Drafting patent claims is time-intensive, costly, and requires professional skill. Therefore, researchers have investigated large language models (LLMs) to assist inventors in writing claims. However, existing work has largely relied on datasets from the United States Patent and Trademark Office (USPTO). To enlarge research scope regarding various jurisdictions, drafting conventions, and legal standards, we introduce EPD, a European patent dataset. EPD presents rich textual data and structured metadata to support multiple patent-related tasks, including claim generation. This dataset enriches the field in three critical aspects: (1) Jurisdictional diversity: Patents from different offices vary in legal and drafting conventions. EPD fills a critical gap by providing a benchmark for European patents to enable more comprehensive evaluation. (2) Quality improvement: EPD offers high-quality granted patents with finalized and legally approved texts, whereas others consist of patent applications that are unexamined or provisional. Experiments show that LLMs fine-tuned on EPD significantly outperform those trained on previous datasets and even GPT-4o in claim quality and cross-domain generalization. (3) Real-world simulation: We propose a difficult subset of EPD to better reflect real-world challenges of claim generation. Results reveal that all tested LLMs perform substantially worse on these challenging samples, which highlights the need for future research.

Foundation models have shown remarkable success across scientific domains, yet their impact in chemistry remains limited due to the absence of diverse, large-scale, high-quality datasets that reflect the field's multifaceted nature. We present the ChemPile, an open dataset containing over 75 billion tokens of curated chemical data, specifically built for training and evaluating general-purpose models in the chemical sciences. The dataset mirrors the human learning journey through chemistry -- from educational foundations to specialized expertise -- spanning multiple modalities and content types including structured data in diverse chemical representations (SMILES, SELFIES, IUPAC names, InChI, molecular renderings), scientific and educational text, executable code, and chemical images. ChemPile integrates foundational knowledge (textbooks, lecture notes), specialized expertise (scientific articles and language-interfaced data), visual understanding (molecular structures, diagrams), and advanced reasoning (problem-solving traces and code) -- mirroring how human chemists develop expertise through diverse learning materials and experiences. Constructed through hundreds of hours of expert curation, the ChemPile captures both foundational concepts and domain-specific complexity. We provide standardized training, validation, and test splits, enabling robust benchmarking. ChemPile is openly released via HuggingFace with a consistent API, permissive license, and detailed documentation. We hope the ChemPile will serve as a catalyst for chemical AI, enabling the development of the next generation of chemical foundation models.

Large language models (LLMs) increasingly power mental-health chatbots, yet the field still lacks a scalable, theory-grounded way to decide which model is most effective to deploy. We present ESC-Judge, the first end-to-end evaluation framework that (i) grounds head-to-head comparisons of emotional-support LLMs in Clara Hill's established Exploration-Insight-Action counseling model, providing a structured and interpretable view of performance, and (ii) fully automates the evaluation pipeline at scale. ESC-Judge operates in three stages: first, it synthesizes realistic help-seeker roles by sampling empirically salient attributes such as stressors, personality, and life history; second, it has two candidate support agents conduct separate sessions with the same role, isolating model-specific strategies; and third, it asks a specialized judge LLM to express pairwise preferences across rubric-anchored skills that span the Exploration, Insight, and Action spectrum. In our study, ESC-Judge matched PhD-level annotators on 85 percent of Exploration, 83 percent of Insight, and 86 percent of Action decisions, demonstrating human-level reliability at a fraction of the cost. All code, prompts, synthetic roles, transcripts, and judgment scripts are released to promote transparent progress in emotionally supportive AI.

Outside of transfer learning settings, reinforcement learning agents start their learning process from a clean slate. As a result, such agents have to go through a slow process to learn even the most obvious skills required to solve a problem. In this paper, we present InnateCoder, a system that leverages human knowledge encoded in foundation models to provide programmatic policies that encode "innate skills" in the form of temporally extended actions, or options. In contrast to existing approaches to learning options, InnateCoder learns them from the general human knowledge encoded in foundation models in a zero-shot setting, and not from the knowledge the agent gains by interacting with the environment. Then, InnateCoder searches for a programmatic policy by combining the programs encoding these options into larger and more complex programs. We hypothesized that InnateCoder's way of learning and using options could improve the sampling efficiency of current methods for learning programmatic policies. Empirical results in MicroRTS and Karel the Robot support our hypothesis, since they show that InnateCoder is more sample efficient than versions of the system that do not use options or learn them from experience.

