United States

To evaluate knowledge in large language models (LLMs), current methods query the model and then evaluate its generated responses. In this work, we ask whether evaluation can be done $\textit{before}$ the model has generated any text. Concretely, is it possible to estimate how knowledgeable a model is about a certain entity, only from its internal computation? We study this question with two tasks: given a subject entity, the goal is to predict (a) the ability of the model to answer common questions about the entity, and (b) the factuality of open-ended responses generated by the model about the entity. Experiments with a variety of LLMs show that KEEN, a simple probe trained over internal subject representations, succeeds at both tasks --- correlating with both the QA accuracy of the model per-subject and FActScore, a recent factuality metric in open-ended generation. Moreover, KEEN naturally aligns with the model&#39;s hedging behavior and faithfully reflects changes in the model&#39;s knowledge after fine-tuning. Lastly, we show a more interpretable yet equally performant variant of KEEN, which highlights a small set of tokens indicative of clusters and gaps in the model&#39;s knowledge. Being simple and lightweight, KEEN can be leveraged to guide decisions such as when it is appropriate to apply further training or augment queries with retrieval.

EMNLP 2024

Estimating Knowledge in Large Language Models Without Generating a Single Token

formal language

autoformalization

mathematical reasoning

### 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!

ACL 2024

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.

To evaluate knowledge in large language models (LLMs), current methods query the model and then evaluate its generated responses. In this work, we ask whether evaluation can be done $\textit{before}$ the model has generated any text. Concretely, is it possible to estimate how knowledgeable a model is about a certain entity, only from its internal computation? We study this question with two tasks: given a subject entity, the goal is to predict (a) the ability of the model to answer common questions about the entity, and (b) the factuality of open-ended responses generated by the model about the entity. Experiments with a variety of LLMs show that KEEN, a simple probe trained over internal subject representations, succeeds at both tasks --- correlating with both the QA accuracy of the model per-subject and FActScore, a recent factuality metric in open-ended generation. Moreover, KEEN naturally aligns with the model's hedging behavior and faithfully reflects changes in the model's knowledge after fine-tuning. Lastly, we show a more interpretable yet equally performant variant of KEEN, which highlights a small set of tokens indicative of clusters and gaps in the model's knowledge. Being simple and lightweight, KEEN can be leveraged to guide decisions such as when it is appropriate to apply further training or augment queries with retrieval.

poster

## Welcome to EMNLP 2024! 
We are excited to welcome you to one of the most prominent conferences in the field of Natural Language Processing. This year, EMNLP 2024 is being held in a hybrid format,
offering both virtual and in-person participation in beautiful Miami. Due to a record-breaking number of submissions, we've expanded the total number of accepted papers to accommodate more cutting-edge research from around the globe.
### [Conference Handbook](https://drive.google.com/file/d/1WPROgxjLAC96AJL7Ugy0tEnYm7dkrbHt/view?usp=sharing)

You are required to register for this event. **Please register [here](https://2024.emnlp.org/registration/).** The EMNLP 2024 event page on Underline will be open to public one week prior to the event.

EMNLP 2024 will take place in Miami, Florida from Nov 12th to Nov 16th, 2024, at the Hyatt Regency Miami Hote and on Underline for remote participants.

Multilingual language models are widely used to extend NLP systems to low-resource languages. However, concrete evidence for the effects of multilinguality on language modeling performance in individual languages remains scarce. Here, we pre-train over 10,000 monolingual and multilingual language models for over 250 languages, including multiple language families that are under-studied in NLP. We assess how language modeling performance in each language varies as a function of (1) monolingual dataset size, (2) added multilingual dataset size, (3) linguistic similarity of the added languages, and (4) model size (up to 45M parameters). We find that in moderation, adding multilingual data improves low-resource language modeling performance, similar to increasing low-resource dataset sizes by up to 33%. Improvements depend on the syntactic similarity of the added multilingual data, with marginal additional effects of vocabulary overlap. However, high-resource languages consistently perform worse in multilingual pre-training scenarios. As dataset sizes increase, adding multilingual data begins to hurt performance for both low-resource and high-resource languages, likely due to limited model capacity (the "curse of multilinguality"). These results suggest that massively multilingual pre-training may not be optimal for any languages involved, but that more targeted models can significantly improve performance.

