discourse parsing is a crucial task in natural language processing that aims to reveal the higher-level relations in a text. Despite growing interest in cross-lingual discourse parsing, challenges persist due to limited parallel data and inconsistencies in the Rhetorical Structure Theory (RST) application across languages and corpora. To address this, we introduce a parallel Russian annotation for the large and diverse English GUM RST corpus. Leveraging recent advances, our end-to-end RST parser achieves state-of-the-art results on both English and Russian corpora. It demonstrates effectiveness in both monolingual and bilingual settings, successfully transferring even with limited second-language annotation. To the best of our knowledge, this work is the first to evaluate the potential of cross-lingual end-to-end RST parsing on a manually annotated parallel corpus.

Using novel approaches to dataset development, the Biasly dataset captures the nuance and subtlety of misogyny in ways that are unique within the literature. Built in collaboration with multi-disciplinary experts and annotators themselves, the dataset contains annotations of movie subtitles, capturing colloquial expressions of misogyny in North American film. The open-source dataset can be used for a range of NLP tasks, including binary and multi-label classification, severity score regression, and text generation for rewrites. In this paper, we discuss the methodology used, analyze the annotations obtained, provide baselines for each task using common NLP algorithms, and furnish error analyses to give insight into model behaviour when fine-tuned on the Biasly dataset.

Summarization is an important application of large language models (LLMs). Most previous evaluation of summarization models has focused on their content selection, faithfulness, grammaticality and coherence. However, it is well known that LLMs can reproduce and reinforce harmful social biases. This raises the question: Do biases affect model outputs in a constrained setting like summarization?To help answer this question, we first motivate and introduce a number of definitions for biased behaviours in summarization models, along with practical operationalizations. Since we find that biases inherent to input documents can confound bias analysis in summaries, we propose a method to generate input documents with carefully controlled demographic attributes. This allows us to study summarizer behavior in a controlled setting, while still working with realistic input documents.We measure gender bias in English summaries generated by both purpose-built summarization models and general purpose chat models as a case study. We find content selection in single document summarization to be largely unaffected by gender bias, while hallucinations exhibit evidence of bias.To demonstrate the generality of our approach, we additionally investigate racial bias, including intersectional settings.

The knowledge graph-to-text (KG-to-text) generation task aims to synthesize coherent and engaging sentences that accurately convey the complex information derived from an input knowledge graph. Existing methods generate the whole target text based on all KG triples at once and may incorporate incorrect KG triples for each sentence. To this end, we propose the bi-directional multi-granularity generation framework. Instead of generating the whole text at a time, we construct the sentence level generation based on the corresponding triples and generate the graph-level text as a result. Moreover, we design a backward relation extraction task to enhance the correctness of relational information. Our method achieves the new state-of-the-art in benchmark dataset WebNLG and further analysis shows the efficiency of different modules.

Large Language Models (LLMs) have shown human-like reasoning abilities but still face challenges in solving complex logical problems. Existing unidirectional chaining methods, such as forward chaining and backward chaining, suffer from issues like low prediction accuracy and efficiency. To address these, we propose a bidirectional chaining method, Bi-Chainer, which dynamically switches to depth-first reasoning in the opposite reasoning direction when it encounters multiple branching options within the current direction. Thus, the intermediate reasoning results can be utilized as guidance to facilitate the reasoning process. We show that Bi-Chainer achieves sizable accuracy boots over unidirectional chaining frameworks on four challenging logical reasoning datasets. Moreover, Bi-Chainer enhances the accuracy of intermediate proof steps and reduces the average number of inference calls, resulting in more efficient and accurate reasoning.

We analyze the behaviors of open large language models (LLMs) on the task of data-to-text (D2T) generation, i.e., generating coherent and relevant text from structured data. To avoid the issue of LLM training data contamination with standard benchmarks, we design Quintd - a tool for collecting novel structured data records from public APIs. We find that open LLMs (Llama 2, Mistral, and Zephyr) can generate fluent and coherent texts in zero-shot settings from data in common formats collected with Quintd. However, we show that the semantic accuracy of the outputs is a major issue: both according to human annotators and our reference-free metric based on GPT-4, more than 80% of the outputs of open LLMs contain at least one semantic error. We publicly release the code, data, and model outputs.

Supervised machine-learning models for predicting user behavior offer a challenging classification problem with lower average prediction performance scores than other text classification tasks. This study evaluates multi-task learning frameworks grounded in Cognitive Appraisal Theory to predict user behavior as a function of users’ self-expression and psychological attributes. Our experiments show that users’ language and traits improve predictions above and beyond models predicting only from text. Our findings highlight the importance of integrating psychological constructs into NLP to enhance the understanding and prediction of user actions. We close with a discussion of the implications for future applications of large language models for computational psychology.

Recent mainstream event argument extraction methods process each event in isolation, resulting in inefficient inference and ignoring the correlations among multiple events. To address these limitations, here we propose a multiple-event argument extraction model DEEIA (Dependency-guided Encoding and Event-specific Information Aggregation), capable of extracting arguments from all events within a document simultaneously. The proposed DEEIA model employs a multi-event prompt mechanism, comprising DE and EIA modules. The DE module is designed to improve the correlation between prompts and their corresponding event contexts, whereas the EIA module provides event-specific information to improve contextual understanding. Extensive experiments show that our method achieves new state-of-the-art performance on four public datasets (RAMS, WikiEvents, MLEE, and ACE05), while significantly saving the inference time compared to the baselines. Further analyses demonstrate the effectiveness of the proposed modules.