The impressive ability of large language models to generate natural text across various tasks has led to critical challenges in authorship authentication. Although numerous detection methods have been developed to differentiate between machine-generated texts (MGT) and human-generated texts (HGT), the explainability of these methods remains a significant gap. Traditional explainability techniques often fall short in capturing the complex word relationships that distinguish HGT from MGT. To address this limitation, we present LM$^2$otifs, a novel explainable framework for MGT detection. Inspired by probabilistic graphical models, we provide a theoretical rationale for the effectiveness. LM$^2$otifs utilizes eXplainable Graph Neural Networks to achieve both accurate detection and interpretability. The LM$^2$otifs pipeline operates in three key stages: first, it transforms text into graphs based on word co-occurrence to represent lexical dependencies; second, graph neural networks are used for prediction; and third, a post-hoc explainability method extracts interpretable motifs, offering multi-level explanations from individual words to sentence structures. Extensive experiments on multiple benchmark datasets demonstrate the comparable performance of LM$^2$otifs. The empirical evaluation of the extracted explainable motifs confirms their effectiveness in differentiating HGT and MGT. Furthermore, qualitative analysis reveals distinct and visible linguistic fingerprints characteristic of MGT.

Recent advances in rule-based reinforcement learning (RL) have significantly improved the reasoning capability of language models (LMs) with rule-based rewards. However, existing RL methods -- such as GRPO, REINFORCE++, and RLOO -- often suffer from training instability, where large policy updates and improper clipping can lead to training collapse. To address this issue, we propose Clipped Policy Gradient Optimization with Policy Drift (CPGD), a novel algorithm designed to stabilize policy learning in LMs. CPGD introduces a policy drift constraint based on KL divergence to dynamically regularize policy updates, and leverages a clip mechanism on the logarithm of the ratio to prevent excessive policy updates. We provide theoretical justification for CPGD and demonstrate through empirical analysis that it mitigates the instability observed in prior approaches. Furthermore, we show that CPGD significantly improves performance while maintaining training stability. Our implementation balances theoretical rigor with practical usability, offering a robust alternative for RL in the post-training of LMs. We release our code at https://github.com/ModalMinds/MM-EUREKA.

--Recently, large language models (LLMs) have demonstrated impressive performance in Knowledge Graph Question Answering (KGQA) tasks, which aim to find answers based on knowledge graphs (KGs) for natural language questions. Existing LLMs-based KGQA methods typically follow the Graph Retrieval-Augmented Generation (GraphRAG) paradigm, which first retrieves reasoning paths from the large KGs, and then generates the answers based on them. However, these methods emphasize the exploration of new optimal reasoning paths in KGs while ignoring the exploitation of historical reasoning paths, which may lead to sub-optimal reasoning paths. Additionally, the complex semantics contained in questions may lead to the retrieval of inaccurate reasoning paths. T o address these issues, this paper proposes a novel and training-free framework for KGQA tasks called Reward-guided Tree Search on Graph (RTSoG). RTSoG decomposes an original question into a series of simpler and well-defined sub-questions to handle the complex semantics. Then, a Self-Critic Monte Carlo Tree Search (SC-MCTS) guided by a reward model is introduced to iteratively retrieve weighted reasoning paths as contextual knowledge. Finally, it stacks the weighted reasoning paths according to their weights to generate the final answers. Extensive experiments on four datasets demonstrate the effectiveness of RTSoG. Notably, it achieves 8.7% and 7.0% performance improvement over the state-of-the-art method on the GrailQA and the WebQSP respectively. Knowledge Graph Question Answering (KGQA) has drawn much attention in recent years.

