Instruction-tuned LLMs are able to provide an explanation about their output to users by generating self-explanations that do not require gradient computations or the application of possibly complex XAI methods. In this paper, we analyse whether this ability results in a good explanation by evaluating self-explanations in the form of input rationales with respect to their plausibility to humans as well as their faithfulness to models. For this, we apply two text classification tasks: sentiment classification and forced labour detection. Next to English, we further include Danish and Italian translations of the sentiment classification task and compare self-explanations to human annotations for all samples. To allow for direct comparisons, we also compute post-hoc feature attribution, i.e., layer-wise relevance propagation (LRP) and apply this pipeline to 4 LLMs (Llama2, Llama3, Mistral and Mixtral). Our results show that self-explanations align more closely with human annotations compared to LRP, while maintaining a comparable level of faithfulness.

Recently introduced by some of the authors, the in-context identification paradigm aims at estimating, offline and based on synthetic data, a meta-model that describes the behavior of a whole class of systems. Once trained, this meta-model is fed with an observed input/output sequence (context) generated by a real system to predict its behavior in a zero-shot learning fashion. In this paper, we enhance the original meta-modeling framework through three key innovations: by formulating the learning task within a probabilistic framework; by managing non-contiguous context and query windows; and by adopting recurrent patching to effectively handle long context sequences. The efficacy of these modifications is demonstrated through a numerical example focusing on the Wiener-Hammerstein system class, highlighting the model's enhanced performance and scalability.

In this work, we show a fundamental limitation in vocabulary adaptation approaches that use Byte-Pair Encoding (BPE) tokenization scheme for fine-tuning pretrained language models (PLMs) to expert domains. Current approaches trivially append the target domain-specific vocabulary at the end of the PLM vocabulary. This approach leads to a lower priority score and causes sub-optimal tokenization in BPE that iteratively uses merge rules to tokenize a given text. To mitigate this issue, we propose AdaptBPE where the BPE tokenization initialization phase is modified to first perform the longest string matching on the added (target) vocabulary before tokenizing at the character level. We perform an extensive evaluation of AdaptBPE versus the standard BPE over various classification and summarization tasks; AdaptBPE improves by 3.57% (in terms of accuracy) and 1.87% (in terms of Rouge-L), respectively. AdaptBPE for MEDVOC works particularly well when reference summaries have high OOV concentration or are longer in length. We also conduct a human evaluation, revealing that AdaptBPE generates more relevant and more faithful summaries as compared to MEDVOC. We make our codebase publicly available at https://github.com/gb-kgp/adaptbpe.

Data annotation refers to the labeling or tagging of textual data with relevant information. A large body of works have reported positive results on leveraging LLMs as an alternative to human annotators. However, existing studies focus on classic NLP tasks, and the extent to which LLMs as data annotators perform in domains requiring expert knowledge remains underexplored. In this work, we investigate comprehensive approaches across three highly specialized domains and discuss practical suggestions from a cost-effectiveness perspective. To the best of our knowledge, we present the first systematic evaluation of LLMs as expert-level data annotators.

Large Language Models (LLMs) have shown impressive abilities in code generation, but they may generate erroneous programs. Reading a program takes ten times longer than writing it. Showing these erroneous programs to developers will waste developers' energies and introduce security risks to software. To address the above limitations, we propose HonestCoder, a novel LLM-based code generation approach. HonestCoder selectively shows the generated programs to developers based on LLMs' confidence. The confidence provides valuable insights into the correctness of generated programs. To achieve this goal, we propose a novel approach to estimate LLMs' confidence in code generation. It estimates confidence by measuring the multi-modal similarity between LLMs-generated programs. We collect and release a multilingual benchmark named TruthCodeBench, which consists of 2,265 samples and covers two popular programming languages (i.e., Python and Java). We apply HonestCoder to four popular LLMs (e.g., DeepSeek-Coder and Code Llama) and evaluate it on TruthCodeBench. Based on the experiments, we obtain the following insights. (1) HonestCoder can effectively estimate LLMs' confidence and accurately determine the correctness of generated programs. For example, HonestCoder outperforms the state-of-the-art baseline by 27.79% in AUROC and 63.74% in AUCPR. (2) HonestCoder can decrease the number of erroneous programs shown to developers. Compared to eight baselines, it can show more correct programs and fewer erroneous programs to developers. (3) Compared to showing code indiscriminately, HonestCoder only adds slight time overhead (approximately 0.4 seconds per requirement). (4) We discuss future directions to facilitate the application of LLMs in software development. We hope this work can motivate broad discussions about measuring the reliability of LLMs' outputs in performing code-related tasks.

