Multiple-choice benchmarks, consisting of various prompts and choices, are among the most widely used methods to assess a language model's natural language understanding capability. Given a specific prompt, we typically compute $P(Choice|Prompt)$ to evaluate how likely a language model is to generate the correct choice compared to incorrect ones. However, we observe that performance measured using this approach reflects not only the model's comprehension of the prompt but also its inherent biases for certain choices regardless of the prompt. This issue makes it challenging to accurately measure a model's natural language understanding, as models may select the answer without fully understanding the prompt. To address this limitation, we propose a novel metric called ANPMI, which normalizes Pointwise Mutual Information (PMI) by $-\log P(Choice)$. ANPMI provides a more accurate assessment of the model's natural language understanding by ensuring that it is challenging to answer a question without properly understanding the prompt.
Stance detection on social media aims to identify attitudes expressed in tweets towards specific targets. Current studies prioritize Large Language Models (LLMs) over Small Language Models (SLMs) due to the overwhelming performance improving provided by LLMs. However, heavily relying on LLMs for stance detection, regardless of the cost, is impractical for real-world social media monitoring systems that require vast data analysis. To this end, we propose \textbf{\underline{Co}}llaborative Stance Detection via Small-Large Language Model Consistency \textbf{\underline{Ver}}ification (\textbf{CoVer}) framework, which enhances LLM utilization via context-shared batch reasoning and logical verification between LLM and SLM. Specifically, instead of processing each text individually, CoVer processes texts batch-by-batch, obtaining stance predictions and corresponding explanations via LLM reasoning in a shared context. Then, to exclude the bias caused by context noises, CoVer introduces the SLM for logical consistency verification. Finally, texts that repeatedly exhibit low logical consistency are classified using consistency-weighted aggregation of prior LLM stance predictions. Our experiments show that CoVer outperforms state-of-the-art methods across multiple benchmarks in the zero-shot setting, achieving 0.54 LLM queries per tweet while significantly enhancing performance. Our CoVer offers a more practical solution for LLM deploying for social media stance detection.
Language models are often miscalibrated, leading to confidently incorrect answers. We introduce GRACE, a benchmark for language model calibration that incorporates comparison with human calibration. GRACE consists of question-answer pairs, in which each question contains a series of clues that gradually become easier, all leading to the same answer; models must answer correctly as early as possible as the clues are revealed. This setting permits granular measurement of model calibration based on how early, accurately, and confidently a model answers. After collecting these questions, we host live human vs. model competitions to gather 1,749 data points on human and model teams' timing, accuracy, and confidence. We propose a metric, CalScore, that uses GRACE to analyze model calibration errors and identify types of model miscalibration that differ from human behavior. We find that although humans are less accurate than models, humans are generally better calibrated. Since state-of-the-art models struggle on GRACE, it effectively evaluates progress on improving model calibration.
Transformers consist of diverse building blocks, such as embedding layers, normalization layers, self-attention mechanisms, and point-wise feedforward networks. Thus, understanding the differences and interactions among these blocks is important. In this paper, we uncover a clear Sharpness Disparity across these blocks, which emerges early in training and intriguingly persists throughout the training process. Motivated by this finding, we propose Blockwise Learning Rate (LR), a strategy that tailors the LR to each block's sharpness, accelerating large language model (LLM) pre-training. By integrating Blockwise LR into AdamW, we consistently achieve lower terminal loss and nearly $2\times$ speedup compared to vanilla AdamW. We demonstrate this acceleration across GPT-2 and LLaMA, with model sizes ranging from 0.12B to 2B and datasets of OpenWebText, MiniPile, and C4. Finally, we incorporate Blockwise LR into Adam-mini (Zhang et al., 2024), a recently proposed memory-efficient variant of Adam, achieving a combined $2\times$ speedup and $2\times$ memory saving. These results underscore the potential of exploiting the sharpness disparity to improve LLM training.
