Authors: Thomas P. Zollo, Todd Morrill, Zhun Deng, Jake C. Snell, Toniann Pitassi, Richard Zemel
The recent explosion in the capabilities of large language models has led to a wave of interest in how best to prompt a model to perform a given task. While it may be tempting to simply choose a prompt based on average performance on a validation set, this can lead to a deployment where unexpectedly poor responses are generated, especially for the worst-off users. To mitigate this prospect, we propose Prompt Risk Control, a lightweight framework for selecting a prompt based on rigorous upper bounds on families of informative risk measures. We offer methods for producing bounds on a diverse set of metrics, including quantities that measure worst-case responses and disparities in generation quality across the population of users. In addition, we extend the underlying statistical bounding techniques to accommodate the possibility of distribution shifts in deployment. Experiments on applications such as open-ended chat, medical question summarization, and code generation highlight how such a framework can foster responsible deployment by reducing the risk of the worst outcomes.
Authors: John X. Morris, Wenting Zhao, Justin T. Chiu, Vitaly Shmatikov, Alexander M. Rush
Language models produce a distribution over the next token; can we use this information to recover the prompt tokens? We consider the problem of language model inversion and show that next-token probabilities contain a surprising amount of information about the preceding text. Often we can recover the text in cases where it is hidden from the user, motivating a method for recovering unknown prompts given only the model's current distribution output. We consider a variety of model access scenarios, and show how even without predictions for every token in the vocabulary we can recover the probability vector through search. On Llama-2 7b, our inversion method reconstructs prompts with a BLEU of $59$ and token-level F1 of $78$ and recovers $27\%$ of prompts exactly. Code for reproducing all experiments is available at this http URL
Authors: Nathan Brown, Ashton Williamson, Tahj Anderson, Logan Lawrence
As pretrained transformer language models continue to achieve state-of-the-art performance, the Natural Language Processing community has pushed for advances in model compression and efficient attention mechanisms to address high computational requirements and limited input sequence length. Despite these separate efforts, no investigation has been done into the intersection of these two fields. In this work, we provide an evaluation of model compression via knowledge distillation on efficient attention transformers. We provide cost-performance trade-offs for the compression of state-of-the-art efficient attention architectures and the gains made in performance in comparison to their full attention counterparts. Furthermore, we introduce a new long-context Named Entity Recognition dataset, GONERD, to train and test the performance of NER models on long sequences. We find that distilled efficient attention transformers can preserve a significant amount of original model performance, preserving up to 98.6% across short-context tasks (GLUE, SQUAD, CoNLL-2003), up to 94.6% across long-context Question-and-Answering tasks (HotpotQA, TriviaQA), and up to 98.8% on long-context Named Entity Recognition (GONERD), while decreasing inference times by up to 57.8%. We find that, for most models on most tasks, performing knowledge distillation is an effective method to yield high-performing efficient attention models with low costs.
Authors: Stephanie L. Hyland, Shruthi Bannur, Kenza Bouzid, Daniel C. Castro, Mercy Ranjit, Anton Schwaighofer, Fernando Pérez-García, Valentina Salvatelli, Shaury Srivastav, Anja Thieme, Noel Codella, Matthew P. Lungren, Maria Teodora Wetscherek, Ozan Oktay, Javier Alvarez-Valle
We present a radiology-specific multimodal model for the task for generating radiological reports from chest X-rays (CXRs). Our work builds on the idea that large language model(s) can be equipped with multimodal capabilities through alignment with pre-trained vision encoders. On natural images, this has been shown to allow multimodal models to gain image understanding and description capabilities. Our proposed model (MAIRA-1) leverages a CXR-specific image encoder in conjunction with a fine-tuned large language model based on Vicuna-7B, and text-based data augmentation, to produce reports with state-of-the-art quality. In particular, MAIRA-1 significantly improves on the radiologist-aligned RadCliQ metric and across all lexical metrics considered. Manual review of model outputs demonstrates promising fluency and accuracy of generated reports while uncovering failure modes not captured by existing evaluation practices. More information and resources can be found on the project website: https://aka.ms/maira.
Authors: Weiwei Li, Xing Liu, Wei Wang, Lu Chen, Sizhe Li, Hui Fan
To address the challenge of identifying and understanding hidden dangers in substations from unstructured text data, a novel dynamic analysis method is proposed. This approach begins by analyzing and extracting data from the unstructured text related to hidden dangers. It then leverages a flexible, distributed data search engine built on Elastic-Search to handle this information. Following this, the hidden Markov model is employed to train the data within the engine. The Viterbi algorithm is integrated to decipher the hidden state sequences, facilitating the segmentation and labeling of entities related to hidden dangers. The final step involves using the Neo4j graph database to dynamically create a knowledge map that visualizes hidden dangers in the substation. This method's effectiveness is demonstrated through an example analysis using data from a specific substation's hidden dangers.
Authors: Shashank Gupta, Xuguang Ai, Ramakanth Kavuluru
End-to-end relation extraction (E2ERE) is an important and realistic application of natural language processing (NLP) in biomedicine. In this paper, we aim to compare three prevailing paradigms for E2ERE using a complex dataset focused on rare diseases involving discontinuous and nested entities. We use the RareDis information extraction dataset to evaluate three competing approaches (for E2ERE): NER $\rightarrow$ RE pipelines, joint sequence to sequence models, and generative pre-trained transformer (GPT) models. We use comparable state-of-the-art models and best practices for each of these approaches and conduct error analyses to assess their failure modes. Our findings reveal that pipeline models are still the best, while sequence-to-sequence models are not far behind; GPT models with eight times as many parameters are worse than even sequence-to-sequence models and lose to pipeline models by over 10 F1 points. Partial matches and discontinuous entities caused many NER errors contributing to lower overall E2E performances. We also verify these findings on a second E2ERE dataset for chemical-protein interactions. Although generative LM-based methods are more suitable for zero-shot settings, when training data is available, our results show that it is better to work with more conventional models trained and tailored for E2ERE. More innovative methods are needed to marry the best of the both worlds from smaller encoder-decoder pipeline models and the larger GPT models to improve E2ERE. As of now, we see that well designed pipeline models offer substantial performance gains at a lower cost and carbon footprint for E2ERE. Our contribution is also the first to conduct E2ERE for the RareDis dataset.
Authors: Zhichao Yang, Sanjit Singh Batra, Joel Stremmel, Eran Halperin
Recent advances in large language models (LLMs) show potential for clinical applications, such as clinical decision support and trial recommendations. However, the GPT-4 LLM predicts an excessive number of ICD codes for medical coding tasks, leading to high recall but low precision. To tackle this challenge, we introduce LLM-codex, a two-stage approach to predict ICD codes that first generates evidence proposals using an LLM and then employs an LSTM-based verification stage. The LSTM learns from both the LLM's high recall and human expert's high precision, using a custom loss function. Our model is the only approach that simultaneously achieves state-of-the-art results in medical coding accuracy, accuracy on rare codes, and sentence-level evidence identification to support coding decisions without training on human-annotated evidence according to experiments on the MIMIC dataset.
Authors: Milad Baghalzadeh Shishehgarkhaneh, Robert C. Moehler, Yihai Fang, Amer A. Hijazi, Hamed Aboutorab
The construction industry in Australia is characterized by its intricate supply chains and vulnerability to myriad risks. As such, effective supply chain risk management (SCRM) becomes imperative. This paper employs different transformer models, and train for Named Entity Recognition (NER) in the context of Australian construction SCRM. Utilizing NER, transformer models identify and classify specific risk-associated entities in news articles, offering a detailed insight into supply chain vulnerabilities. By analysing news articles through different transformer models, we can extract relevant entities and insights related to specific risk taxonomies local (milieu) to the Australian construction landscape. This research emphasises the potential of NLP-driven solutions, like transformer models, in revolutionising SCRM for construction in geo-media specific contexts.
Authors: Dongjun Jang, Sangah Lee, Sungjoo Byun, Jinwoong Kim, Jean Seo, Minseok Kim, Soyeon Kim, Chaeyoung Oh, Jaeyoon Kim, Hyemi Jo, Hyopil Shin
This paper presents the DaG LLM (David and Goliath Large Language Model), a language model specialized for Korean and fine-tuned through Instruction Tuning across 41 tasks within 13 distinct categories.
Authors: Madelon Hulsebos, Paul Groth, Çağatay Demiralp
Understanding the semantics of relational tables is instrumental for automation in data exploration and preparation systems. A key source for understanding a table is the semantics of its columns. With the rise of deep learning, learned table representations are now available, which can be applied for semantic type detection and achieve good performance on benchmarks. Nevertheless, we observe a gap between this performance and its applicability in practice. In this paper, we propose AdaTyper to address one of the most critical deployment challenges: adaptation. AdaTyper uses weak-supervision to adapt a hybrid type predictor towards new semantic types and shifted data distributions at inference time, using minimal human feedback. The hybrid type predictor of AdaTyper combines rule-based methods and a light machine learning model for semantic column type detection. We evaluate the adaptation performance of AdaTyper on real-world database tables hand-annotated with semantic column types through crowdsourcing and find that the f1-score improves for new and existing types. AdaTyper approaches an average precision of 0.6 after only seeing 5 examples, significantly outperforming existing adaptation methods based on human-provided regular expressions or dictionaries.