The context window of large language models (LLMs) has been extended significantly in recent years. However, while the context length that the LLM can process has grown, the capability of the model to accurately reason over that context degrades noticeably. This occurs because modern LLMs often become overwhelmed by the vast amount of information in the context; when answering questions, the model must identify and reason over relevant evidence sparsely distributed throughout the text. To alleviate the challenge of long-context reasoning, we develop a retrieve-then-reason framework, enabling LLMs to reason over relevant evidence collected during an intermediate retrieval step. We find that modern LLMs struggle to accurately retrieve relevant facts and instead, often hallucinate "retrieved facts", resulting in flawed reasoning and the production of incorrect answers. Through extensive experiments on long-context QA benchmarks, we find our method to outperform competitive baselines by large margins, achieving at least 8.4 and 7.9 EM gains on the long-context versions of HotpotQA and SQuAD datasets, respectively. While these developments are promising, in our preliminary study, we show that the long-context performance of LLMs varied significantly across different tasks. We observe that, when tasked to generate answers by directly reasoning over the full context, performance degrades as the input context grows. In contrast, when tasked with retrieving the set of evidence relevant to the question, the performance of LLMs is only mildly affected by the growth of the input context.

Industrial machine fault diagnosis is a critical component of operational efficiency and safety in manufacturing environments. Traditional methods rely heavily on expert knowledge and specific machine learning models, which can be limited in their adaptability and require extensive labeled data. This paper introduces a novel approach leveraging Large Language Models (LLMs), specifically through a structured multi-round prompting technique, to improve fault diagnosis accuracy. By dynamically crafting prompts, our method enhances the model's ability to synthesize information from diverse data sources, leading to improved contextual understanding and actionable recommendations. Experimental results demonstrate that our approach outperforms baseline models, achieving an accuracy of 91% in diagnosing various fault types. The findings underscore the potential of LLMs in revolutionizing industrial fault consultation practices, paving the way for more effective maintenance strategies in complex environments.

In this paper, we study whether transformer-based language models can extract predicate argument structure from simple sentences. We firstly show that language models sometimes confuse which predicates apply to which objects. To mitigate this, we explore two tasks: question answering (Q/A), and first order logic (FOL) translation, and two regimes, prompting and finetuning. In FOL translation, we finetune several large language models on synthetic datasets designed to gauge their generalization abilities. For Q/A, we finetune encoder models like BERT and RoBERTa and use prompting for LLMs. The results show that FOL translation for LLMs is better suited to learn predicate argument structure.

Automatic question generation (QG) serves a wide range of purposes, such as augmenting question-answering (QA) corpora, enhancing chatbot systems, and developing educational materials. Despite its importance, most existing datasets predominantly focus on English, resulting in a considerable gap in data availability for other languages. Cross-lingual transfer for QG (XLT-QG) addresses this limitation by allowing models trained on high-resource language datasets to generate questions in low-resource languages. In this paper, we propose a simple and efficient XLT-QG method that operates without the need for monolingual, parallel, or labeled data in the target language, utilizing a small language model. Our model, trained solely on English QA datasets, learns interrogative structures from a limited set of question exemplars, which are then applied to generate questions in the target language. Experimental results show that our method outperforms several XLT-QG baselines and achieves performance comparable to GPT-3.5-turbo across different languages. Additionally, the synthetic data generated by our model proves beneficial for training multilingual QA models. With significantly fewer parameters than large language models and without requiring additional training for target languages, our approach offers an effective solution for QG and QA tasks across various languages.

Text-to-speech (TTS) systems that scale up the amount of training data have achieved significant improvements in zero-shot speech synthesis. However, these systems have certain limitations: they require a large amount of training data, which increases costs, and often overlook prosody similarity. To address these issues, we propose MultiVerse, a zero-shot multi-task TTS system that is able to perform TTS or speech style transfer in zero-shot and cross-lingual conditions. MultiVerse requires much less training data than traditional data-driven approaches. To ensure zero-shot performance even with limited data, we leverage source-filter theory-based disentanglement, utilizing the prompt for modeling filter-related and source-related representations. Additionally, to further enhance prosody similarity, we adopt a prosody modeling approach combining prompt-based autoregressive and non-autoregressive methods. Evaluations demonstrate the remarkable zero-shot multi-task TTS performance of MultiVerse and show that MultiVerse not only achieves zero-shot TTS performance comparable to data-driven TTS systems with much less data, but also significantly outperforms other zero-shot TTS systems trained with the same small amount of data. In particular, our novel prosody modeling technique significantly contributes to MultiVerse's ability to generate speech with high prosody similarity to the given prompts. Our samples are available at https://nc-ai.github.io/speech/publications/multiverse/index.html