Large language models (LLMs) excel at generating human-like responses but often struggle with interactive tasks that require access to real-time information. This limitation poses challenges in finance, where models must access up-to-date information, such as recent news or price movements, to support decision-making. To address this, we introduce Financial Agent, a knowledge-grounding approach for LLMs to handle financial queries using real-time text and tabular data. Our contributions are threefold: First, we develop a Financial Context Dataset of over 50,000 financial queries paired with the required context. Second, we train FinBloom 7B, a custom 7 billion parameter LLM, on 14 million financial news articles from Reuters and Deutsche Presse-Agentur, alongside 12 million Securities and Exchange Commission (SEC) filings. Third, we fine-tune FinBloom 7B using the Financial Context Dataset to serve as a Financial Agent. This agent generates relevant financial context, enabling efficient real-time data retrieval to answer user queries. By reducing latency and eliminating the need for users to manually provide accurate data, our approach significantly enhances the capability of LLMs to handle dynamic financial tasks. Our proposed approach makes real-time financial decisions, algorithmic trading and other related tasks streamlined, and is valuable in contexts with high-velocity data flows.
The emergence of audio language models is empowered by neural audio codecs, which establish critical mappings between continuous waveforms and discrete tokens compatible with language model paradigms. The evolutionary trends from multi-layer residual vector quantizer to single-layer quantizer are beneficial for language-autoregressive decoding. However, the capability to handle multi-domain audio signals through a single codebook remains constrained by inter-domain distribution discrepancies. In this work, we introduce UniCodec, a unified audio codec with a single codebook to support multi-domain audio data, including speech, music, and sound. To achieve this, we propose a partitioned domain-adaptive codebook method and domain Mixture-of-Experts strategy to capture the distinct characteristics of each audio domain. Furthermore, to enrich the semantic density of the codec without auxiliary modules, we propose a self-supervised mask prediction modeling approach. Comprehensive objective and subjective evaluations demonstrate that UniCodec achieves excellent audio reconstruction performance across the three audio domains, outperforming existing unified neural codecs with a single codebook, and even surpasses state-of-the-art domain-specific codecs on both acoustic and semantic representation capabilities.
Conversational speech synthesis (CSS) aims to synthesize both contextually appropriate and expressive speech, and considerable efforts have been made to enhance the understanding of conversational context. However, existing CSS systems are limited to deterministic prediction, overlooking the diversity of potential responses. Moreover, they rarely employ language model (LM)-based TTS backbones, limiting the naturalness and quality of synthesized speech. To address these issues, in this paper, we propose DiffCSS, an innovative CSS framework that leverages diffusion models and an LM-based TTS backbone to generate diverse, expressive, and contextually coherent speech. A diffusion-based context-aware prosody predictor is proposed to sample diverse prosody embeddings conditioned on multimodal conversational context. Then a prosody-controllable LM-based TTS backbone is developed to synthesize high-quality speech with sampled prosody embeddings. Experimental results demonstrate that the synthesized speech from DiffCSS is more diverse, contextually coherent, and expressive than existing CSS systems
We release 70,509 high-quality social networks extracted from multilingual fiction and nonfiction narratives. We additionally provide metadata for $\sim$30,000 of these texts (73\% nonfiction and 27\% fiction) written between 1800 and 1999 in 58 languages. This dataset provides information on historical social worlds at an unprecedented scale, including data for 2,510,021 individuals in 2,805,482 pair-wise relationships annotated for affinity and relationship type. We achieve this scale by automating previously manual methods of extracting social networks; specifically, we adapt an existing annotation task as a language model prompt, ensuring consistency at scale with the use of structured output. This dataset serves as a unique resource for humanities and social science research by providing data on cognitive models of social realities.
