Discrete Tokenization for Multimodal LLMs: A Comprehensive Survey.

The paper studies the problem of software code migration during the transition from the old version of the SAP ERP 6.0 system to modern ERP platforms, in particular SAP S/4HANA. The necessity of automation of the migration process using artificial intelligence methods and tools is substantiated. A comprehensive analysis of existing academic and industrial case studies confirmed the feasibility of using large language models for code migration. A number of advantages of the approach based on large language models are revealed, in particular, the ability to handle various migration scenarios. The key risks and limitations common in the enterprise environment are highlighted, including data security, privacy, high cost of API calls and limited knowledge of models regarding the subject area. A method is developed for selecting a large open source language model that can be safely used in the enterprise landscape, thereby mitigating the risks associated with cloud solutions. The method is based on the developed system of criteria, including: model size, context window size, license type, support for fine-tuning training, focus on code quality (measured using a reference dataset), support for model quantization and community support. The hybrid AHP-TOPSIS method is proposed for systematic ranking of candidate models. The AHP method was used to check the weights of the criteria, while TOPSIS was used to rank the models based on their proximity to the ideal solution. Based on the developed method, three large language models were selected: Qwen 2.5 Coder 14B, DeepSeek-Coder-V2 16B and Llama 3.1 8B. A comprehensive testing plan for the selected models was developed for further experimental studies. The scientific novelty of the work lies in the development of a method for selecting large language models for custom code migration tasks, which differs from the existing system of model selection criteria and the use of a hybrid AHP-TOPSIS approach for ranking candidate models. The practical value of the work is that the developed method allows you to select a model taking into account the constraints of the corporate environment and information security requirements, which will increase the level of automation in the migration of AVAP when switching to new versions of ERP systems from SAP SE. The developed method can be applied to the selection of large language models when upgrading any large computer systems developed in both open source and proprietary programming languages.

As Large Language Models (LLMs) become increasingly integrated into everyday life as general-purpose multimodal AI systems, their capabilities to simulate human understanding are under examination. This study investigates LLMs’ ability to interpret linguistic pragmatics, which involves context and implied meanings. Using Grice’s communication principles, we evaluated both LLMs (GPT-2, GPT-3, GPT-3.5, GPT-4, and Bard) and human subjects (N = 147) on dialogue-based tasks. Human participants included 71 primarily Serbian students and 76 native English speakers from the United States. Findings revealed that LLMs, particularly GPT-4, outperformed humans. GPT-4 achieved the highest score of 4.80, surpassing the best human score of 4.55. Other LLMs performed well: GPT-3.5 scored 4.10, Bard 3.75, and GPT-3 3.25; GPT-2 had the lowest score of 1.05. The average LLM score was 3.39, exceeding the human cohorts’ averages of 2.80 (Serbian students) and 2.34 (U.S. participants). In the ranking of all 155 subjects (including LLMs and humans), GPT-4 secured the top position, while the best human ranked second. These results highlight significant progress in LLMs’ ability to simulate understanding of linguistic pragmatics. Future studies should confirm these findings with more dialogue-based tasks and diverse participants. This research has important implications for advancing general-purpose AI models in various communication-centered tasks, including potential application in humanoid robots in the future.

The subject of the study is the analysis and improvement of methods for determining the relevance of project names to the information content of purchases using large language models. The object of the study is a database containing the names of projects and purchases in the field of electric power industry, collected from open sources. The author examines in detail such aspects of the topic as the use of TF-IDF and cosine similarity metrics for primary data filtering, and also describes in detail the integration and evaluation of the effectiveness of large language models such as GigaChat, GPT-3.5, and GPT-4 in text data matching tasks. Special attention is paid to the methods of clarifying the similarity of names based on reflection introduced into the prompta of large language models, which makes it possible to increase the accuracy of data comparison. The study uses TF-IDF and cosine similarity methods for primary data analysis, as well as large GigaChat, GPT-3.5 and GPT-4 language models for detailed verification of the relevance of project names and purchases, including reflection in model prompta to improve the accuracy of results. The novelty of the research lies in the development of a combined approach to determining the relevance of project names and purchases, combining traditional methods of processing text information (TF-IDF, cosine similarity) with the capabilities of large language models. A special contribution of the author to the research of the topic is the proposed methodology for improving the accuracy of data comparison by clarifying the results of primary selection using GPT-3.5 and GPT-4 models with optimized prompta, including reflection. The main conclusions of the study are confirmation of the prospects of using the developed approach in the tasks of information support for procurement processes and project implementation, as well as the possibility of using the results obtained for the development of text data mining systems in various sectors of the economy. The study showed that the use of language models makes it possible to improve the value of the F2 measure to 0.65, which indicates a significant improvement in the quality of data comparison compared with basic methods.