Authors: Saman Motamed, Danda Pani Paudel, Luc Van Gool
Diffusion models have revolutionized generative content creation and text-to-image (T2I) diffusion models in particular have increased the creative freedom of users by allowing scene synthesis using natural language. T2I models excel at synthesizing concepts such as nouns, appearances, and styles. To enable customized content creation based on a few example images of a concept, methods such as Textual Inversion and DreamBooth invert the desired concept and enable synthesizing it in new scenes. However, inverting more general concepts that go beyond object appearance and style (adjectives and verbs) through natural language, remains a challenge. Two key characteristics of these concepts contribute to the limitations of current inversion methods. 1) Adjectives and verbs are entangled with nouns (subject) and can hinder appearance-based inversion methods, where the subject appearance leaks into the concept embedding and 2) describing such concepts often extends beyond single word embeddings (being frozen in ice, walking on a tightrope, etc.) that current methods do not handle.
In this study, we introduce Lego, a textual inversion method designed to invert subject entangled concepts from a few example images. Lego disentangles concepts from their associated subjects using a simple yet effective Subject Separation step and employs a Context Loss that guides the inversion of single/multi-embedding concepts. In a thorough user study, Lego-generated concepts were preferred over 70% of the time when compared to the baseline. Additionally, visual question answering using a large language model suggested Lego-generated concepts are better aligned with the text description of the concept.
Authors: Jiahao Li, Quan Wang, Chiwei Zhu, Zhendong Mao, Yongdong Zhang
The task of Grammatical Error Correction (GEC) aims to automatically correct grammatical errors in natural texts. Almost all previous works treat annotated training data equally, but inherent discrepancies in data are neglected. In this paper, the inherent discrepancies are manifested in two aspects, namely, accuracy of data annotation and diversity of potential annotations. To this end, we propose MainGEC, which designs token-level and sentence-level training weights based on inherent discrepancies in accuracy and potential diversity of data annotation, respectively, and then conducts mixed-grained weighted training to improve the training effect for GEC. Empirical evaluation shows that whether in the Seq2Seq or Seq2Edit manner, MainGEC achieves consistent and significant performance improvements on two benchmark datasets, demonstrating the effectiveness and superiority of the mixed-grained weighted training. Further ablation experiments verify the effectiveness of designed weights of both granularities in MainGEC.
Authors: Neo Christopher Chung, George Dyer, Lennart Brocki
The global mental health crisis is looming with a rapid increase in mental disorders, limited resources, and the social stigma of seeking treatment. As the field of artificial intelligence (AI) has witnessed significant advancements in recent years, large language models (LLMs) capable of understanding and generating human-like text may be used in supporting or providing psychological counseling. However, the application of LLMs in the mental health domain raises concerns regarding the accuracy, effectiveness, and reliability of the information provided. This paper investigates the major challenges associated with the development of LLMs for psychological counseling, including model hallucination, interpretability, bias, privacy, and clinical effectiveness. We explore potential solutions to these challenges that are practical and applicable to the current paradigm of AI. From our experience in developing and deploying LLMs for mental health, AI holds a great promise for improving mental health care, if we can carefully navigate and overcome pitfalls of LLMs.
Authors: Muneeswaran I, Shreya Saxena, Siva Prasad, M V Sai Prakash, Advaith Shankar, Varun V, Vishal Vaddina, Saisubramaniam Gopalakrishnan
Large Language Models (LLMs) are widely used in critical fields such as healthcare, education, and finance due to their remarkable proficiency in various language-related tasks. However, LLMs are prone to generating factually incorrect responses or "hallucinations," which can lead to a loss of credibility and trust among users. To address this issue, we propose a multi-stage framework that generates the rationale first, verifies and refines incorrect ones, and uses them as supporting references to generate the answer. The generated rationale enhances the transparency of the answer and our framework provides insights into how the model arrived at this answer, by using this rationale and the references to the context. In this paper, we demonstrate its effectiveness in improving the quality of responses to drug-related inquiries in the life sciences industry. Our framework improves traditional Retrieval Augmented Generation (RAG) by enabling OpenAI GPT-3.5-turbo to be 14-25% more faithful and 16-22% more accurate on two datasets. Furthermore, fine-tuning samples based on our framework improves the accuracy of smaller open-access LLMs by 33-42% and competes with RAG on commercial models.
Authors: Bingkang Shi, Xiaodan Zhang, Dehan Kong, Yulei Wu, Zongzhen Liu, Honglei Lyu, Longtao Huang
The social biases and unwelcome stereotypes revealed by pretrained language models are becoming obstacles to their application. Compared to numerous debiasing methods targeting word level, there has been relatively less attention on biases present at phrase level, limiting the performance of debiasing in discipline domains. In this paper, we propose an automatic multi-token debiasing pipeline called \textbf{General Phrase Debiaser}, which is capable of mitigating phrase-level biases in masked language models. Specifically, our method consists of a \textit{phrase filter stage} that generates stereotypical phrases from Wikipedia pages as well as a \textit{model debias stage} that can debias models at the multi-token level to tackle bias challenges on phrases. The latter searches for prompts that trigger model's bias, and then uses them for debiasing. State-of-the-art results on standard datasets and metrics show that our approach can significantly reduce gender biases on both career and multiple disciplines, across models with varying parameter sizes.
Authors: John Mendonça, Patrícia Pereira, Miguel Menezes, Vera Cabarrão, Ana C. Farinha, Helena Moniz, João Paulo Carvalho, Alon Lavie, Isabel Trancoso
Task-oriented conversational datasets often lack topic variability and linguistic diversity. However, with the advent of Large Language Models (LLMs) pretrained on extensive, multilingual and diverse text data, these limitations seem overcome. Nevertheless, their generalisability to different languages and domains in dialogue applications remains uncertain without benchmarking datasets. This paper presents a holistic annotation approach for emotion and conversational quality in the context of bilingual customer support conversations. By performing annotations that take into consideration the complete instances that compose a conversation, one can form a broader perspective of the dialogue as a whole. Furthermore, it provides a unique and valuable resource for the development of text classification models. To this end, we present benchmarks for Emotion Recognition and Dialogue Quality Estimation and show that further research is needed to leverage these models in a production setting.
Authors: Jiří Bednář, Jakub Náplava, Petra Barančíková, Ondřej Lisický
This article focuses on the development and evaluation of Small-sized Czech sentence embedding models. Small models are important components for real-time industry applications in resource-constrained environments. Given the limited availability of labeled Czech data, alternative approaches, including pre-training, knowledge distillation, and unsupervised contrastive fine-tuning, are investigated. Comprehensive intrinsic and extrinsic analyses are conducted, showcasing the competitive performance of our models compared to significantly larger counterparts, with approximately 8 times smaller size and 5 times faster speed than conventional Base-sized models. To promote cooperation and reproducibility, both the models and the evaluation pipeline are made publicly accessible. Ultimately, this article presents practical applications of the developed sentence embedding models in Seznam.cz, the Czech search engine. These models have effectively replaced previous counterparts, enhancing the overall search experience for instance, in organic search, featured snippets, and image search. This transition has yielded improved performance.
Authors: Daryna Dementieva, Daniil Moskovskiy, David Dale, Alexander Panchenko
Text detoxification is the task of transferring the style of text from toxic to neutral. While here are approaches yielding promising results in monolingual setup, e.g., (Dale et al., 2021; Hallinan et al., 2022), cross-lingual transfer for this task remains a challenging open problem (Moskovskiy et al., 2022). In this work, we present a large-scale study of strategies for cross-lingual text detoxification -- given a parallel detoxification corpus for one language; the goal is to transfer detoxification ability to another language for which we do not have such a corpus. Moreover, we are the first to explore a new task where text translation and detoxification are performed simultaneously, providing several strong baselines for this task. Finally, we introduce new automatic detoxification evaluation metrics with higher correlations with human judgments than previous benchmarks. We assess the most promising approaches also with manual markup, determining the answer for the best strategy to transfer the knowledge of text detoxification between languages.
