Thailand

Generation of plausible yet incorrect factual information, termed hallucination, is an unsolved issue in large language models. We study the ability of language models to deliberate on the responses they give in order to correct their mistakes. We develop the Chain-of-Verification (CoVe) method whereby the model first (i) drafts an initial response; then (ii) plans verification questions to fact-check its draft; (iii) answers those questions independently so the answers are not biased by other responses; and (iv) generates its final verified response. In experiments, we show CoVe decreases hallucinations across a variety of tasks, from list-based questions from Wikidata, closed book MultiSpanQA and longform text generation.

ACL 2024

Chain-of-Verification Reduces Hallucination in Large Language Models

llms

hallucination

reasoning

poster

### Welcome!
The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) will take place in Bangkok, Thailand from August 11th to 16th, 2024. Our Virtual Poster Sessions will take place online Thursday, August 22, 2024.

You are required to register for this event. **Please register [here](https://2024.aclweb.org/registration). **

If you have already registered, please check your inbox for an email from Underline granting you access to ACL 2024 content.

Please register!

The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) will take place in Bangkok, Thailand from August 11th to 16th, 2024. More information will be announced soon.

The detection of machine-generated text, especially from large language models (LLMs), is crucial in preventing serious social problems resulting from their misuse. Some methods train dedicated detectors on specific datasets but fall short in generalizing to unseen test data, while other zero-shot ones often yield suboptimal performance. Although the recent DetectGPT has shown promising detection performance, it suffers from significant inefficiency issues, as detecting a single candidate requires querying the source LLM with hundreds of its perturbations. This paper aims to bridge this gap. Concretely, we propose to incorporate a Bayesian surrogate model, which allows us to select typical samples based on Bayesian uncertainty and interpolate scores from typical samples to other samples, to improve query efficiency. Empirical results demonstrate that our method significantly outperforms existing approaches under a low query budget. Notably, when detecting the text generated by LLaMA family models, our method with just 2 or 3 queries can outperform DetectGPT with 200 queries.

Efficient Detection of LLM-generated Texts with a Bayesian Surrogate Model

As artificial neural networks grow in complexity, understanding their inner workings becomes increasingly challenging, which is particularly important in healthcare applications. The intrinsic evaluation metrics of autoregressive neural language models (NLMs), perplexity (PPL), can reflect how ``surprised'' an NLM model is at novel input. PPL has been widely used to understand the behavior of NLMs. Previous findings show that changes in PPL when masking attention layers in pre-trained transformer-based NLMs reflect linguistic anomalies associated with Alzheimer's disease dementia. Building upon this, we explore a novel bidirectional attention head ablation method that exhibits properties attributed to the concepts of cognitive and brain reserve in human brain studies, which postulate that people with more neurons in the brain and more efficient processing are more resilient to neurodegeneration. Our results show that larger GPT-2 models require a disproportionately larger share of attention heads to be masked/ablated to display degradation of similar magnitude to masking in smaller models. These results suggest that the attention mechanism in transformer models may present an analogue to the notions of cognitive and brain reserve and could potentially be used to model certain aspects of the progression of neurodegenerative disorders and aging.

Too Big to Fail: Larger Language Models are Disproportionately Resilient to Induction of Dementia-Related Linguistic Anomalies

Pretrained large language models (LLMs) are currently state-of-the-art for solving the vast majority of natural language processing tasks. While many real-world applications still require fine-tuning to reach satisfactory levels of performance, many of them are in the low-data regime, making fine-tuning challenging. To address this, we propose LLM2LLM, a targeted and iterative data augmentation strategy that uses a teacher LLM to enhance a small seed dataset by augmenting additional data that can be used for fine-tuning on a specific task. LLM2LLM (1) fine-tunes a baseline student LLM on the initial seed data, (2) evaluates and extracts data points that the model gets wrong, and (3) uses a teacher LLM to generate synthetic data based on these incorrect data points, which are then added back into the training data. This approach amplifies the signal from incorrectly predicted data points by the LLM during training and reintegrates them into the dataset to focus on more challenging examples for the LLM. Our results show that LLM2LLM significantly enhances the performance of LLMs in the low-data regime, outperforming both traditional fine-tuning and other data augmentation baselines. LLM2LLM reduces the dependence on labor-intensive data curation and paves the way for more scalable and performant LLM solutions, allowing us to tackle data-constrained domains and tasks. We achieve improvements up to 24.2% on the GSM8K dataset, 32.6% on CaseHOLD, 32.0% on SNIPS, 52.6% on TREC and 39.8% on SST-2 over regular fine-tuning in the low-data regime using a Llama-2-7B student model. Our code is available at https://github.com/SqueezeAILab/LLM2LLM.