Model scaling is becoming the default choice for many language tasks due to the success of large language models (LLMs). However, it can fall short in specific scenarios where simple customized methods excel. In this paper, we delve into the patent approval prediction task and unveil that simple domain-specific graph methods outperform enlarging the model, using the intrinsic dependencies within the patent data. Specifically, we first extend the embedding-based state-of-the-art (SOTA) by scaling up its backbone model with various sizes of open-source LLMs, then explore prompt-based methods to harness proprietary LLMs’ potential, but find the best results close to random guessing, underlining the ineffectiveness of model scaling-up. Hence, we propose a novel Fine-grained cLAim depeNdency (FLAN) Graph through meticulous patent data analyses, capturing the inherent dependencies across segments of the patent text. As it is model-agnostic, we apply cost-effective graph models to our FLAN Graph to obtain representations for approval prediction. Extensive experiments and detailed analyses prove that incorporating FLAN Graph via various graph models consistently outperforms all LLM baselines significantly. We hope that our observations and analyses in this paper can bring more attention to this challenging task and prompt further research into the limitations of LLMs.

In argumentation theory, argument schemes are a characterisation of stereotypical patterns of inference. There has been little work done to develop computational approaches to identify these schemes in natural language. Moreover, advancements in recognizing textual entailment lack a standardized definition of inference, which makes it challenging to compare methods trained on different datasets and rely on the generalisability of their results. In this work, we propose a rigorous approach to align entailment recognition with argumentation theory. Wagemans’ Periodic Table of Arguments (PTA), a taxonomy of argument schemes, provides the appropriate framework to unify these two fields. To operationalise the theoretical model, we introduce a tool to assist humans in annotating arguments according to the PTA. Beyond providing insights into non-expert annotator training, we present Kialo-PTA24, the first multi-topic dataset for the PTA. Finally, we benchmark the performance of pre-trained language models on various aspects of argument analysis. Our experiments show that the task of argument canonicalisation poses a significant challenge for state-of-the-art models, suggesting an inability to represent argumentative reasoning and a direction for future investigation.

This paper presents a novel Dialectal Sound and Vowelization Recovery framework, designed to recognize borrowed and dialectal sounds within phonologically diverse and dialect-rich languages, that extends beyond its standard orthographic sound sets. The proposed framework utilized quantized sequence of input with(out) continuous pretrained self-supervised representation. We show the efficacy of the pipeline using limited data for Arabic, a dialect-rich language containing more than 22 major dialects. Phonetically correct transcribed speech resources for dialectal Arabic is scare. Therefore, we introduce ArabVoice15, a first of its kind, curated test set featuring 5 hours of dialectal speech across 15 Arab countries, with phonetically accurate transcriptions, including borrowed and dialect-specific sounds. We described in detail the annotation guideline along with the analysis of the dialectal confusion pairs. Our extensive evaluation includes both subjective – human perception tests and objective measures. Our empirical results, reported with three test sets, show that with only one and half hours of training data, our model improve character error rate by ≈7% in ArabVoice15 compared to the baseline.

A single language model, even when aligned with labelers through reinforcement learning from human feedback (RLHF), may not suit all human preferences. Recent approaches therefore prefer customization, gathering multi-dimensional feedback, and creating distinct reward models for each dimension.Different language models are then optimized for various preferences using multi-objective RLHF (MORLHF) with varying reward weights.However, RL fine-tuning is unstable and resource-heavy, especially with diverse and usually conflicting objectives.In this paper, we present Multi-Objective Direct Preference Optimization (MODPO), an RL-free extension of Direct Preference Optimization (DPO) for multiple alignment objectives.Essentially, MODPO folds language modeling directly into reward modeling, training language models as implicit collective reward models that combine all objectives with specific weights. MODPO theoretically yields the same optimal solutions as MORLHF but is practically more stable and efficient.Empirical results in safety alignment and long-form question answering show that MODPO matches or outperforms existing methods, producing a Pareto front of language models catering to diverse preferences with three times less computational resources compared to MORLHF.Code is available at https://github.com/ZHZisZZ/modpo.

Recent developments in Language Models (LMs) have shown their effectiveness in NLP tasks, particularly in knowledge-intensive tasks.However, the mechanisms underlying knowledge storage and memory access within their parameters remain elusive. In this paper, we investigate whether a generative LM (e.g., GPT-2) is able to access its memory sequentially or randomly. Through carefully-designed synthetic tasks, covering the scenarios of full recitation, selective recitation and grounded question answering, we reveal that LMs manage to sequentially access their memory while encountering challenges in randomly accessing memorized content. We find that techniques including recitation and permutation improve the random memory access capability of LMs. Furthermore, by applying this intervention to realistic scenarios of open-domain question answering, we validate that enhancing random access by recitation leads to notable improvements in question answering. The code to reproduce our experiments can be found at https://github.com/sail-sg/lm-random-memory-access.