Authors: Wentao Ge, Shunian Chen, Guiming Chen, Junying Chen, Zhihong Chen, Shuo Yan, Chenghao Zhu, Ziyue Lin, Wenya Xie, Xidong Wang, Anningzhe Gao, Zhiyi Zhang, Jianquan Li, Xiang Wan, Benyou Wang
In the pursuit of Artificial General Intelligence (AGI), the integration of vision in language models has marked a significant milestone. The advent of vision-language models (MLLMs) like GPT-4V have expanded AI applications, aligning with the multi-modal capabilities of the human brain. However, evaluating the efficacy of MLLMs poses a substantial challenge due to the subjective nature of tasks that lack definitive answers. Existing automatic evaluation methodologies on multi-modal large language models rely on objective queries that have standard answers, inadequately addressing the nuances of creative and associative multi-modal tasks. To address this, we introduce MLLM-Bench, an innovative benchmark inspired by Vicuna, spanning a diverse array of scenarios, including Perception, Understanding, Applying, Analyzing, Evaluating, and Creation along with the ethical consideration. MLLM-Bench is designed to reflect user experience more accurately and provide a more holistic assessment of model performance. Comparative evaluations indicate a significant performance gap between existing open-source models and GPT-4V. We posit that MLLM-Bench will catalyze progress in the open-source community towards developing user-centric vision-language models that meet a broad spectrum of real-world applications. See online leaderboard in \url{https://mllm-bench.llmzoo.com}.
Authors: Yueqi Zeng, Ziqiang Li, Pengfei Xia, Lei Liu, Bin Li
With the boom in the natural language processing (NLP) field these years, backdoor attacks pose immense threats against deep neural network models. However, previous works hardly consider the effect of the poisoning rate. In this paper, our main objective is to reduce the number of poisoned samples while still achieving a satisfactory Attack Success Rate (ASR) in text backdoor attacks. To accomplish this, we propose an efficient trigger word insertion strategy in terms of trigger word optimization and poisoned sample selection. Extensive experiments on different datasets and models demonstrate that our proposed method can significantly improve attack effectiveness in text classification tasks. Remarkably, our approach achieves an ASR of over 90% with only 10 poisoned samples in the dirty-label setting and requires merely 1.5% of the training data in the clean-label setting.
Authors: Shulin Cao, Jiajie Zhang, Jiaxin Shi, Xin Lv, Zijun Yao, Qi Tian, Juanzi Li, Lei Hou
Large language models (LLMs) are capable of answering knowledge-intensive complex questions with chain-of-thought (CoT) reasoning. However, they tend to generate factually incorrect reasoning steps when the required knowledge is not available or up-to-date in models' parameters. Recent works turn to retrieving external knowledge to augment CoT reasoning. Despite being promising, these chain-based methods suffer from: 1) Negative retrieval. Unnecessary or incorrect retrieval may mislead the reasoning; 2) Limited sight. Lacking the ability to look backward or forward, a local error in one step will propagate along the chain.
In this paper, we propose a novel approach: Probabilistic Tree-of-thought Reasoning (ProbTree). First, LLMs translate a complex question into a query tree, in which each non-root node denotes a sub-question of its parent node. Then, probabilistic reasoning is conducted over the tree, by solving questions from leaf to root considering the confidence of both question decomposing and answering. During reasoning, for leaf nodes, LLMs choose a more confident answer from Closed-book QA that employs parametric knowledge and Open-book QA that employs retrieved external knowledge, thus eliminating the negative retrieval problem. For non-leaf nodes, with the hierarchical structure, LLMs have broader sights and are able to globally reason with the information from child nodes, thus recovering from local errors. The experiments on three Complex QA datasets under the open-domain setting show that our approach outperforms SOTA methods significantly, demonstrating the effect of probabilistic tree-of-thought reasoning.
Authors: Ondřej Cífka, Constantinos Dimitriou, Cheng-i Wang, Hendrik Schreiber, Luke Miner, Fabian-Robert Stöter
Current automatic lyrics transcription (ALT) benchmarks focus exclusively on word content and ignore the finer nuances of written lyrics including formatting and punctuation, which leads to a potential misalignment with the creative products of musicians and songwriters as well as listeners' experiences. For example, line breaks are important in conveying information about rhythm, emotional emphasis, rhyme, and high-level structure. To address this issue, we introduce Jam-ALT, a new lyrics transcription benchmark based on the JamendoLyrics dataset. Our contribution is twofold. Firstly, a complete revision of the transcripts, geared specifically towards ALT evaluation by following a newly created annotation guide that unifies the music industry's guidelines, covering aspects such as punctuation, line breaks, spelling, background vocals, and non-word sounds. Secondly, a suite of evaluation metrics designed, unlike the traditional word error rate, to capture such phenomena. We hope that the proposed benchmark contributes to the ALT task, enabling more precise and reliable assessments of transcription systems and enhancing the user experience in lyrics applications such as subtitle renderings for live captioning or karaoke.
Authors: Yasser Abdelaziz Dahou Djilali, Sanath Narayan, Eustache Le Bihan, Haithem Boussaid, Ebtessam Almazrouei, Merouane Debbah
The Lip Reading Sentences-3 (LRS3) benchmark has primarily been the focus of intense research in visual speech recognition (VSR) during the last few years. As a result, there is an increased risk of overfitting to its excessively used test set, which is only one hour duration. To alleviate this issue, we build a new VSR test set named WildVSR, by closely following the LRS3 dataset creation processes. We then evaluate and analyse the extent to which the current VSR models generalize to the new test data. We evaluate a broad range of publicly available VSR models and find significant drops in performance on our test set, compared to their corresponding LRS3 results. Our results suggest that the increase in word error rates is caused by the models inability to generalize to slightly harder and in the wild lip sequences than those found in the LRS3 test set. Our new test benchmark is made public in order to enable future research towards more robust VSR models.
Authors: Armand Stricker, Patrick Paroubek
As a recent development, task-oriented dialogues (TODs) have been enriched with chitchat in an effort to make dialogues more diverse and engaging. This enhancement is particularly valuable as TODs are often confined to narrow domains, making the mitigation of repetitive and predictable responses a significant challenge. This paper presents a comparative analysis of three chitchat enhancements, aiming to identify the most effective approach in terms of diversity. Additionally, we quantify the divergence between the added chitchat, the original task-oriented language, and chitchat typically found in chitchat datasets, highlighting the top 20 divergent keywords for each comparison. Our findings drive a discussion on future enhancements for augmenting TODs, emphasizing the importance of grounding dialogues beyond the task to achieve more diverse and natural exchanges.
Authors: Armand Stricker, Patrick Paroubek
Most existing dialogue corpora and models have been designed to fit into 2 predominant categories : task-oriented dialogues portray functional goals, such as making a restaurant reservation or booking a plane ticket, while chit-chat/open-domain dialogues focus on holding a socially engaging talk with a user. However, humans tend to seamlessly switch between modes and even use chitchat to enhance task-oriented conversations. To bridge this gap, new datasets have recently been created, blending both communication modes into conversation examples. The approaches used tend to rely on adding chit-chat snippets to pre-existing, human-generated task-oriented datasets. Given the tendencies observed in humans, we wonder however if the latter do not \textit{already} hold chit-chat sequences. By using topic modeling and searching for topics which are most similar to a set of keywords related to social talk, we explore the training sets of Schema-Guided Dialogues and MultiWOZ. Our study shows that sequences related to social talk are indeed naturally present, motivating further research on ways chitchat is combined into task-oriented dialogues.
Authors: Armand Stricker
Although general question answering has been well explored in recent years, temporal question answering is a task which has not received as much focus. Our work aims to leverage a popular approach used for general question answering, answer extraction, in order to find answers to temporal questions within a paragraph. To train our model, we propose a new dataset, inspired by SQuAD, specifically tailored to provide rich temporal information. We chose to adapt the corpus WikiWars, which contains several documents on history's greatest conflicts. Our evaluation shows that a deep learning model trained to perform pattern matching, often used in general question answering, can be adapted to temporal question answering, if we accept to ask questions whose answers must be directly present within a text.
Authors: Yan Tao, Olga Viberg, Ryan S. Baker, Rene F. Kizilcec
Culture fundamentally shapes people's reasoning, behavior, and communication. Generative artificial intelligence (AI) technologies may cause a shift towards a dominant culture. As people increasingly use AI to expedite and even automate various professional and personal tasks, cultural values embedded in AI models may bias authentic expression. We audit large language models for cultural bias, comparing their responses to nationally representative survey data, and evaluate country-specific prompting as a mitigation strategy. We find that GPT-4, 3.5 and 3 exhibit cultural values resembling English-speaking and Protestant European countries. Our mitigation strategy reduces cultural bias in recent models but not for all countries/territories. To avoid cultural bias in generative AI, especially in high-stakes contexts, we suggest using culture matching and ongoing cultural audits.
Authors: Vincent Dumoulin, Daniel D. Johnson, Pablo Samuel Castro, Hugo Larochelle, Yann Dauphin
Learning from human feedback (LHF) -- and in particular learning from pairwise preferences -- has recently become a crucial ingredient in training large language models (LLMs), and has been the subject of much research. Most recent works frame it as a reinforcement learning problem, where a reward function is learned from pairwise preference data and the LLM is treated as a policy which is adapted to maximize the rewards, often under additional regularization constraints. We propose an alternative interpretation which centers on the generative process for pairwise preferences and treats LHF as a density estimation problem. We provide theoretical and empirical results showing that for a family of generative processes defined via preference behavior distribution equations, training a reward function on pairwise preferences effectively models an annotator's implicit preference distribution. Finally, we discuss and present findings on "annotator misspecification" -- failure cases where wrong modeling assumptions are made about annotator behavior, resulting in poorly-adapted models -- suggesting that approaches that learn from pairwise human preferences could have trouble learning from a population of annotators with diverse viewpoints.