This paper discusses a multimodal AI system applied to legal reasoning for tax law. The results given here are very general and apply to systems developed for other areas besides tax law. A central goal of this work is to gain a better understanding of the relationships between LLMs (Large Language Models) and automated theorem-proving methodologies. To do this, we suppose (1) two cases for the theorem-proving system: one where it has a countable number of total meanings for its countable number of atoms and the other is where it has an uncountable number of total meanings for its countable number of atoms, and (2) LLMs can have an uncountable number of token meanings. With this in mind, the results given in this paper use the downward and upward Löwenheim–Skolem theorems and logical model theory to contrast these two AI modalities. One modality focuses on syntactic proofs and the other focuses on logical semantics based on LLMs. Particularly, one modality uses a rule-based first-order logic theorem-proving system to perform legal reasoning. The objective of this theorem-proving system is to provide proofs as evidence of valid legal reasoning when enacted laws are applied to particular situations. These proofs are syntactic structures that can be presented in the form of narrative explanations of how the answer to the legal question was determined. The second modality uses LLMs to analyze and transform a user’s tax query so this query can be sent to a first-order logic theorem-proving system to perform its legal reasoning function. The main goal of our application of LLMs is to enhance and simplify user input and output for the theorem-proving system. Using logical model theory, we show how there can exist an equivalence between laws represented in logic of the theorem-proving system, fixed in time when the theorem-proving system was set up, and new semantics given by LLMs. These results are based on logical model theory and Löwenheim–Skolem theorems.

Generative large language models (LLMs) are a subset of transformers-based neural network architecture models. LLMs have successfully leveraged a combination of an increased number of parameters, improvements in computational efficiency, and large pre-training datasets to perform a wide spectrum of natural language processing (NLP) tasks. Using a few examples (few-shot) or no examples (zero-shot) for prompt-tuning has enabled LLMs to achieve state-of-the-art performance in a broad range of NLP applications. This article by the American Medical Informatics Association (AMIA) NLP Working Group characterizes the opportunities, challenges, and best practices for our community to leverage and advance the integration of LLMs in downstream NLP applications effectively. This can be accomplished through a variety of approaches, including augmented prompting, instruction prompt tuning, and reinforcement learning from human feedback (RLHF). Our focus is on making LLMs accessible to the broader biomedical informatics community, including clinicians and researchers who may be unfamiliar with NLP. Additionally, NLP practitioners may gain insight from the described best practices. We focus on 3 broad categories of NLP tasks, namely natural language understanding, natural language inferencing, and natural language generation. We review the emerging trends in prompt tuning, instruction fine-tuning, and evaluation metrics used for LLMs while drawing attention to several issues that impact biomedical NLP applications, including falsehoods in generated text (confabulation/hallucinations), toxicity, and dataset contamination leading to overfitting. We also review potential approaches to address some of these current challenges in LLMs, such as chain of thought prompting, and the phenomena of emergent capabilities observed in LLMs that can be leveraged to address complex NLP challenge in biomedical applications.

Game theory establishes a fundamental framework for analyzing strategic interactions among rational decision-makers. The rapid advancement of large language models (LLMs) has sparked extensive research exploring the intersection of these two fields. Specifically, game-theoretic methods are being applied to evaluate and enhance LLM capabilities, while LLMs themselves are reshaping classic game models. This paper presents a comprehensive survey of the intersection of these fields, exploring a bidirectional relationship from three perspectives: (1) Establishing standardized game-based benchmarks for evaluating LLM behavior; (2) Leveraging game-theoretic methods to improve LLM performance through algorithmic innovations; (3) Characterizing the societal impacts of LLMs through game modeling. Among these three aspects, we also highlight how the equilibrium analysis for traditional game models is impacted by LLMs' advanced language understanding, which in turn extends the study of game theory. Finally, we identify key challenges and future research directions, assessing their feasibility based on the current state of the field. By bridging theoretical rigor with emerging AI capabilities, this survey aims to foster interdisciplinary collaboration and drive progress in this evolving research area.

Background: Recent advancements in large language model (LLM) technologies have introduced powerful open-source instruction-tuned LLMs that match the text generation quality of leading models like GPT-4. Despite accelerating LLM adoption in sensitive-information environments, the lack of disclosed training data hinders replication and makes these achievements exclusive to specific models. Purpose: Given the multilingual nature of the latest iteration of open-source LLMs, the benefits of training language-specific LLMs diminish, leaving computational efficiency as the sole guaranteed advantage of this computationally-expensive procedure. This work aims to address the language-adaptation limitations posed by restricted access to high-quality instruction-tuning data, offering a more cost-effective pipeline. Method: To tackle language-adaptation challenges, we introduce Learned Embedding Propagation (LEP), a novel method with lower training data requirements and minimal disruption of existing LLM knowledge. LEP employs an innovative embedding propagation technique, bypassing the need for instruction-tuning and directly integrating new language knowledge into any instruct-tuned LLM variant. Additionally, we developed Darumeru, a new benchmark for evaluating text generation robustness during training, specifically tailored for Russian adaptation. Results: We applied the LEP method to adapt LLaMa-3-8B and Mistral-7B for Russian, testing four different vocabulary adaptation scenarios. Evaluation demonstrates that LEP achieves competitive performance levels, comparable to OpenChat 3.5 and LLaMa-3-8B-Instruct. Further improvements were observed through self-calibration and additional instruction-tuning steps, enhancing task-solving capabilities beyond the original models. Conclusion: LEP offers a viable and efficient alternative to traditional language-specific instruction-tuning, significantly reducing the costs associated with language adaptation while maintaining or surpassing the performance benchmarks set by contemporary LLMs.

Forgetting is the complementary process to memory, enabling both biological and artificial systems to prioritize relevant information while discarding outdated or redundant data. In large language models (LLMs), forgetting occurs through computational mechanisms such as attention bottlenecks, catastrophic interference, and optimization dynamics, mirroring neurobiological processes such as synaptic decay and memory consolidation. While existing research has extensively studied memory mechanisms in LLMs, there remains a critical gap in systematically analyzing forgetting phenomena and their implications for AI system design. This study presents a comprehensive review of forgetting mechanisms in LLMs through a neuroscience-inspired lens. This study first establishes parallels between human forgetting processes and computational forgetting patterns in LLMs. Building on these foundations, this study proposes a novel taxonomy that classifies LLM forgetting across four key dimensions. This taxonomy is further supported by the introduction of quantitative evaluation metrics to provide additional evidence. Furthermore, this study demonstrates how controlled forgetting can be transformed from a limitation into a design feature through bio-inspired optimization strategies. These approaches enable critical applications in privacy-compliant AI, adaptive education systems, and sustainable computing. The study concludes by identifying open challenges in generalization, ethical governance of machine forgetting, and computational efficiency, while outlining future directions for developing LLMs with human-like memory adaptability.

As the Large Language Model (LLM) becomes increasingly important in various domains, the performance of LLM inference is crucial to massive LLM applications. However, centering around the computational efficiency and the memory utilization, the following challenges remain unsolved in achieving high-throughput LLM inference: (1) Synchronous partial softmax update. The softmax operation requires a synchronous update operation among each partial softmax result, leading to ~20% overheads for the attention computation in LLMs. (2) Under-utilized computation of flat GEMM. The shape of matrices performing GEMM in LLM inference tends to be flat, leading to under-utilized computation and 50% performance loss after padding zeros in previous designs (e.g., cuBLAS, CUTLASS, etc.). (3) Memory redundancy caused by activations. Dynamic allocation of activations during inference leads to redundant storage of useless variables, bringing 22% more memory consumption. <p xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">We present <i>FlashDecoding</i>++<i>Next</i>, a high-throughput inference engine supporting mainstream LLMs and hardware backends. To tackle the above challenges, <i>FlashDecoding</i>++<i>Next</i> creatively proposes: (1) Asynchronous softmax with unified maximum. <i>FlashDecoding</i>++<i>Next</i> introduces a unified maximum technique for different partial softmax computations to avoid synchronization. Based on this, a fine-grained pipelining is proposed, leading to 1.18× and 1.14× for the <i>prefill</i> and <i>decode</i> phases in LLM inference, respectively. (2) Flat GEMM optimization with double buffering. <i>FlashDecoding</i>++<i>Next</i> points out that flat GEMMs with different shapes face varied bottlenecks. Then, techniques like double buffering are introduced, resulting in up to 52% speedup for the flat GEMM operation. (3) Buffer reusing and unified memory management. <i>FlashDecoding</i>++<i>Next</i> reuses the pre-allocated activation buffers throughout the inference process to remove redundancy. Based on that, we unify the management of different types of storage to further exploit the reusing opportunity. The memory optimization enables up to 1.57× longer sequence to be processed. <i>FlashDecoding</i>++<i>Next</i> demonstrates remarkable throughput improvement, delivering up to 68.88× higher throughput compared to the HuggingFace [1] implementation. On average, <i>FlashDecoding</i>++<i>Next</i> achieves 1.25× and 1.46× higher throughput compared to vLLM [2] and TensorRT-LLM [3] on mainstream LLMs.

IntroductionPatients missing their appointments (no-shows) are a persistent issue that results in idle resources while delaying critical patient prognosis. Likewise, long waiting times increase frustration for patients, leaving a negative impression on the appointment. In this paper, we explore 3 modalities of diagnostic and interventional radiology appointments for pediatric patients at the Hospital for Sick Children (SickKids), Toronto, ON, Canada. Our goal was to survey machine learning methods that best predict the risk of patient no-shows and long wait-times exceeding 1 hour for scheduling teams to propose targeted downstream accommodations.MethodsWe experimented with 6 predictive model types separately trained on both tasks which included extreme gradient boosting (XGBoost), Random Forest (RF), Support Vector Machine, Logistic Regression, Artificial Neural Network, and a pre-trained large language model (LLM). Utilizing 20 features containing a mixture of patient demographics and appointment related data, we experimented with different data balancing methods including instance hardness threshold (IHT) and class weighting to reduce bias in prediction. We then conducted a comparative study of the improvements made by utilizing continuous contextual data in our LLM which boasted a 51% improvement in F1 score for the wait-time model.ResultsOur XGBoost model had the best combination of AUC and F1 scores (0.96 and 0.62, respectively) for predicting no-show while RF had the best AUC and F1 scores (0.83 and 0.61, respectively) for wait-time prediction. The LLMs also performed well for 90% probability thresholds (high risk patients) while being robustly calibrated on unseen test data.DiscussionOur results surveyed multiple algorithms and data balancing methods to propose the greatest performing models on our tasks, implemented a unique methodology to use LLMs on heterogeneous data within this domain, and demonstrated the greater importance of contextual appointment data over patient demographic features for a more equitable prediction algorithm. Going forward, the predictive output (calibrated probabilities of events) can be used as stochastic input for risk-based optimized scheduling to provide accommodation for patients less likely to receive quality access to healthcare.

This article explores various applications of artificial intelligence (AI) technologies in dairy farming, including the use of computer vision systems (CVS) for animal identification, BCS and body shape analysis, and potential uses of large language models (LLM) in the dairy industry. Among recent advancements in precision livestock farming tools, CVS have gained popularity as powerful solutions for individual animal monitoring. These systems can capture phenotypes from multiple animals simultaneously using a single device in an automated and nonintrusive manner. To match animals with their corresponding predicted phenotypes, these systems require individual animal identification, which can be achieved through external identification systems or computer vision-based animal identification algorithms. Additionally, modern natural language processing techniques, such as LLM, offer opportunities for advanced data integration, including unstructured textual data. Furthermore, we discuss the challenges associated with integrating data from different sources and modalities, such as images, text, and tabular data, into multimodal machine learning systems for phenotype prediction, which also represents a key area of AI application. Digital technologies such as CVS and LLM have the potential to transform dairy farming; CVS can provide individual and objective assessments of animal health, whereas LLM can integrate diverse data sources for phenotype prediction. Although there are many potential challenges ahead, these technologies offer significant opportunities for advancing animal health monitoring, farm management, and individual phenotyping.

Large language models (LLMs) such as ChatGPT o1, ChatGPT o3, and DeepSeek R1 have shown great potential in solving difficult problems. However, current LLM evaluation benchmarks are limited to one-step interactions. Some of the existing sequence decision-making environments, such as TextStarCraftII and LLM-PySC2, are too complicated and require hours of interaction to complete a game. In this paper, we introduce LLM-Cave, a benchmark and light environment for LLM reasoning and decision-making systems. This environment is a classic instance in the era of Symbolism. Artificial intelligence enables the agent to explore the environment and avoid potential losses by reasoning about nearby dangers using partial observable state information. In the experiment, we evaluated the sequential reasoning ability, decision-making performance and computational efficiency of mainstream large language models (LLMs) such as GPT-4o-mini, o1-mini, and DeepSeek-R1. Experiments show that while Deepseek-R1 achieved the highest success rate on complex reasoning tasks, smaller models like 4o-mini significantly narrowed the performance gap on challenges by employing Chain of Speculation and Planner-Critic strategies, at the expense of reduced computational efficiency. This indicates that structured, multi-step reasoning combined with an LLM-based feedback mechanism can substantially enhance an LLM's decision-making capabilities, providing a promising direction for improving reasoning in weaker models and suggesting a new reasoning-centered benchmark for LLM assessment. Our code is open-sourced in https://github.com/puleya1277/CaveEnv.