Abstract--Distantly supervised named entity recognition (DS-NER) has emerged as a cheap and convenient alternative to traditional human annotation methods, enabling the automatic generation of training data by aligning text with external resources. Despite the many efforts in noise measurement methods, few works focus on the latent noise distribution between different distant annotation methods. In this work, we explore the effectiveness and robustness of DS-NER by two aspects: (1) distant annotation techniques, which encompasses both traditional rule-based methods and the innovative large language model supervision approach, and (2) noise assessment, for which we introduce a novel framework. This framework addresses the challenges by distinctly categorizing them into the unlabeled-entity problem (UEP) and the noisy-entity problem (NEP), subsequently providing specialized solutions for each. Our proposed method achieves significant improvements on eight real-world distant supervision datasets originating from three different data sources and involving four distinct annotation techniques, confirming its superiority over current state-of-the-art methods. Index Terms --Distantly supervised learning, Named entity recognition, Noise measurement. With the prosperous development of neural techniques [3]-[5], the past decade has witnessed the tremendous success of the NER tasks. T o achieve a high performance, the need for massive high-quality data is inevitable, whether for the previous fully supervised methods or the recent fine-tuned T ask-Specific Large Language Models like UniNER [6]. However, obtaining massive data with high-quality annotations is either inapplicable or unaffordable. Thus, NER under distant supervision (DS) has been a popular alternative [7], [8]. Distantly supervised NER first aims to annotate an unlabeled dataset utilizing external resources, such as knowledge bases and dictionaries, then train a model on the distantly annotated data. Recently, LLMs have been demonstrated to be proficient annotators for numerous NLP tasks [9]. However, regardless of the distantly supervised method employed, whether traditional rule-based annotating methods like KB-Matching [8] and Dict-Matching [10] or LLMbased annotating methods, considerable label noise is injected into the datasets. Consequently, devising a strategy to train a high-performance NER model on a noisy NER dataset becomes critically important. W e begin with a preliminary study to compare and assess the annotation capabilities of different annotation methods. Y uyang Ding and Dan Qiao contribute equally. Xiaofang Zhou is with the Hong Kong University of Science and T echnol-ogy, Hong Kong, China.

Rotary Position Embedding (RoPE) is an efficient position encoding approach and is widely utilized in numerous large language models (LLMs). Recently, a lot of methods have been put forward to further expand the context window based on RoPE. The core concept of those methods is to predefine or search for a set of factors to rescale the base frequencies of RoPE. Nevertheless, it is quite a challenge for existing methods to predefine an optimal factor due to the exponential search space. In view of this, we introduce PSC (Phase Shift Calibration), a small module for calibrating the frequencies predefined by existing methods. With the employment of PSC, we demonstrate that many existing methods can be further enhanced, like PI, YaRN, and LongRoPE. We conducted extensive experiments across multiple models and tasks. The results demonstrate that (1) when PSC is enabled, the comparative reductions in perplexity increase as the context window size is varied from 16k, to 32k, and up to 64k. (2) Our approach is broadly applicable and exhibits robustness across a variety of models and tasks. The code can be found at https://github.com/WNQzhu/PSC.

Hierarchical data arise in countless domains, from biological taxonomies and organizational charts to legal codes and knowledge graphs. Residual Quantization (RQ) is widely used to generate discrete, multitoken representations for such data by iteratively quantizing residuals in a multilevel codebook. However, its reliance on Euclidean geometry can introduce fundamental mismatches that hinder modeling of hierarchical branching, necessary for faithful representation of hierarchical data. In this work, we propose Hyperbolic Residual Quantization (HRQ), which embeds data natively in a hyperbolic manifold and performs residual quantization using hyperbolic operations and distance metrics. By adapting the embedding network, residual computation, and distance metric to hyperbolic geometry, HRQ imparts an inductive bias that aligns naturally with hierarchical branching. We claim that HRQ in comparison to RQ can generate more useful for downstream tasks discrete hierarchical representations for data with latent hierarchies. We evaluate HRQ on two tasks: supervised hierarchy modeling using WordNet hypernym trees, where the model is supervised to learn the latent hierarchy - and hierarchy discovery, where, while latent hierarchy exists in the data, the model is not directly trained or evaluated on a task related to the hierarchy. Across both scenarios, HRQ hierarchical tokens yield better performance on downstream tasks compared to Euclidean RQ with gains of up to $20\%$ for the hierarchy modeling task. Our results demonstrate that integrating hyperbolic geometry into discrete representation learning substantially enhances the ability to capture latent hierarchies.