Authors: Wu Zekun, Sahan Bulathwela, Adriano Soares Koshiyama
Large Language Models (LLM) have made significant advances in the recent past becoming more mainstream in Artificial Intelligence (AI) enabled human-facing applications. However, LLMs often generate stereotypical output inherited from historical data, amplifying societal biases and raising ethical concerns. This work introduces i) the Multi-Grain Stereotype Dataset, which includes 52,751 instances of gender, race, profession and religion stereotypic text and ii) a novel stereotype classifier for English text. We design several experiments to rigorously test the proposed model trained on the novel dataset. Our experiments show that training the model in a multi-class setting can outperform the one-vs-all binary counterpart. Consistent feature importance signals from different eXplainable AI tools demonstrate that the new model exploits relevant text features. We utilise the newly created model to assess the stereotypic behaviour of the popular GPT family of models and observe the reduction of bias over time. In summary, our work establishes a robust and practical framework for auditing and evaluating the stereotypic bias in LLM.
Authors: Ramon Pires, Thales Sales Almeida, Hugo Abonizio, Rodrigo Nogueira
Recent advancements in language models have showcased human-comparable performance in academic entrance exams. However, existing studies often overlook questions that require the integration of visual comprehension, thus compromising the full spectrum and complexity inherent in real-world scenarios. To address this gap, we present a comprehensive framework to evaluate language models on entrance exams, which incorporates both textual and visual elements. We evaluate the two most recent editions of Exame Nacional do Ensino M\'edio (ENEM), the main standardized entrance examination adopted by Brazilian universities. Our study not only reaffirms the capabilities of GPT-4 as the state of the art for handling complex multidisciplinary questions, but also pioneers in offering a realistic assessment of multimodal language models on Portuguese examinations. One of the highlights is that text captions transcribing visual content outperform the direct use of images, suggesting that the vision model has room for improvement. Yet, despite improvements afforded by images or captions, mathematical questions remain a challenge for these state-of-the-art models. The code and data used on experiments are available at https://github.com/piresramon/gpt-4-enem.
Authors: Milan Straka, Jana Straková
We describe the winning submission to the CRAC 2022 Shared Task on Multilingual Coreference Resolution. Our system first solves mention detection and then coreference linking on the retrieved spans with an antecedent-maximization approach, and both tasks are fine-tuned jointly with shared Transformer weights. We report results of fine-tuning a wide range of pretrained models. The center of this contribution are fine-tuned multilingual models. We found one large multilingual model with sufficiently large encoder to increase performance on all datasets across the board, with the benefit not limited only to the underrepresented languages or groups of typologically relative languages. The source code is available at https://github.com/ufal/crac2022-corpipe.
Authors: Sebastian Koch, Stefan Pasch
This paper introduces supervised machine learning to the literature measuring corporate culture from text documents. We compile a unique data set of employee reviews that were labeled by human evaluators with respect to the information the reviews reveal about the firms' corporate culture. Using this data set, we fine-tune state-of-the-art transformer-based language models to perform the same classification task. In out-of-sample predictions, our language models classify 16 to 28 percent points more of employee reviews in line with human evaluators than traditional approaches of text classification. We make our models publicly available.
Authors: Steven H. Wang, Antoine Scardigli, Leonard Tang, Wei Chen, Dimitry Levkin, Anya Chen, Spencer Ball, Thomas Woodside, Oliver Zhang, Dan Hendrycks
Reading comprehension of legal text can be a particularly challenging task due to the length and complexity of legal clauses and a shortage of expert-annotated datasets. To address this challenge, we introduce the Merger Agreement Understanding Dataset (MAUD), an expert-annotated reading comprehension dataset based on the American Bar Association's 2021 Public Target Deal Points Study, with over 39,000 examples and over 47,000 total annotations. Our fine-tuned Transformer baselines show promising results, with models performing well above random on most questions. However, on a large subset of questions, there is still room for significant improvement. As the only expert-annotated merger agreement dataset, MAUD is valuable as a benchmark for both the legal profession and the NLP community.
Authors: Hugo Sousa, Alípio Jorge, Ricardo Campos
Temporal information extraction (TIE) has attracted a great deal of interest over the last two decades, leading to the development of a significant number of datasets. Despite its benefits, having access to a large volume of corpora makes it difficult when it comes to benchmark TIE systems. On the one hand, different datasets have different annotation schemes, thus hindering the comparison between competitors across different corpora. On the other hand, the fact that each corpus is commonly disseminated in a different format requires a considerable engineering effort for a researcher/practitioner to develop parsers for all of them. This constraint forces researchers to select a limited amount of datasets to evaluate their systems which consequently limits the comparability of the systems. Yet another obstacle that hinders the comparability of the TIE systems is the evaluation metric employed. While most research works adopt traditional metrics such as precision, recall, and $F_1$, a few others prefer temporal awareness -- a metric tailored to be more comprehensive on the evaluation of temporal systems. Although the reason for the absence of temporal awareness in the evaluation of most systems is not clear, one of the factors that certainly weights this decision is the necessity to implement the temporal closure algorithm in order to compute temporal awareness, which is not straightforward to implement neither is currently easily available. All in all, these problems have limited the fair comparison between approaches and consequently, the development of temporal extraction systems. To mitigate these problems, we have developed tieval, a Python library that provides a concise interface for importing different corpora and facilitates system evaluation. In this paper, we present the first public release of tieval and highlight its most relevant features.
Authors: Renze Lou, Kai Zhang, Wenpeng Yin
Task semantics can be expressed by a set of input-to-output examples or a piece of textual instruction. Conventional machine learning approaches for natural language processing (NLP) mainly rely on the availability of large-scale sets of task-specific examples. Two issues arise: first, collecting task-specific labeled examples does not apply to scenarios where tasks may be too complicated or costly to annotate, or the system is required to handle a new task immediately; second, this is not user-friendly since end-users are probably more willing to provide task description rather than a set of examples before using the system. Therefore, the community is paying increasing interest in a new supervision-seeking paradigm for NLP: learning from task instructions. Despite its impressive progress, there are some common issues that the community struggles with. This survey paper tries to summarize and provide insights into the current research on instruction learning, particularly by answering the following questions: (i) What is task instruction, and what instruction types exist? (ii) How to model instructions? (iii) What factors influence and explain the instructions' performance? (iv) What challenges remain in instruction learning? To our knowledge, this is the first comprehensive survey about textual instructions.
Authors: Wayne Xin Zhao, Kun Zhou, Junyi Li, Tianyi Tang, Xiaolei Wang, Yupeng Hou, Yingqian Min, Beichen Zhang, Junjie Zhang, Zican Dong, Yifan Du, Chen Yang, Yushuo Chen, Zhipeng Chen, Jinhao Jiang, Ruiyang Ren, Yifan Li, Xinyu Tang, Zikang Liu, Peiyu Liu, Jian-Yun Nie, Ji-Rong Wen
Language is essentially a complex, intricate system of human expressions governed by grammatical rules. It poses a significant challenge to develop capable AI algorithms for comprehending and grasping a language. As a major approach, language modeling has been widely studied for language understanding and generation in the past two decades, evolving from statistical language models to neural language models. Recently, pre-trained language models (PLMs) have been proposed by pre-training Transformer models over large-scale corpora, showing strong capabilities in solving various NLP tasks. Since researchers have found that model scaling can lead to performance improvement, they further study the scaling effect by increasing the model size to an even larger size. Interestingly, when the parameter scale exceeds a certain level, these enlarged language models not only achieve a significant performance improvement but also show some special abilities that are not present in small-scale language models. To discriminate the difference in parameter scale, the research community has coined the term large language models (LLM) for the PLMs of significant size. Recently, the research on LLMs has been largely advanced by both academia and industry, and a remarkable progress is the launch of ChatGPT, which has attracted widespread attention from society. The technical evolution of LLMs has been making an important impact on the entire AI community, which would revolutionize the way how we develop and use AI algorithms. In this survey, we review the recent advances of LLMs by introducing the background, key findings, and mainstream techniques. In particular, we focus on four major aspects of LLMs, namely pre-training, adaptation tuning, utilization, and capacity evaluation. Besides, we also summarize the available resources for developing LLMs and discuss the remaining issues for future directions.
Authors: Tianle Wang, Zihan Wang, Weitang Liu, Jingbo Shang
State-of-the-art weakly supervised text classification methods, while significantly reduced the required human supervision, still requires the supervision to cover all the classes of interest. This is never easy to meet in practice when human explore new, large corpora without complete pictures. In this paper, we work on a novel yet important problem of weakly supervised open-world text classification, where supervision is only needed for a few examples from a few known classes and the machine should handle both known and unknown classes in test time. General open-world classification has been studied mostly using image classification; however, existing methods typically assume the availability of sufficient known-class supervision and strong unknown-class prior knowledge (e.g., the number and/or data distribution). We propose a novel framework WOT-Class that lifts those strong assumptions. Specifically, it follows an iterative process of (a) clustering text to new classes, (b) mining and ranking indicative words for each class, and (c) merging redundant classes by using the overlapped indicative words as a bridge. Extensive experiments on 7 popular text classification datasets demonstrate that WOT-Class outperforms strong baselines consistently with a large margin, attaining 23.33% greater average absolute macro-F1 over existing approaches across all datasets. Such competent accuracy illuminates the practical potential of further reducing human effort for text classification.
Authors: Shahar Katz, Yonatan Belinkov
Recent advances in interpretability suggest we can project weights and hidden states of transformer-based language models (LMs) to their vocabulary, a transformation that makes them more human interpretable. In this paper, we investigate LM attention heads and memory values, the vectors the models dynamically create and recall while processing a given input. By analyzing the tokens they represent through this projection, we identify patterns in the information flow inside the attention mechanism. Based on our discoveries, we create a tool to visualize a forward pass of Generative Pre-trained Transformers (GPTs) as an interactive flow graph, with nodes representing neurons or hidden states and edges representing the interactions between them. Our visualization simplifies huge amounts of data into easy-to-read plots that can reflect the models' internal processing, uncovering the contribution of each component to the models' final prediction. Our visualization also unveils new insights about the role of layer norms as semantic filters that influence the models' output, and about neurons that are always activated during forward passes and act as regularization vectors.
Authors: Kranti Chalamalasetti, Jana Götze, Sherzod Hakimov, Brielen Madureira, Philipp Sadler, David Schlangen
Recent work has proposed a methodology for the systematic evaluation of "Situated Language Understanding Agents"-agents that operate in rich linguistic and non-linguistic contexts-through testing them in carefully constructed interactive settings. Other recent work has argued that Large Language Models (LLMs), if suitably set up, can be understood as (simulators of) such agents. A connection suggests itself, which this paper explores: Can LLMs be evaluated meaningfully by exposing them to constrained game-like settings that are built to challenge specific capabilities? As a proof of concept, this paper investigates five interaction settings, showing that current chat-optimised LLMs are, to an extent, capable to follow game-play instructions. Both this capability and the quality of the game play, measured by how well the objectives of the different games are met, follows the development cycle, with newer models performing better. The metrics even for the comparatively simple example games are far from being saturated, suggesting that the proposed instrument will remain to have diagnostic value. Our general framework for implementing and evaluating games with LLMs is available at https://github.com/clembench .
Authors: Dachuan Shi, Chaofan Tao, Anyi Rao, Zhendong Yang, Chun Yuan, Jiaqi Wang
Recent vision-language models have achieved tremendous progress far beyond what we ever expected. However, their computational costs are also dramatically growing with rapid development, especially for the large models. It makes model acceleration exceedingly critical in a scenario of limited resources. Although extensively studied for unimodal models, the acceleration for multimodal models, especially the vision-language Transformers, is relatively under-explored. To pursue more efficient and accessible vision-language Transformers, this paper introduces \textbf{Cross}-\textbf{G}uided \textbf{E}nsemble of \textbf{T}okens (\textbf{\emph{CrossGET}}), a universal acceleration framework for vision-language Transformers. This framework adaptively combines tokens through real-time, cross-modal guidance, thereby achieving substantial acceleration while keeping high performance. \textit{CrossGET} has two key innovations: 1) \textit{Cross-Guided Matching and Ensemble}. \textit{CrossGET} incorporates cross-modal guided token matching and ensemble to exploit cross-modal information effectively, only introducing cross-modal tokens with negligible extra parameters. 2) \textit{Complete-Graph Soft Matching}. In contrast to the existing bipartite soft matching approach, \textit{CrossGET} introduces a complete-graph soft matching policy to achieve more reliable token-matching results while maintaining parallelizability and high efficiency. Extensive experiments are conducted on various vision-language tasks, including image-text retrieval, visual reasoning, image captioning, and visual question answering. Performance on both classic multimodal architectures and emerging multimodal LLMs demonstrate the effectiveness and versatility of the proposed \textit{CrossGET} framework. The code will be at \url{https://github.com/sdc17/CrossGET}.
Authors: Yu Pan, Yanni Hu, Yuguang Yang, Wen Fei, Jixun Yao, Heng Lu, Lei Ma, Jianjun Zhao
Contrastive cross-modality pretraining has recently exhibited impressive success in diverse fields, whereas there is limited research on their merits in speech emotion recognition (SER). In this paper, we propose GEmo-CLAP, a kind of gender-attribute-enhanced contrastive language-audio pretraining (CLAP) method for SER. Specifically, we first construct an effective emotion CLAP (Emo-CLAP) for SER, using pre-trained text and audio encoders. Second, given the significance of gender information in SER, two novel multi-task learning based GEmo-CLAP (ML-GEmo-CLAP) and soft label based GEmo-CLAP (SL-GEmo-CLAP) models are further proposed to incorporate gender information of speech signals, forming more reasonable objectives. Experiments on IEMOCAP indicate that our proposed two GEmo-CLAPs consistently outperform Emo-CLAP with different pre-trained models. Remarkably, the proposed WavLM-based SL-GEmo-CLAP obtains the best UAR of 81.43\% and WAR of 83.16\%, which performs better than state-of-the-art SER methods.
Authors: Humza Naveed, Asad Ullah Khan, Shi Qiu, Muhammad Saqib, Saeed Anwar, Muhammad Usman, Naveed Akhtar, Nick Barnes, Ajmal Mian
Large Language Models (LLMs) have recently demonstrated remarkable capabilities in natural language processing tasks and beyond. This success of LLMs has led to a large influx of research contributions in this direction. These works encompass diverse topics such as architectural innovations, better training strategies, context length improvements, fine-tuning, multi-modal LLMs, robotics, datasets, benchmarking, efficiency, and more. With the rapid development of techniques and regular breakthroughs in LLM research, it has become considerably challenging to perceive the bigger picture of the advances in this direction. Considering the rapidly emerging plethora of literature on LLMs, it is imperative that the research community is able to benefit from a concise yet comprehensive overview of the recent developments in this field. This article provides an overview of the existing literature on a broad range of LLM-related concepts. Our self-contained comprehensive overview of LLMs discusses relevant background concepts along with covering the advanced topics at the frontier of research in LLMs. This review article is intended to not only provide a systematic survey but also a quick comprehensive reference for the researchers and practitioners to draw insights from extensive informative summaries of the existing works to advance the LLM research.
Authors: Jiashuo Sun, Chengjin Xu, Lumingyuan Tang, Saizhuo Wang, Chen Lin, Yeyun Gong, Lionel M. Ni, Heung-Yeung Shum, Jian Guo
Although large language models (LLMs) have achieved significant success in various tasks, they often struggle with hallucination problems, especially in scenarios requiring deep and responsible reasoning. These issues could be partially addressed by introducing external knowledge graphs (KG) in LLM reasoning. In this paper, we propose a new LLM-KG integrating paradigm ``$\hbox{LLM}\otimes\hbox{KG}$'' which treats the LLM as an agent to interactively explore related entities and relations on KGs and perform reasoning based on the retrieved knowledge. We further implement this paradigm by introducing a new approach called Think-on-Graph (ToG), in which the LLM agent iteratively executes beam search on KG, discovers the most promising reasoning paths, and returns the most likely reasoning results. We use a number of well-designed experiments to examine and illustrate the following advantages of ToG: 1) compared with LLMs, ToG has better deep reasoning power; 2) ToG has the ability of knowledge traceability and knowledge correctability by leveraging LLMs reasoning and expert feedback; 3) ToG provides a flexible plug-and-play framework for different LLMs, KGs and prompting strategies without any additional training cost; 4) the performance of ToG with small LLM models could exceed large LLM such as GPT-4 in certain scenarios and this reduces the cost of LLM deployment and application. As a training-free method with lower computational cost and better generality, ToG achieves overall SOTA in 6 out of 9 datasets where most previous SOTAs rely on additional training.
Authors: Maciej Besta, Nils Blach, Ales Kubicek, Robert Gerstenberger, Lukas Gianinazzi, Joanna Gajda, Tomasz Lehmann, Michal Podstawski, Hubert Niewiadomski, Piotr Nyczyk, Torsten Hoefler
We introduce Graph of Thoughts (GoT): a framework that advances prompting capabilities in large language models (LLMs) beyond those offered by paradigms such as Chain-of-Thought or Tree of Thoughts (ToT). The key idea and primary advantage of GoT is the ability to model the information generated by an LLM as an arbitrary graph, where units of information ("LLM thoughts") are vertices, and edges correspond to dependencies between these vertices. This approach enables combining arbitrary LLM thoughts into synergistic outcomes, distilling the essence of whole networks of thoughts, or enhancing thoughts using feedback loops. We illustrate that GoT offers advantages over state of the art on different tasks, for example increasing the quality of sorting by 62% over ToT, while simultaneously reducing costs by >31%. We ensure that GoT is extensible with new thought transformations and thus can be used to spearhead new prompting schemes. This work brings the LLM reasoning closer to human thinking or brain mechanisms such as recurrence, both of which form complex networks.