This paper examines the potential of integrating Large Language Models (LLMs) with knowledge bases to imrpove the accuracy and reliability of their responses. The advantages of such a combination are evaluated, particularly in reducing the risk of hallucinations – the phenomenon where models generate erroneous or fabricated information. Various methodologies for combining LLMs with knowledge bases are analyzed, along with their respective advantages and limitations. The prospects and challenges of implementing this technology in diverse fields—such as information retrieval, decision support, and automated content creation—are discussed. The paper presents an overview of the current state of research in this domain and delineates directions for future investigation.The integration of LLMs with knowledge bases represents a significant advancement in artificial intelligence, aiming one of the key concerns regarding LLMs—their tendency to generate inaccurate or fabricated information, commonly referred to as hallucinations. This approach leverages the vast language understanding and generation capabilities of LLMs while grounding their outputs in structured and verified information from knowledge bases. The synergy between these two technologies has the potential to significantly enhance the reliability and factual accuracy of AI-generated responses across a wide range of applications. The methodologies for combining LLMs with knowledge bases differ in their implementation and effectiveness. Some approaches involve pre-training LLMs on curated knowledge bases, while others reference knowledge bases externally during the inference process. Each method presents its own set of advantages and challenges, such as balancing computational efficiency against accuracy and maintaining the fluency of LLM outputs while adhering strictly to factual information. The application of this integrated technology extends beyond mere information retrieval, showing promise in complex decision support systems, automated content creation for specialized domains, and contributing to the advancement of explainable AI by providing traceable sources for generated information. As research in this area progresses, it is expected to open new avenues for developing more trustworthy and capable AI systems across various industries and academic disciplines.

Large language models (LLMs) are rapidly advancing natural language processing and driving increasing interest across biomedical research, clinical care, and healthcare operations. This systematic review synthesizes current empirical evidence on LLM applications in healthcare and biomedical informatics, focusing on their capabilities, evaluation practices, and reported outcomes. Existing studies highlight promising uses in research, where LLMs assist literature-informed genomic interpretation, functional annotation, and biological hypothesis generation, as well as support tasks related to protein and multiomics analysis. In clinical contexts, LLMs are primarily evaluated for natural language-driven tasks, including electronic health record summarization, clinical documentation support, medical question answering, and information extraction, rather than autonomous diagnostic or therapeutic decision-making. Early investigations also describe potential value in healthcare operations, patient communication, clinician education, and drug discovery workflows largely through knowledge retrieval, text generation, and semantic search. However, current evidence remains preliminary: most studies are retrospective, benchmark-based, simulation-driven, or limited to controlled research settings. Key challenges include hallucination and unreliable reasoning, bias and inequitable performance across populations, data privacy and security constraints, reproducibility limitations, and lack of prospective clinical validation and regulatory guidance. Emerging strategies, such as domain-specific pretraining, retrieval augmentation, multi-modal architectures, and privacy-preserving learning, aim to improve reliability, safety, and real-world applicability. This review concludes by outlining methodological, infrastructural, and governance requirements needed to responsibly integrate LLMs into biomedical workflows, emphasizing that clinical deployment remains exploratory and must be rigorously evaluated.

Human beings capable of making and using tools can accomplish tasks far beyond their innate abilities, and this paradigm of integration with tools may not be limited to humans themselves. Recently, the large language model (LLM) has demonstrated immense potential across various fields with its unique planning and reasoning abilities. However, there are still many challenges beyond its capabilities due to deficiencies in its training data and inherent illusions. Thus, integrating LLMs and tools into tool learning agents has become a new emerging research direction. To this end, we present a systematic investigation and comprehensive review of tool-learning agents in this paper. We start by introducing the definition of the tool learning task for Agents and then illustrating the typical architecture of the tool-learning models. Since these tools are all defined by users, LLM does not know what tools there are and what their functions are. Thus, LLMs should first find appropriate tools and split the tool retrieval methods into two categories: training-based and non-training-based. To accurately complete the user task, it is important to decompose the task into several sub-tasks and execute them in the correct order. Following that, we introduce the tool planning methods and organize these works by whether they rely on the model’s inherent reasoning capabilities for planning or utilize external reasoning tools. Due to the rapid development of this field, we also introduce an emerging frontier direction: using multimodal tools for LLM. In addition, we compile current open-source benchmarks and evaluation metrics, focusing on their scale, composition, calculation methods, and assessment dimensions. Next, we introduce several application scenarios for the LLM-based tool learning methods. Finally, we discuss the safety and ethical issues involved in tool learning.