LLM2LLM: Boosting LLMs with Novel Iterative Data Enhancement

Structural extraction of events within discourse is critical since it avails a deeper understanding of communication patterns and behavior trends. Event argument extraction (EAE), at the core of event-centric understanding, is the task of identifying role-specific text spans (i.e., arguments) for a given event. Document-level EAE (DocEAE) focuses on arguments that are scattered across an entire document. In this work, we explore open-source Large Language Models (LLMs) for DocEAE, and propose ULTRA, a hierarchical framework that extracts event arguments more cost-effectively. Further, it alleviates the positional bias issue intrinsic to LLMs. ULTRA sequentially reads text chunks of a document to generate a candidate argument set, upon which non-pertinent candidates are dropped through self-refinement. We introduce LEAFER to address the challenge LLMs face in locating the exact boundary of an argument. ULTRA outperforms strong baselines, including strong supervised models and ChatGPT, by 9.8% when evaluated by Exact Match (EM).

ULTRA: Unleash LLMs’ Potential for Event Argument Extraction through Hierarchical Modeling and Pair-wise Self-Refinement

Instruction-tuning language models has become a crucial step in aligning them for general use. Typically, this process involves extensive training on large datasets, incurring high training costs. In this paper, we introduce a novel training data selection based on the learning percentage of the samples. We assert that current language models possess the capability to autonomously select high-quality training data, leading to comparable or improved performance compared to training on the entire dataset. Our experiments span different-sized models, revealing that this characteristic holds for models ranging from 1B (small) to 13B (large) in size. Moreover, we demonstrate an interesting finding that the data hardness transfers across model sizes, and a smaller 350M model can effectively curate high-quality training data with hard samples for a larger 13B model, resulting in an equally or superior instruction-tuned model compared to training on the complete dataset. Utilizing open-sourced OPT and Llama-2 models up to 13B in size, two publicly available instruction-tuning training datasets and evaluated by both automatic metrics & humans, our paper introduces a novel approach to training data selection, showcasing a more efficient alternative.

Smaller Language Models are capable of selecting Instruction-Tuning Training Data for Larger Language Models

Large Language Models (LLM) have revolutionized Natural Language Processing (NLP), improving state-of-the-art and exhibiting emergent capabilities across various tasks. However, their application in extracting information from visually rich documents, which is at the core of many document processing workflows and involving the extraction of key entities from semi-structured documents, has not yet been successful. The main obstacles to adopting LLMs for this task include the absence of layout encoding within LLMs, which is critical for high quality extraction, and the lack of a grounding mechanism to localize the predicted entities within the document. In this paper, we introduce Language Model-based Document Information EXtraction and Localization (LMDX), a methodology to reframe the document information extraction task for a LLM. LMDX enables extraction of singular, repeated, and hierarchical entities, both with and without training data, while providing grounding guarantees and localizing the entities within the document. Finally, we apply LMDX to the PaLM 2-S and Gemini Pro LLMs and evaluate it on VRDU and CORD benchmarks, setting a new state-of-the-art and showing how LMDX enables the creation of high quality, data-efficient parsers.