Authors: Shanglin Lei, Guanting Dong, Xiaoping Wang, Keheng Wang, Sirui Wang
The development of emotion recognition in dialogue (ERC) has been consistently hindered by the complexity of pipeline designs, leading to ERC models that often overfit to specific datasets and dialogue patterns. In this study, we propose a novel approach, namely
InstructERC, to reformulates the ERC task from a discriminative framework to a generative framework based on Large Language Models (LLMs) . InstructERC has two significant contributions: Firstly, InstructERC introduces a simple yet effective retrieval template module, which helps the model explicitly integrate multi-granularity dialogue supervision information by concatenating the historical dialog content, label statement, and emotional domain demonstrations with high semantic similarity. Furthermore, we introduce two additional emotion alignment tasks, namely speaker identification and emotion prediction tasks, to implicitly model the dialogue role relationships and future emotional tendencies in conversations. Our LLM-based plug-and-play plugin framework significantly outperforms all previous models and achieves comprehensive SOTA on three commonly used ERC datasets. Extensive analysis of parameter-efficient and data-scaling experiments provide empirical guidance for applying InstructERC in practical scenarios. Our code will be released after blind review.
Authors: Tudor Groza, Harry Caufield, Dylan Gration, Gareth Baynam, Melissa A Haendel, Peter N Robinson, Christopher J Mungall, Justin T Reese
Objective: Clinical deep phenotyping and phenotype annotation play a critical role in both the diagnosis of patients with rare disorders as well as in building computationally-tractable knowledge in the rare disorders field. These processes rely on using ontology concepts, often from the Human Phenotype Ontology, in conjunction with a phenotype concept recognition task (supported usually by machine learning methods) to curate patient profiles or existing scientific literature. With the significant shift in the use of large language models (LLMs) for most NLP tasks, we examine the performance of the latest Generative Pre-trained Transformer (GPT) models underpinning ChatGPT as a foundation for the tasks of clinical phenotyping and phenotype annotation. Materials and Methods: The experimental setup of the study included seven prompts of various levels of specificity, two GPT models (gpt-3.5-turbo and gpt-4.0) and two established gold standard corpora for phenotype recognition, one consisting of publication abstracts and the other clinical observations. Results: Our results show that, with an appropriate setup, these models can achieve state of the art performance. The best run, using few-shot learning, achieved 0.58 macro F1 score on publication abstracts and 0.75 macro F1 score on clinical observations, the former being comparable with the state of the art, while the latter surpassing the current best in class tool. Conclusion: While the results are promising, the non-deterministic nature of the outcomes, the high cost and the lack of concordance between different runs using the same prompt and input make the use of these LLMs challenging for this particular task.
Authors: Lauren Hong, Ting Wang
Parameter-efficient fine-tuning (PEFT) enables efficient adaptation of pre-trained language models (PLMs) to specific tasks. By tuning only a minimal set of (extra) parameters, PEFT achieves performance comparable to full fine-tuning. However, despite its prevalent use, the security implications of PEFT remain largely unexplored. In this paper, we conduct a pilot study revealing that PEFT exhibits unique vulnerability to trojan attacks. Specifically, we present PETA, a novel attack that accounts for downstream adaptation through bilevel optimization: the upper-level objective embeds the backdoor into a PLM while the lower-level objective simulates PEFT to retain the PLM's task-specific performance. With extensive evaluation across a variety of downstream tasks and trigger designs, we demonstrate PETA's effectiveness in terms of both attack success rate and unaffected clean accuracy, even after the victim user performs PEFT over the backdoored PLM using untainted data. Moreover, we empirically provide possible explanations for PETA's efficacy: the bilevel optimization inherently 'orthogonalizes' the backdoor and PEFT modules, thereby retaining the backdoor throughout PEFT. Based on this insight, we explore a simple defense that omits PEFT in selected layers of the backdoored PLM and unfreezes a subset of these layers' parameters, which is shown to effectively neutralize PETA.
Authors: Changhun Lee, Chiehyeon Lim
We study the theoretical aspects of Reinforced Language Models (RLMs) from a bi-objective optimization perspective. Specifically, we consider the RLMs as a Pareto optimization problem that maximizes the two conflicting objectives, i.e., reward objective and likelihood objectives, simultaneously. Our main contribution consists of three parts. First, we establish the theoretical foundations of RLM as a Pareto optimization problem by presenting Reward Upper BOund (RUBO) and Pareto optimality. Our theoretical outcomes are supported by not only deductive proofs but also empirical results. Second, we propose Reward Dropout, a simple yet powerful method that guarantees to improve a bi-objective optimization of RLM. Lastly, we demonstrate that the Reward Dropout is consistently effective across five benchmark datasets and four benchmark LLMs, meaning that the Reward Dropout significantly improves the optimization performance of RLMs.
Authors: Guozheng Li, Peng Wang, Wenjun Ke
Relation extraction (RE) consistently involves a certain degree of labeled or unlabeled data even if under zero-shot setting. Recent studies have shown that large language models (LLMs) transfer well to new tasks out-of-the-box simply given a natural language prompt, which provides the possibility of extracting relations from text without any data and parameter tuning. This work focuses on the study of exploring LLMs, such as ChatGPT, as zero-shot relation extractors. On the one hand, we analyze the drawbacks of existing RE prompts and attempt to incorporate recent prompt techniques such as chain-of-thought (CoT) to improve zero-shot RE. We propose the summarize-and-ask (\textsc{SumAsk}) prompting, a simple prompt recursively using LLMs to transform RE inputs to the effective question answering (QA) format. On the other hand, we conduct comprehensive experiments on various benchmarks and settings to investigate the capabilities of LLMs on zero-shot RE. Specifically, we have the following findings: (i) \textsc{SumAsk} consistently and significantly improves LLMs performance on different model sizes, benchmarks and settings; (ii) Zero-shot prompting with ChatGPT achieves competitive or superior results compared with zero-shot and fully supervised methods; (iii) LLMs deliver promising performance in extracting overlapping relations; (iv) The performance varies greatly regarding different relations. Different from small language models, LLMs are effective in handling challenge none-of-the-above (NoTA) relation.
Authors: Hao Liu, Matei Zaharia, Pieter Abbeel
Training large transformers using next-token prediction has given rise to groundbreaking advancements in AI. While this generative AI approach has produced impressive results, it heavily leans on human supervision. Even state-of-the-art AI models like ChatGPT depend on fine-tuning through human demonstrations, demanding extensive human input and domain expertise. This strong reliance on human oversight poses a significant hurdle to the advancement of AI innovation. To address this limitation, we propose a novel paradigm termed Exploratory AI (EAI) aimed at autonomously generating high-quality training data. Drawing inspiration from unsupervised reinforcement learning (RL) pretraining, EAI achieves exploration within the natural language space. We accomplish this by harnessing large language models to assess the novelty of generated content. Our approach employs two key components: an actor that generates novel content following exploration principles and a critic that evaluates the generated content, offering critiques to guide the actor. Empirical evaluations demonstrate that EAI significantly boosts model performance on complex reasoning tasks, addressing the limitations of human-intensive supervision.
Authors: Andre Ye, Sebastin Santy, Jena D. Hwang, Amy X. Zhang, Ranjay Krishna
Computer vision often treats perception as objective, and this assumption gets reflected in the way that datasets are collected and models are trained. For instance, image descriptions in different languages are typically assumed to be translations of the same semantic content. However, work in cross-cultural psychology and linguistics has shown that individuals differ in their visual perception depending on their cultural background and the language they speak. In this paper, we demonstrate significant differences in semantic content across languages in both dataset and model-produced captions. When data is multilingual as opposed to monolingual, captions have higher semantic coverage on average, as measured by scene graph, embedding, and linguistic complexity. For example, multilingual captions have on average 21.8% more objects, 24.5% more relations, and 27.1% more attributes than a set of monolingual captions. Moreover, models trained on content from different languages perform best against test data from those languages, while those trained on multilingual content perform consistently well across all evaluation data compositions. Our research provides implications for how diverse modes of perception can improve image understanding.