The increasing complexity of software systems has driven significant advancements in program analysis, as traditional methods unable to meet the demands of modern software development. To address these limitations, deep learning techniques, particularly Large Language Models (LLMs), have gained attention due to their context-aware capabilities in code comprehension. Recognizing the potential of LLMs, researchers have extensively explored their application in program analysis since their introduction. Despite existing surveys on LLM applications in cybersecurity, comprehensive reviews specifically addressing their role in program analysis remain scarce. In this survey, we systematically review the application of LLMs in program analysis, categorizing the existing work into static analysis, dynamic analysis, and hybrid approaches. Moreover, by examining and synthesizing recent studies, we identify future directions and challenges in the field. This survey aims to demonstrate the potential of LLMs in advancing program analysis practices and offer actionable insights for security researchers seeking to enhance detection frameworks or develop domain-specific models.
With the urbanization process, an increasing number of sensors are being deployed in transportation systems, leading to an explosion of big data. To harness the power of this vast transportation data, various machine learning (ML) and artificial intelligence (AI) methods have been introduced to address numerous transportation challenges. However, these methods often require significant investment in data collection, processing, storage, and the employment of professionals with expertise in transportation and ML. Additionally, privacy issues are a major concern when processing data for real-world traffic control and management. To address these challenges, the research team proposes an innovative Multi-agent framework named Independent Mobility GPT (IDM-GPT) based on large language models (LLMs) for customized traffic analysis, management suggestions, and privacy preservation. IDM-GPT efficiently connects users, transportation databases, and ML models economically. IDM-GPT trains, customizes, and applies various LLM-based AI agents for multiple functions, including user query comprehension, prompts optimization, data analysis, model selection, and performance evaluation and enhancement. With IDM-GPT, users without any background in transportation or ML can efficiently and intuitively obtain data analysis and customized suggestions in near real-time based on their questions. Experimental results demonstrate that IDM-GPT delivers satisfactory performance across multiple traffic-related tasks, providing comprehensive and actionable insights that support effective traffic management and urban mobility improvement.
Recent advances in deep learning have revolutionized seismic monitoring, yet developing a foundation model that performs well across multiple complex tasks remains challenging, particularly when dealing with degraded signals or data scarcity. This work presents SeisMoLLM, the first foundation model that utilizes cross-modal transfer for seismic monitoring, to unleash the power of large-scale pre-training from a large language model without requiring direct pre-training on seismic datasets. Through elaborate waveform tokenization and fine-tuning of pre-trained GPT-2 model, SeisMoLLM achieves state-of-the-art performance on the DiTing and STEAD datasets across five critical tasks: back-azimuth estimation, epicentral distance estimation, magnitude estimation, phase picking, and first-motion polarity classification. It attains 36 best results out of 43 task metrics and 12 top scores out of 16 few-shot generalization metrics, with many relative improvements ranging from 10% to 50%. In addition to its superior performance, SeisMoLLM maintains efficiency comparable to or even better than lightweight models in both training and inference. These findings establish SeisMoLLM as a promising foundation model for practical seismic monitoring and highlight cross-modal transfer as an exciting new direction for earthquake studies, showcasing the potential of advanced deep learning techniques to propel seismology research forward.
Goal-oriented de novo molecule design, namely generating molecules with specific property or substructure constraints, is a crucial yet challenging task in drug discovery. Existing methods, such as Bayesian optimization and reinforcement learning, often require training multiple property predictors and struggle to incorporate substructure constraints. Inspired by the success of Large Language Models (LLMs) in text generation, we propose ChatMol, a novel approach that leverages LLMs for molecule design across diverse constraint settings. Initially, we crafted a molecule representation compatible with LLMs and validated its efficacy across multiple online LLMs. Afterwards, we developed specific prompts geared towards diverse constrained molecule generation tasks to further fine-tune current LLMs while integrating feedback learning derived from property prediction. Finally, to address the limitations of LLMs in numerical recognition, we referred to the position encoding method and incorporated additional encoding for numerical values within the prompt. Experimental results across single-property, substructure-property, and multi-property constrained tasks demonstrate that ChatMol consistently outperforms state-of-the-art baselines, including VAE and RL-based methods. Notably, in multi-objective binding affinity maximization task, ChatMol achieves a significantly lower KD value of 0.25 for the protein target ESR1, while maintaining the highest overall performance, surpassing previous methods by 4.76%. Meanwhile, with numerical enhancement, the Pearson correlation coefficient between the instructed property values and those of the generated molecules increased by up to 0.49. These findings highlight the potential of LLMs as a versatile framework for molecule generation, offering a promising alternative to traditional latent space and RL-based approaches.