LMDX: Language Model-based Document Information Extraction and Localization

Topic models have a rich history with various applications and have recently been reinvigorated by neural topic modeling. However, these numerous topic models adopt totally distinct datasets, implementations, and evaluations. This impedes quick utilization and fair comparisons, and thereby hinders their research progress and applications. To tackle this challenge, we in this paper propose a Topic Modeling System Toolkit (TopMost). Compared to existing toolkits, TopMost stands out by supporting more extensive features. It covers a broader spectrum of topic modeling scenarios with their complete lifecycles, including datasets, preprocessing, models, training, and evaluations. Thanks to its highly cohesive and decoupled modular design, TopMost enables rapid utilization, fair comparisons, and flexible extensions of diverse cutting-edge topic models. Our code, tutorials, and documentation are available at https://github.com/bobxwu/topmost.

Towards the TopMost: A Topic Modeling System Toolkit

The scaling of Large Language Models (LLMs) for retrieval-based tasks, particularly in Retrieval Augmented Generation (RAG), faces significant memory constraints, especially when fine-tuning extensive prompt sequences. Current open-source libraries support full-model inference and fine-tuning across multiple GPUs but fall short of accommodating the efficient parameter distribution required for retrieved context. Addressing this gap, we introduce a novel framework for PEFT-compatible fine-tuning of GPT models, leveraging distributed training. Our framework uniquely utilizes JAX's just-in-time (JIT) compilation and tensor-sharding for efficient resource management, thereby enabling accelerated fine-tuning with reduced memory requirements. This advancement significantly improves the scalability and feasibility of fine-tuning LLMs for complex RAG applications, even on systems with limited GPU resources. Our experiments show more than 12x improvement in runtime compared to Hugging Face/DeepSpeed implementation with four GPUs while consuming less than half the VRAM per GPU.

JORA: JAX Tensor-Parallel LoRA Library for Retrieval Augmented Fine-Tuning

Many computational analyses require linking information across noisy text datasets. While large language models (LLMs) offer significant promise, approximate string matching packages in popular statistical softwares such as R and Stata remain predominant in academic applications. These packages have simple interfaces and can be easily extended to a diversity of languages and settings, and for academic applications, ease-of-use and extensibility are essential. In contrast, packages for record linkage with LLMs require significant familiarity with deep learning frameworks and often focus on specialized applications of commercial value in English. The open-source package LinkTransformer aims to bridge this gap by providing an end-to-end software for performing record linkage and other data cleaning tasks with transformer LLMs, treating linkage as a text retrieval problem. At its core is an off-the-shelf toolkit for applying transformer models to record linkage. LinkTransformer contains a rich repository of pre-trained models for multiple languages and supports easy integration of any transformer language model from Hugging Face or OpenAI, providing the extensibility required for many scholarly applications. Its APIs also perform common data processing tasks, e.g., aggregation, noisy de-duplication, and translation-free cross-lingual linkage. LinkTransformer contains comprehensive tools for efficient model tuning, allowing for highly customized applications, and users can easily contribute their custom-trained models to its model hub to ensure reproducibility. Using a novel benchmark dataset geared towards academic applications, we show that LinkTransformer - with both custom models and Hugging Face or OpenAI models off-the-shelf - outperforms string matching by a wide margin. By combining transformer LMs with intuitive APIs, LinkTransformer aims to democratize these performance gains for those who lack familiarity with deep learning frameworks.

LinkTransformer: A Unified Package for Record Linkage with Transformer Language Models

Following the rapid progress in natural language processing (NLP) models, language models are applied to increasingly more complex interactive tasks such as negotiations and conversation moderations. 
Having human evaluators directly interact with these NLP models is essential for adequately evaluating the performance on such interactive tasks. 
We develop BotEval, an easily customizable, open-source, evaluation toolkit that focuses on enabling human-bot interactions as part of the evaluation process, as opposed to human evaluators making judgements for a static input. 
BotEval balances flexibility for customization and user-friendliness by providing templates for common use cases that span various degrees of complexity and built-in compatibility with popular crowdsourcing platforms.
We showcase the numerous useful features of BotEval through a study that evaluates the performance of various chatbots on their effectiveness for conversational moderation and discuss how BotEval differs from other annotation tools.

Premium content

Chain-of-Verification Reduces Hallucination in Large Language Models

Downloads

Next from ACL 2024

Efficient Detection of LLM-generated Texts with a Bayesian Surrogate Model

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES