In the rapidly evolving landscape of large language models (LLMs) for medical applications, ensuring the reliability and accuracy of these models in clinical settings is paramount. Existing benchmarks often focus on fixed-format tasks like multiple-choice QA, which fail to capture the complexity of real-world clinical diagnostics. Moreover, traditional evaluation metrics and LLM-based evaluators struggle with misalignment, often providing oversimplified assessments that do not adequately reflect human judgment. To address these challenges, we introduce HDCEval, a **H**ierarchical **D**ivide-and-**C**onquer **Eval**uation framework tailored for fine-grained alignment in medical evaluation. HDCEval is built on a set of fine-grained medical evaluation guidelines developed in collaboration with professional doctors, encompassing Patient Question Relevance, Medical Knowledge Correctness, and Expression. The framework decomposes complex evaluation tasks into specialized subtasks, each evaluated by expert models trained through Attribute-Driven Token Optimization (ADTO) on a meticulously curated preference dataset. This hierarchical approach ensures that each aspect of the evaluation is handled with expert precision, leading to a significant improvement in alignment with human evaluators.

Hierarchical Divide-and-Conquer for Fine-Grained Alignment in LLM-Based Medical Evaluation

Document Information Extraction (DIE) aims to extract structured information from Visually Rich Documents (VRDs). Previous full-training approaches have demonstrated strong performance but may struggle with generalization to unseen data. In contrast, training-free methods leverage powerful pre-trained models like Large Language Models (LLMs) to address various downstream tasks with only a few examples. Nonetheless, training-free methods for DIE encounter two primary challenges: (1) understanding the complex relationship between layout and textual elements in VRDs, and (2) providing accurate guidance to pre-trained models. To address these challenges, we propose Sample-centric In-context Learning (SAIL) for DIE. SAIL introduces a fine-grained entity-level textual similarity to facilitate in-depth text analysis by LLMs and incorporates layout similarity to enhance the analysis of layouts in VRDs. Additionally, SAIL formulates a unified In-Context Learning (ICL) prompt template for various sample-centric examples, enabling tailored prompts that deliver precise guidance to pre-trained models for each sample. Extensive experiments on FUNSD, CORD, and SROIE benchmarks with various base models (e.g., LLMs) indicate that our method outperforms training-free baselines, even closer to the full-training methods. The results show the superiority and generalization of our method.

SAIL: Sample-Centric In-Context Learning for Document Information Extraction

Knowledge editing aims to update outdated or incorrect knowledge in large language models (LLMs). However, current knowledge editing methods have limited scalability for lifelong editing. This study explores the fundamental reason why knowledge editing fails in lifelong editing. We begin with the closed-form solution derived from linear associative memory, which underpins state-of-the-art knowledge editing methods. We extend the solution from single editing to lifelong editing, and through rigorous mathematical derivation, identify an interference term in the final solution, suggesting that editing knowledge may impact irrelevant knowledge. Further analysis of the interference term reveals a close relationship with superposition between knowledge representations. When knowledge superposition does not exist in language models, the interference term vanishes, allowing for lossless knowledge editing. Experiments across numerous language models reveal that knowledge superposition is universal, exhibiting high kurtosis, zero mean, and heavy-tailed distributions with clear scaling laws. Ultimately, by combining theory and experiments, we demonstrate that knowledge superposition is the fundamental reason for the failure of lifelong editing. Moreover, this is the first study to investigate knowledge editing from the perspective of superposition and provides a comprehensive observation of superposition across numerous real-world language models.

Knowledge in Superposition: Unveiling the Failures of Lifelong Knowledge Editing for Large Language Models

Graph Retrieval Augmented Generation (GRAG) is a novel paradigm that takes the naive RAG system a step further by integrating graph information, such as knowledge graph (KGs), into large-scale language models (LLMs) to mitigate hallucination. However, existing GRAG still encounter limitations: 1) simple paradigms usually fail with the complex problems due to the narrow and shallow correlations capture from KGs 2) methods of strong coupling with KGs tend to be high computation cost and time consuming if the graph is dense. In this paper, we propose the Fast Think-on-Graph (FastToG), an innovative paradigm for enabling LLMs to think "community by community" within KGs. To do this, FastToG employs  community detection for deeper correlation capture and two stages community pruning - coarse and fine pruning for faster retrieval. Furthermore, we also develop two Community-to-Text methods to convert the graph structure of communities into textual form for better understanding by LLMs. Experimental results on real-world datasets demonstrate the effectiveness of FastToG, showcasing higher accuracy, faster reasoning, and better explainability compared to baseline. Our code is open-source and available at https://xxxx.

Fast Think-on-Graph: Wider, Deeper and Faster Reasoning of Large Language Model on Knowledge Graph

Complex question answering (QA) is a challenging task in artificial intelligence research which requires reasoning based on related knowledge. The retrieval-augmented generation (RAG) based on large language models (LLMs) have become one promising solution in QA. To facilitate RAG more effectively, the LLM needs to precisely evaluate knowledge required in QA. That is, first, the LLM needs to examine its knowledge boundary (what the LLM does not know) to retrieve external knowledge as supplement. Second, the LLM needs to evaluate the utility of the retrieved knowledge (whether it helps in reasoning) for robust RAG. To this end, in this paper, we propose a novel Question Answering with Knowledge Evaluation (KEQA) framework to promote the effectiveness and efficiency of RAG in QA. First, inspired by quizzes in classroom, we propose a quiz-based method to precisely examine the knowledge state of the uninterpretable LLM for QA. We ask indicative quizzes on each required knowledge, and inspect whether the LLM can consistently answer the quiz to examine its knowledge boundary. Second, we retrieve the unknown knowledge from external source, and evaluate its utility to pick the helpful ones for reasoning. We design a reasoning-based metric to evaluate utility, and construct a demonstration set in training data for reference to guide knowledge picking in inference. We conduct extensive experiments on four widely-used QA datasets, and the results demonstrate the effectiveness of the proposed method.

Explore What LLM Does Not Know in Complex Question Answering

Continual Semantic Parsing (CSP) aims to train parsers to convert natural language questions into SQL across tasks with limited annotated examples, adapting to the real-world scenario of dynamically updated databases. Previous studies mitigate this challenge by replaying historical data or employing parameter-efficient tuning (PET), but they often violate data privacy or rely on ideal continual learning settings. To address these problems, we propose a new **L**arge Language Model (LLM)-**E**nhanced **C**ontinuous **S**emantic **P**arsing method, named **LECSP**, which alleviates forgetting while encouraging generalization, without requiring real data replay or ideal settings. Specifically, it first analyzes the commonalities and differences between tasks from the SQL syntax perspective to guide LLMs in reconstructing key memories and improving memory accuracy through a calibration strategy. Then, it uses a task-aware dual-teacher distillation framework to promote the accumulation and transfer of knowledge during sequential training.
Experimental results on two CSP benchmarks show that our method significantly outperforms existing methods, even those utilizing data replay or ideal settings. Additionally, we achieve generalization performance beyond the upper limits, better adapting to unseen tasks.

Filling Memory Gaps: Enhancing Continual Semantic Parsing via SQL Syntax Variance-Guided LLMs Without Real Data Replay

Large Language Models (LLMs) excel in linguistic tasks but struggle with mathematical reasoning, particularly in non- English languages like Hindi. This research aims to en- hance the mathematical reasoning skills of smaller, resource- efficient open-source LLMs in both Hindi and English. We evaluate models like OpenHathi 7B, LLaMA-2 7B, Wizard- Math 7B, Mistral 7B, LLeMMa 7B, MAmmoTH 7B, Gemini Pro, and GPT-4 using zero-shot, few-shot chain-of-thought (CoT) methods, and supervised fine-tuning. Our approach in- corporates curriculum learning, progressively training mod- els on increasingly difficult problems, a novel Decompo- sition Strategy to simplify complex arithmetic operations, and a Structured Solution Design that divides solutions into phases. Our experiments result in notable performance en- hancements. WizardMath 7B exceeds Gemini’s accuracy on English datasets by +6% and matches Gemini’s performance on Hindi datasets. Adopting a bilingual approach that com- bines English and Hindi samples achieves results comparable to individual language models, demonstrating the capability to learn mathematical reasoning in both languages. This re- search highlights the potential for improving mathematical reasoning in open-source LLMs.

Multilingual Mathematical Reasoning: Advancing Open-Source LLMs in Hindi and English

Recent advancements in large language models (LLMs) have broadened their application scope but revealed challenges in balancing capabilities across general knowledge, coding, and mathematics. To address this, we introduce a Collaborative and Semantic Experts (CoE) approach for supervised fine-tuning (SFT), which employs a two-phase training strategy. Initially, expert training fine-tunes the feed-forward network on specialized datasets, developing distinct experts in targeted domains. Subsequently, expert leveraging synthesizes these trained experts into a structured model with semantic guidance to activate specific experts, enhancing performance and interpretability. Evaluations on comprehensive benchmarks across MMLU, HumanEval, GSM8K, MT-Bench, and AlpacaEval confirm CoE's efficacy, demonstrating improved performance and expert collaboration in diverse tasks, significantly outperforming traditional SFT methods. We commit to releasing all training data and models, anticipating that this will contribute to the advancement of open-source LLMs.