When Is Multilinguality a Curse? Language Modeling for 250 High- and Low-Resource Languages

Do LLMs have political leanings and are LLMs able to shift our political views? This paper explores these questions in the context of the 2024 U.S. presidential election. Through a voting simulation, we demonstrate 18 open-weight and closed-source LLMs' political preference for Biden over Trump. We show how Biden-leaning becomes more pronounced in instruction-tuned and reinforced models compared to their base versions by analyzing their responses to political questions related to the two nominees. We further explore the potential impact of LLMs on voter choice by recruiting 935 U.S. registered voters. Participants interacted with LLMs (Claude-3, Llama-3, and GPT-4) over five exchanges. Intriguingly, although LLMs were not asked to persuade users to support Biden, about 20% of Trump supporters reduced their support for Trump after LLM interaction. This result is noteworthy given that many studies on the persuasiveness of political campaigns have shown minimal effects in presidential elections. Many users also expressed a desire for further interaction with LLMs on political subjects. Further research on how LLMs affect users' political views is required, as their use becomes more widespread.

Hidden Persuaders: How LLM Political Bias Could Sway Our Elections

Large Language Models (LLMs) have highlighted the necessity of effective unlearning mechanisms to comply with data regulations and ethical AI practices. LLM unlearning aims at removing undesired data influences and associated model capabilities without compromising utility beyond the scope of unlearning. While interest in studying LLM unlearning is growing, the impact of the optimizer choice for LLM unlearning remains unexplored. In this work, we shed light on the significance of optimizer selection in LLM unlearning for the first time, establishing a clear connection between second-order optimization and influence unlearning (a classical approach using influence functions to update the model for data influence removal). This insight propels us to develop a second-order optimization-based LLM unlearning framework, termed Second-Order UnLearning (SOUL), which extends the static, one-shot model update using influence unlearning to a dynamic, iterative unlearning process. Our extensive experiments show that SOUL consistently outperforms conventional first-order methods across various unlearning tasks, models, and metrics, indicating that second-order optimization offers an effective and broadly applicable solution for LLM unlearning.

SOUL: Unlocking the Power of Second-Order Optimization for LLM Unlearning

FActScore has gained popularity as a metric to estimate the factuality of long-form texts generated by Large Language Models (LLMs) in English. However, there has not been any work in studying the behavior of FActScore in other languages. This paper studies the limitations of each component in the four-component pipeline of FActScore in the multilingual setting. We introduce a new dataset for FActScore on texts generated by strong multilingual LLMs. Our evaluation shows that LLMs exhibit distinct behaviors in both fact extraction and fact scoring tasks. No LLM produces consistent and reliable FActScore across languages of varying levels of resources. We also find that the knowledge source plays an important role in the quality of the estimated FActScore. Using Wikipedia as the knowledge source may hinder the true FActScore of long-form text due to its limited coverage in medium- and low-resource languages. We also incorporate 3 mitigations to our knowledge source that ultimately improve FActScore estimation across all languages.

An Analysis of Multilingual FActScore

Yoruba---an African language with roughly 47 million speakers---encompasses a continuum with several dialects. Recent efforts to develop NLP technologies for African languages have focused on their standard dialects, resulting in disparities for dialects and varieties for which there are little to no resources or tools. We take steps towards bridging this gap by introducing a new high-quality parallel text and speech corpus; YORULECT across three domains and four regional yoruba dialects. To develop this corpus, we engaged native speakers, traveling to communities where these dialects are spoken, to collect text and speech data. Using our newly created corpus, we conducted extensive experiments on (text) machine translation, automatic speech recognition, and speech-to-text translation. Our results reveal substantial performance disparities between standard yoruba and the other dialects across all tasks. However, we also show that with dialect-adaptive finetuning, we are able to narrow this gap. We believe our dataset and experimental analysis will contribute greatly to developing NLP tools for Yoruba and its dialects, and potentially for other African languages, by improving our understanding of existing challenges and offering a high-quality dataset for further development. We will release YORULECT dataset and models publicly under an open license.

Voices Unheard: NLP Resources and Models for Yorùbá Regional Dialects

Large Multimodal Models (LMMs) excel at comprehending human instructions and demonstrate remarkable results across a broad spectrum of tasks. Reinforcement Learning from Human Feedback (RLHF) and AI Feedback (RLAIF) further refine LLMs by aligning them with specific preferences. These methods primarily use ranking-based feedback for entire generations. With advanced AI models (Teacher), such as GPT-4 and Claude 3 Opus, we can request various types of detailed feedback that are expensive for humans to provide. We propose a two-stage algorithm ARES that Alternates REinforcement Learning (RL) and Supervised Fine-Tuning (SFT). First, we ask the Teacher to score how much each sentence contributes to solving the problem in a Chain-of-Thought (CoT). This sentence-level feedback allows us to consider individual valuable segments, providing more granular rewards for the RL procedure. Second, we ask the Teacher to correct wrong reasoning after the RL stage. The RL procedure requires substantial hyperparameter tuning and often generates errors such as repetitive words and incomplete sentences. With correction feedback, we stabilize the RL fine-tuned model through SFT. We conduct experiments on the multi-modal datasets ScienceQA and A-OKVQA to demonstrate the effectiveness of our proposal. The ARES rationale achieves around 70% win rate compared to baseline models judged by GPT-4o. Additionally, we observe that the improved rationale reasoning leads to a 2.5% increase in inference answer accuracy on average for the multi-modal datasets.

ARES: Alternating Reinforcement Learning and Supervised Fine-Tuning for Enhanced Multi-Modal Chain-of-Thought Reasoning Through Diverse AI Feedback

In this paper, we address the data scarcity problem in automatic data-driven glossing for low-resource languages by coordinating multiple sources of linguistic expertise. We enhance models by incorporating both token-level and sentence-level translations, utilizing the extensive linguistic capabilities of modern LLMs, and incorporating available dictionary resources. Our enhancements lead to an average absolute improvement of 5%-points in word-level accuracy over the previous state of the art on a typologically diverse dataset spanning six low-resource languages. The improvements are particularly noticeable for the lowest-resourced language Gitksan, where we achieve a 10%-point improvement. Furthermore, in a simulated ultra-low resource setting for the same six languages, training on fewer than 100 glossed sentences, we establish an average 10%-point improvement in word-level accuracy over the previous state-of-the-art system.

Multiple Sources are Better Than One: Incorporating External Knowledge in Low-Resource Glossing

In translation, a concept represented by a single word in a source language can have multiple variations in a target language. The task of lexical selection requires using context to identify which variation is most appropriate for a source text. We work with native speakers of nine languages to create DTAiLS, a dataset of 1,377 sentence pairs that exhibit cross-lingual concept variation when translating from English. We evaluate recent LLMs and neural machine translation systems on DTAiLS, with the best-performing model, GPT-4, achieving from 67 to 85\% accuracy across languages. Finally, we use language models to generate English rules describing target-language concept variations. Providing weaker models with high-quality lexical rules improves accuracy substantially, in some cases reaching or outperforming GPT-4.

Using Language Models to Disambiguate Lexical Choices in Translation

The dynamic nature of real-world information necessitates knowledge editing (KE) in large language models (LLMs). The edited knowledge should propagate and facilitate the deduction of new information based on existing model knowledge. We term the existing related knowledge in LLM serving as the origination of knowledge propagation as ''deduction anchors''. However, current KE approaches, which only operate on (subject, relation, object) triple. We both theoretically and empirically observe that this simplified setting often leads to uncertainty when determining the deduction anchors, causing low confidence in their answers. To mitigate this issue, we propose a novel task of event-based knowledge editing that pairs facts with event descriptions. This task manifests not only a closer simulation of real-world editing scenarios but also a more logically sound setting, implicitly defining the deduction anchor and enabling LLMs to propagate knowledge confidently. We curate a new benchmark dataset Evedit derived from the CounterFact dataset and validate its superiority in improving model confidence. Moreover, while we observe that the event-based setting is significantly challenging for existing approaches, we propose a novel approach Self-Edit that showcases stronger performance, achieving 55.6% consistency improvement while maintaining the naturalness of generation.

EVEDIT: Event-based Knowledge Editing for Deterministic Knowledge Propagation

Instruction tuning, or supervised finetuning on extensive task-specific data, is necessary for Large Vision-Language Models (LVLMs) to generalize well across a broad range of vision-language (VL) tasks. However, training on large VL datasets can become prohibitively expensive. In this work, we introduce COINCIDE, an effective and scalable data selection technique that uses a small model as a reference model to select visual instruction tuning data for efficient finetuning of a target LVLM, focusing on diversity and transferability. Specifically, we cluster the training data using internal activations from a small model, which identifies VL concept-skill compositions needed by a target LVLM. We then sample data from these diverse clusters by considering their density and transferability, or the ability to transfer well to other concept-skill compositions. This approach ensures the diversity of these compositions, which is vital for LVLM generalization. Extensive experiments demonstrate that COINCIDE achieves superior performance and data selection efficiency against 8 strong baselines on two distinct datasets: LLaVA-1.5 and Vision-Flan. Using only 20% of the LLaVA-1.5 dataset, COINCIDE achieves performance comparable to the LVLM finetuned on the whole dataset, with 70% reduction of the wall-clock running time. On the Vision-Flan dataset, our method achieves superior results with only 16.7% of the training data.

Downloads

Next from EMNLP 2024

When Is Multilinguality a Curse? Language Modeling for 250 High- and Low-Resource Languages

Similar lecture

Context versus Prior Knowledge in Language Models