Authors: Yonatan Oren, Nicole Meister, Niladri Chatterji, Faisal Ladhak, Tatsunori B. Hashimoto
Large language models are trained on vast amounts of internet data, prompting concerns and speculation that they have memorized public benchmarks. Going from speculation to proof of contamination is challenging, as the pretraining data used by proprietary models are often not publicly accessible. We show that it is possible to provide provable guarantees of test set contamination in language models without access to pretraining data or model weights. Our approach leverages the fact that when there is no data contamination, all orderings of an exchangeable benchmark should be equally likely. In contrast, the tendency for language models to memorize example order means that a contaminated language model will find certain canonical orderings to be much more likely than others. Our test flags potential contamination whenever the likelihood of a canonically ordered benchmark dataset is significantly higher than the likelihood after shuffling the examples. We demonstrate that our procedure is sensitive enough to reliably prove test set contamination in challenging situations, including models as small as 1.4 billion parameters, on small test sets of only 1000 examples, and datasets that appear only a few times in the pretraining corpus. Using our test, we audit five popular publicly accessible language models for test set contamination and find little evidence for pervasive contamination.
Authors: Xiaoyu Tian, Liangyu Chen, Na Liu, Yaxuan Liu, Wei Zou, Kaijiang Chen, Ming Cui
Inspired by the dual-process theory of human cognition, we introduce DUMA, a novel conversational agent framework that embodies a dual-mind mechanism through the utilization of two generative Large Language Models (LLMs) dedicated to fast and slow thinking respectively. The fast thinking model serves as the primary interface for external interactions and initial response generation, evaluating the necessity for engaging the slow thinking model based on the complexity of the complete response. When invoked, the slow thinking model takes over the conversation, engaging in meticulous planning, reasoning, and tool utilization to provide a well-analyzed response. This dual-mind configuration allows for a seamless transition between intuitive responses and deliberate problem-solving processes based on the situation. We have constructed a conversational agent to handle online inquiries in the real estate industry. The experiment proves that our method balances effectiveness and efficiency, and has a significant improvement compared to the baseline.
Authors: Veronica Chatrath, Oluwanifemi Bamgbose, Shaina Raza
As the use of large language models (LLMs) increases within society, as does the risk of their misuse. Appropriate safeguards must be in place to ensure LLM outputs uphold the ethical standards of society, highlighting the positive role that artificial intelligence technologies can have. Recent events indicate ethical concerns around conventionally trained LLMs, leading to overall unsafe user experiences. This motivates our research question: how do we ensure LLM alignment? In this work, we introduce a test suite of unique prompts to foster the development of aligned LLMs that are fair, safe, and robust. We show that prompting LLMs at every step of the development pipeline, including data curation, pre-training, and fine-tuning, will result in an overall more responsible model. Our test suite evaluates outputs from four state-of-the-art language models: GPT-3.5, GPT-4, OPT, and LLaMA-2. The assessment presented in this paper highlights a gap between societal alignment and the capabilities of current LLMs. Additionally, implementing a test suite such as ours lowers the environmental overhead of making models safe and fair.
Authors: Antonin Sulc, Raimund Kammering, Annika Eichler, Tim Wilksen
Navigating the landscape of particle accelerators has become increasingly challenging with recent surges in contributions. These intricate devices challenge comprehension, even within individual facilities. To address this, we introduce PACuna, a fine-tuned language model refined through publicly available accelerator resources like conferences, pre-prints, and books. We automated data collection and question generation to minimize expert involvement and make the data publicly available. PACuna demonstrates proficiency in addressing intricate accelerator questions, validated by experts. Our approach shows adapting language models to scientific domains by fine-tuning technical texts and auto-generated corpora capturing the latest developments can further produce pre-trained models to answer some intricate questions that commercially available assistants cannot and can serve as intelligent assistants for individual facilities.
Authors: Rusheb Shah, Quentin Feuillade--Montixi, Soroush Pour, Arush Tagade, Stephen Casper, Javier Rando
Despite efforts to align large language models to produce harmless responses, they are still vulnerable to jailbreak prompts that elicit unrestricted behaviour. In this work, we investigate persona modulation as a black-box jailbreaking method to steer a target model to take on personalities that are willing to comply with harmful instructions. Rather than manually crafting prompts for each persona, we automate the generation of jailbreaks using a language model assistant. We demonstrate a range of harmful completions made possible by persona modulation, including detailed instructions for synthesising methamphetamine, building a bomb, and laundering money. These automated attacks achieve a harmful completion rate of 42.5% in GPT-4, which is 185 times larger than before modulation (0.23%). These prompts also transfer to Claude 2 and Vicuna with harmful completion rates of 61.0% and 35.9%, respectively. Our work reveals yet another vulnerability in commercial large language models and highlights the need for more comprehensive safeguards.
Authors: Yuxiang Zhou, Yulan He
Adjusting for latent covariates is crucial for estimating causal effects from observational textual data. Most existing methods only account for confounding covariates that affect both treatment and outcome, potentially leading to biased causal effects. This bias arises from insufficient consideration of non-confounding covariates, which are relevant only to either the treatment or the outcome. In this work, we aim to mitigate the bias by unveiling interactions between different variables to disentangle the non-confounding covariates when estimating causal effects from text. The disentangling process ensures covariates only contribute to their respective objectives, enabling independence between variables. Additionally, we impose a constraint to balance representations from the treatment group and control group to alleviate selection bias. We conduct experiments on two different treatment factors under various scenarios, and the proposed model significantly outperforms recent strong baselines. Furthermore, our thorough analysis on earnings call transcripts demonstrates that our model can effectively disentangle the variables, and further investigations into real-world scenarios provide guidance for investors to make informed decisions.
Authors: Yuanhe Tian, Ruyi Gan, Yan Song, Jiaxing Zhang, Yongdong Zhang
Recently, the increasing demand for superior medical services has highlighted the discrepancies in the medical infrastructure. With big data, especially texts, forming the foundation of medical services, there is an exigent need for effective natural language processing (NLP) solutions tailored to the healthcare domain. Conventional approaches leveraging pre-trained models present promising results in this domain and current large language models (LLMs) offer advanced foundation for medical text processing. However, most medical LLMs are trained only with supervised fine-tuning (SFT), even though it efficiently empowers LLMs to understand and respond to medical instructions but is ineffective in learning domain knowledge and aligning with human preference. Another engineering barrier that prevents current medical LLM from better text processing ability is their restricted context length (e.g., 2,048 tokens), making it hard for the LLMs to process long context, which is frequently required in the medical domain. In this work, we propose ChiMed-GPT, a new benchmark LLM designed explicitly for Chinese medical domain, with enlarged context length to 4,096 tokens and undergoes a comprehensive training regime with pre-training, SFT, and RLHF. Evaluations on real-world tasks including information extraction, question answering, and dialogue generation demonstrate ChiMed-GPT's superior performance over general domain LLMs. Furthermore, we analyze possible biases through prompting ChiMed-GPT to perform attitude scales regarding discrimination of patients, so as to contribute to further responsible development of LLMs in the medical domain. The code and model are released at https://github.com/synlp/ChiMed-GPT.
Authors: Zhang Li, Biao Yang, Qiang Liu, Zhiyin Ma, Shuo Zhang, Jingxu Yang, Yabo Sun, Yuliang Liu, Xiang Bai
Large Multimodal Models (LMMs) have shown promise in vision-language tasks but struggle with high-resolution input and detailed scene understanding. Addressing these challenges, we introduce Monkey to enhance LMM capabilities. Firstly, Monkey processes input images by dividing them into uniform patches, each matching the size (e.g., 448x448) used in the original training of the well-trained vision encoder. Equipped with individual adapter for each patch, Monkey can handle higher resolutions up to 1344x896 pixels, enabling the detailed capture of complex visual information. Secondly, it employs a multi-level description generation method, enriching the context for scene-object associations. This two-part strategy ensures more effective learning from generated data: the higher resolution allows for a more detailed capture of visuals, which in turn enhances the effectiveness of comprehensive descriptions. Extensive ablative results validate the effectiveness of our designs. Additionally, experiments on 18 datasets further demonstrate that Monkey surpasses existing LMMs in many tasks like Image Captioning and various Visual Question Answering formats. Specially, in qualitative tests focused on dense text question answering, Monkey has exhibited encouraging results compared with GPT4V. Code is available at https://github.com/Yuliang-Liu/Monkey.
Authors: Haoyuan Li, Hao Jiang, Tianke Zhang, Zhelun Yu, Aoxiong Yin, Hao Cheng, Siming Fu, Yuhao Zhang, Wanggui He
Training AI models has always been challenging, especially when there is a need for custom models to provide personalized services. Algorithm engineers often face a lengthy process to iteratively develop models tailored to specific business requirements, making it even more difficult for non-experts. The quest for high-quality and efficient model development, along with the emergence of Large Language Model (LLM) Agents, has become a key focus in the industry. Leveraging the powerful analytical, planning, and decision-making capabilities of LLM, we propose a TrainerAgent system comprising a multi-agent framework including Task, Data, Model and Server agents. These agents analyze user-defined tasks, input data, and requirements (e.g., accuracy, speed), optimizing them comprehensively from both data and model perspectives to obtain satisfactory models, and finally deploy these models as online service. Experimental evaluations on classical discriminative and generative tasks in computer vision and natural language processing domains demonstrate that our system consistently produces models that meet the desired criteria. Furthermore, the system exhibits the ability to critically identify and reject unattainable tasks, such as fantastical scenarios or unethical requests, ensuring robustness and safety. This research presents a significant advancement in achieving desired models with increased efficiency and quality as compared to traditional model development, facilitated by the integration of LLM-powered analysis, decision-making, and execution capabilities, as well as the collaboration among four agents. We anticipate that our work will contribute to the advancement of research on TrainerAgent in both academic and industry communities, potentially establishing it as a new paradigm for model development in the field of AI.
Authors: Fei Zheng
Recently, large language models (LLMs) have drawn extensive attention from academia and the public, due to the advent of the ChatGPT. While LLMs show their astonishing ability in text generation for various tasks, privacy concerns limit their usage in real-life businesses. More specifically, either the user's inputs (the user sends the query to the model-hosting server) or the model (the user downloads the complete model) itself will be revealed during the usage. Vertical federated learning (VFL) is a promising solution to this kind of problem. It protects both the user's input and the knowledge of the model by splitting the model into a bottom part and a top part, which is maintained by the user and the model provider, respectively. However, in this paper, we demonstrate that in LLMs, VFL fails to protect the user input since it is simple and cheap to reconstruct the input from the intermediate embeddings. Experiments show that even with a commercial GPU, the input sentence can be reconstructed in only one second. We also discuss several possible solutions to enhance the privacy of vertical federated LLMs.
Authors: Haoran Wang, Kai Shu
To ensure AI safety, instruction-tuned Large Language Models (LLMs) are specifically trained to ensure alignment, which refers to making models behave in accordance with human intentions. While these models have demonstrated commendable results on various safety benchmarks, the vulnerability of their safety alignment has not been extensively studied. This is particularly troubling given the potential harm that LLMs can inflict. Existing attack methods on LLMs often rely on poisoned training data or the injection of malicious prompts. These approaches compromise the stealthiness and generalizability of the attacks, making them susceptible to detection. Additionally, these models often demand substantial computational resources for implementation, making them less practical for real-world applications. Inspired by recent success in modifying model behavior through steering vectors without the need for optimization, and drawing on its effectiveness in red-teaming LLMs, we conducted experiments employing activation steering to target four key aspects of LLMs: truthfulness, toxicity, bias, and harmfulness - across a varied set of attack settings. To establish a universal attack strategy applicable to diverse target alignments without depending on manual analysis, we automatically select the intervention layer based on contrastive layer search. Our experiment results show that activation attacks are highly effective and add little or no overhead to attack efficiency. Additionally, we discuss potential countermeasures against such activation attacks. Our code and data are available at https://github.com/wang2226/Backdoor-Activation-Attack Warning: this paper contains content that can be offensive or upsetting.
Authors: Ilaria Manco, Benno Weck, SeungHeon Doh, Minz Won, Yixiao Zhang, Dmitry Bogdanov, Yusong Wu, Ke Chen, Philip Tovstogan, Emmanouil Benetos, Elio Quinton, György Fazekas, Juhan Nam
We introduce the Song Describer dataset (SDD), a new crowdsourced corpus of high-quality audio-caption pairs, designed for the evaluation of music-and-language models. The dataset consists of 1.1k human-written natural language descriptions of 706 music recordings, all publicly accessible and released under Creative Common licenses. To showcase the use of our dataset, we benchmark popular models on three key music-and-language tasks (music captioning, text-to-music generation and music-language retrieval). Our experiments highlight the importance of cross-dataset evaluation and offer insights into how researchers can use SDD to gain a broader understanding of model performance.
Authors: Vukasin Bozic, Danilo Dordevic, Daniele Coppola, Joseph Thommes, Sidak Pal Singh
This work presents an analysis of the effectiveness of using standard shallow feed-forward networks to mimic the behavior of the attention mechanism in the original Transformer model, a state-of-the-art architecture for sequence-to-sequence tasks. We substitute key elements of the attention mechanism in the Transformer with simple feed-forward networks, trained using the original components via knowledge distillation. Our experiments, conducted on the IWSLT2017 dataset, reveal the capacity of these "attentionless Transformers" to rival the performance of the original architecture. Through rigorous ablation studies, and experimenting with various replacement network types and sizes, we offer insights that support the viability of our approach. This not only sheds light on the adaptability of shallow feed-forward networks in emulating attention mechanisms but also underscores their potential to streamline complex architectures for sequence-to-sequence tasks.
Authors: Yining Ye, Xin Cong, Shizuo Tian, Jiannan Cao, Hao Wang, Yujia Qin, Yaxi Lu, Heyang Yu, Huadong Wang, Yankai Lin, Zhiyuan Liu, Maosong Sun
From ancient water wheels to robotic process automation (RPA), automation technology has evolved throughout history to liberate human beings from arduous tasks. Yet, RPA struggles with tasks needing human-like intelligence, especially in elaborate design of workflow construction and dynamic decision-making in workflow execution. As Large Language Models (LLMs) have emerged human-like intelligence, this paper introduces Agentic Process Automation (APA), a groundbreaking automation paradigm using LLM-based agents for advanced automation by offloading the human labor to agents associated with construction and execution. We then instantiate ProAgent, an LLM-based agent designed to craft workflows from human instructions and make intricate decisions by coordinating specialized agents. Empirical experiments are conducted to detail its construction and execution procedure of workflow, showcasing the feasibility of APA, unveiling the possibility of a new paradigm of automation driven by agents. Our code is public at https://github.com/OpenBMB/ProAgent.
Authors: Nishant Mishra, Gaurav Sahu, Iacer Calixto, Ameen Abu-Hanna, Issam H. Laradji
Generating high-quality summaries for chat dialogs often requires large labeled datasets. We propose a method to efficiently use unlabeled data for extractive summarization of customer-agent dialogs. In our method, we frame summarization as a question-answering problem and use state-of-the-art large language models (LLMs) to generate pseudo-labels for a dialog. We then use these pseudo-labels to fine-tune a chat summarization model, effectively transferring knowledge from the large LLM into a smaller specialized model. We demonstrate our method on the \tweetsumm dataset, and show that using 10% of the original labelled data set we can achieve 65.9/57.0/61.0 ROUGE-1/-2/-L, whereas the current state-of-the-art trained on the entire training data set obtains 65.16/55.81/64.37 ROUGE-1/-2/-L. In other words, in the worst case (i.e., ROUGE-L) we still effectively retain 94.7% of the performance while using only 10% of the data.
Authors: Xiaodong Cui, Ashish Mittal, Songtao Lu, Wei Zhang, George Saon, Brian Kingsbury
Soft random sampling (SRS) is a simple yet effective approach for efficient training of large-scale deep neural networks when dealing with massive data. SRS selects a subset uniformly at random with replacement from the full data set in each epoch. In this paper, we conduct a theoretical and empirical analysis of SRS. First, we analyze its sampling dynamics including data coverage and occupancy. Next, we investigate its convergence with non-convex objective functions and give the convergence rate. Finally, we provide its generalization performance. We empirically evaluate SRS for image recognition on CIFAR10 and automatic speech recognition on Librispeech and an in-house payload dataset to demonstrate its effectiveness. Compared to existing coreset-based data selection methods, SRS offers a better accuracy-efficiency trade-off. Especially on real-world industrial scale data sets, it is shown to be a powerful training strategy with significant speedup and competitive performance with almost no additional computing cost.
Authors: Yaodong Yu, Sam Buchanan, Druv Pai, Tianzhe Chu, Ziyang Wu, Shengbang Tong, Hao Bai, Yuexiang Zhai, Benjamin D. Haeffele, Yi Ma
In this paper, we contend that a natural objective of representation learning is to compress and transform the distribution of the data, say sets of tokens, towards a low-dimensional Gaussian mixture supported on incoherent subspaces. The goodness of such a representation can be evaluated by a principled measure, called sparse rate reduction, that simultaneously maximizes the intrinsic information gain and extrinsic sparsity of the learned representation. From this perspective, popular deep network architectures, including transformers, can be viewed as realizing iterative schemes to optimize this measure. Particularly, we derive a transformer block from alternating optimization on parts of this objective: the multi-head self-attention operator compresses the representation by implementing an approximate gradient descent step on the coding rate of the features, and the subsequent multi-layer perceptron sparsifies the features. This leads to a family of white-box transformer-like deep network architectures, named CRATE, which are mathematically fully interpretable. We show, by way of a novel connection between denoising and compression, that the inverse to the aforementioned compressive encoding can be realized by the same class of CRATE architectures. Thus, the so-derived white-box architectures are universal to both encoders and decoders. Experiments show that these networks, despite their simplicity, indeed learn to compress and sparsify representations of large-scale real-world image and text datasets, and achieve performance very close to highly engineered transformer-based models: ViT, MAE, DINO, BERT, and GPT2. We believe the proposed computational framework demonstrates great potential in bridging the gap between theory and practice of deep learning, from a unified perspective of data compression. Code is available at: https://ma-lab-berkeley.github.io/CRATE .