Fine-Tuning Language Models with Collaborative and Semantic Experts

Emotion Recognition in Conversations (ERC) refers to the task of automatically identifying the emotion of each utterance in conversations. Unfortunately, because the emotion of an utterance is influenced by the contextual information of the conversation, to label each utterance in ERC, the annotators are required to repeatedly read the whole conversation, making the fine-grained labeling process very costly. In this work, we suggest an efficient coarse-grained labeling pattern for ERC, where it asks annotators to assign a set of emotions for each conversation. The empirical results of our preliminary evaluations indicate the accuracy and efficiency of this conversation-level labeling pattern. Accordingly, we suggest that inducing ERC predictors with conversation-level emotion sets in a weakly-supervised learning manner can be a potential candidate for ERC, and we formally refer to this new weakly-supervised learning task as Dataless ERC (DERC). To validate its potential, we propose a simple-yet-flexible DERC framework, namely DERC with Progressive Learning (DERC-PL). We jointly update pseudo-utterance-level emotions and the ERC predictor in a self-training manner, and in particular, progressively update the ERC predictor from training subsets with lower noise densities to the ones with higher noise densities. We specify several versions of DERC-PL by applying various off-the-shelf ERC methods. Empirical results indicate that DERC-PL can be on par with existing weakly-supervised learning baselines and supervised learning ERC methods.

Utterance-level Emotion Recognition in Conversation with Conversation-level Supervision

Instruction Fine-Tuning (IFT) significantly enhances the zero-shot capabilities of pretrained Large Language Models (LLMs). While coding data is known to boost LLM reasoning abilities during pretraining, its role in activating internal reasoning capacities during IFT remains understudied. This paper investigates a key question: How does coding data impact LLMs' reasoning capacities during IFT stage? To explore this, we thoroughly examine the impact of coding data across different coding data proportions, model families, sizes, and reasoning domains, from various perspectives. Specifically, we create three IFT datasets with increasing coding data proportions, fine-tune six LLM backbones across different families and scales on these datasets, evaluate the tuned models' performance across twelve tasks in three reasoning domains, and analyze the outcomes from three broad-to-granular perspectives: overall, domain-level, and task-specific.
Our holistic analysis provides valuable insights into each perspective. First, coding data tuning enhances the overall reasoning capabilities of LLMs across different model families and scales. Moreover, while the impact of coding data varies by domain, it shows consistent trends within each domain across different model families and scales. Additionally, coding data generally provides comparable task-specific benefits across model families, with optimal proportions in IFT datasets being task-dependent.

Unveiling the Impact of Coding Data Instruction Fine-Tuning on Large Language Models Reasoning

Despite the rapid progress that existing automated feedback methods have made in correcting the output of large language models (LLMs), these methods cannot be well applied to the relation extraction (RE) task due to their designated feedback objectives and correction manner. To address this problem, we propose a novel automated feedback framework for RE, which presents a rationale supervisor to verify the rationale and provide re-selected demonstrations as feedback to correct the initial prediction. Specifically, we first design a causal intervention and observation method for to collect biased/unbiased rationales for contrastive training the rationale supervisor. Then, we present a verification-feedback-correction procedure to iteratively enhance LLMs' capability of handling the RE task. Extensive experiments prove that our proposed framework significantly outperforms existing methods.

Enhancing Relation Extraction via Supervised Rationale Verification and Feedback

Large Language Models (LLMs) have significantly advanced natural language processing (NLP), providing versatile capabilities across various applications. However, their application to complex, domain-specific tasks, such as cyber-security, often faces substantial challenges. In this study, we introduce SecKnowledge and CyberPal.AI to address these challenges and train security-expert LLMs.
SecKnowledge is a domain-knowledge-driven cyber-security instruction dataset, meticulously designed using years of accumulated expert knowledge in the domain through a multi-phase generation process. CyberPal.AI refers to a family of LLMs fine-tuned using SecKnowledge, aimed at building security-specialized LLMs capable of answering and following complex security-related instructions. 
Additionally, we introduce SecKnowledge-Eval, a comprehensive and diverse cyber-security evaluation benchmark, composed of an extensive set of cyber-security tasks we specifically developed to assess LLMs in the field of cyber-security, along with other publicly available security benchmarks.
Our results show a significant average improvement of up to 24% over the baseline models, underscoring the benefits of our expert-driven instruction dataset generation process.
These findings contribute to the advancement of AI-based cyber-security applications, paving the way for security-expert LLMs that can enhance threat-hunting and investigation processes.

Underline digital video library