We present WinoMTDE, a new gender bias evaluation test set designed to assess occupational stereotyping and underrepresentation in German machine translation (MT) systems. Building on the automatic evaluation method introduced by arXiv:1906.00591v1 [cs.CL], we extend the approach to German, a language with grammatical gender. The WinoMTDE dataset comprises 288 German sentences that are balanced in regard to gender, as well as stereotype, which was annotated using German labor statistics. We conduct a large-scale evaluation of five widely used MT systems and a large language model. Our results reveal persistent bias in most models, with the LLM outperforming traditional systems. The dataset and evaluation code are publicly available under https://github.com/michellekappl/mt_gender_german.
Long-context transformers face significant efficiency challenges due to the quadratic cost of self-attention. However, many modern applications-from multi-turn dialogue to high-resolution vision-require contexts spanning tens of thousands of tokens. We introduce SPECTRE, a method that replaces each attention head with a fast real FFT, a content-adaptive spectral gate, and an inverse FFT, reducing per-layer complexity from $\mathcal{O}(L^{2})$ to $O(L\log L)$ while preserving the surrounding architecture. We extend this efficiency to autoregressive generation through our Prefix-FFT cache and enhance local feature representation with an optional wavelet module that adds negligible computational overhead. Our experiments demonstrate that SPECTRE operates up to 7$\times$ faster than FlashAttention-2 on 128k-token contexts while matching or exceeding baseline performance on PG-19 language modeling and ImageNet-1k classification tasks. SPECTRE achieves these improvements by adding fewer than 6\% parameters to the base model, making hundred-kilotoken context processing feasible on commodity GPUs without specialized hardware.
Grading student assignments in STEM courses is a laborious and repetitive task for tutors, often requiring a week to assess an entire class. For students, this delay of feedback prevents iterating on incorrect solutions, hampers learning, and increases stress when exercise scores determine admission to the final exam. Recent advances in AI-assisted education, such as automated grading and tutoring systems, aim to address these challenges by providing immediate feedback and reducing grading workload. However, existing solutions often fall short due to privacy concerns, reliance on proprietary closed-source models, lack of support for combining Markdown, LaTeX and Python code, or excluding course tutors from the grading process. To overcome these limitations, we introduce PyEvalAI, an AI-assisted evaluation system, which automatically scores Jupyter notebooks using a combination of unit tests and a locally hosted language model to preserve privacy. Our approach is free, open-source, and ensures tutors maintain full control over the grading process. A case study demonstrates its effectiveness in improving feedback speed and grading efficiency for exercises in a university-level course on numerics.
In the competitive landscape of advertising, success hinges on effectively navigating and leveraging complex interactions among consumers, advertisers, and advertisement platforms. These multifaceted interactions compel advertisers to optimize strategies for modeling consumer behavior, enhancing brand recall, and tailoring advertisement content. To address these challenges, we present MindMem, a multimodal predictive model for advertisement memorability. By integrating textual, visual, and auditory data, MindMem achieves state-of-the-art performance, with a Spearman's correlation coefficient of 0.631 on the LAMBDA and 0.731 on the Memento10K dataset, consistently surpassing existing methods. Furthermore, our analysis identified key factors influencing advertisement memorability, such as video pacing, scene complexity, and emotional resonance. Expanding on this, we introduced MindMem-ReAd (MindMem-Driven Re-generated Advertisement), which employs Large Language Model-based simulations to optimize advertisement content and placement, resulting in up to a 74.12% improvement in advertisement memorability. Our results highlight the transformative potential of Artificial Intelligence in advertising, offering advertisers a robust tool to drive engagement, enhance competitiveness, and maximize impact in a rapidly evolving market.
In this paper, we introduce LDGen, a novel method for integrating large language models (LLMs) into existing text-to-image diffusion models while minimizing computational demands. Traditional text encoders, such as CLIP and T5, exhibit limitations in multilingual processing, hindering image generation across diverse languages. We address these challenges by leveraging the advanced capabilities of LLMs. Our approach employs a language representation strategy that applies hierarchical caption optimization and human instruction techniques to derive precise semantic information,. Subsequently, we incorporate a lightweight adapter and a cross-modal refiner to facilitate efficient feature alignment and interaction between LLMs and image features. LDGen reduces training time and enables zero-shot multilingual image generation. Experimental results indicate that our method surpasses baseline models in both prompt adherence and image aesthetic quality, while seamlessly supporting multiple languages. Project page: https://zrealli.github.io/LDGen.
In recent years, machine learning has profoundly reshaped the field of chemistry, facilitating significant advancements across various applications, including the prediction of molecular properties and the generation of molecular structures. Language models and graph-based models are extensively utilized within this domain, consistently achieving state-of-the-art results across an array of tasks. However, the prevailing practice of representing chemical compounds in the SMILES format -- used by most datasets and many language models -- presents notable limitations as a training data format. In contrast, chemical fingerprints offer a more physically informed representation of compounds, thereby enhancing their suitability for model training. This study aims to develop a language model that is specifically trained on fingerprints. Furthermore, we introduce a bimodal architecture that integrates this language model with a graph model. Our proposed methodology synthesizes these approaches, utilizing RoBERTa as the language model and employing Graph Isomorphism Networks (GIN), Graph Convolutional Networks (GCN) and Graphormer as graph models. This integration results in a significant improvement in predictive performance compared to conventional strategies for tasks such as Quantitative Structure-Activity Relationship (QSAR) and the prediction of nuclear magnetic resonance (NMR) spectra, among others.
In this demonstration, we develop IntentRec4Maps, a system to recognise users' intentions in interactive maps for real-world navigation. IntentRec4Maps uses the Google Maps Platform as the real-world interactive map, and a very effective approach for recognising users' intentions in real-time. We showcase the recognition process of IntentRec4Maps using two different Path-Planners and a Large Language Model (LLM). GitHub: https://github.com/PeijieZ/IntentRec4Maps
Recent research across mathematical problem solving, proof assistant programming and multimodal jailbreaking documents a striking finding: when (multimodal) language model tackle a suite of tasks with multiple attempts per task -- succeeding if any attempt is correct -- then the negative log of the average success rate scales a power law in the number of attempts. In this work, we identify an apparent puzzle: a simple mathematical calculation predicts that on each problem, the failure rate should fall exponentially with the number of attempts. We confirm this prediction empirically, raising a question: from where does aggregate polynomial scaling emerge? We then answer this question by demonstrating per-problem exponential scaling can be made consistent with aggregate polynomial scaling if the distribution of single-attempt success probabilities is heavy tailed such that a small fraction of tasks with extremely low success probabilities collectively warp the aggregate success trend into a power law - even as each problem scales exponentially on its own. We further demonstrate that this distributional perspective explains previously observed deviations from power law scaling, and provides a simple method for forecasting the power law exponent with an order of magnitude lower relative error, or equivalently, ${\sim}2-4$ orders of magnitude less inference compute. Overall, our work contributes to a better understanding of how neural language model performance improves with scaling inference compute and the development of scaling-predictable evaluations of (multimodal) language models.