Reasoning large language models are rapidly evolving across various domains. However, their capabilities in handling complex financial tasks still require in-depth exploration. In this paper, we introduce Fin-R1, a reasoning large language model specifically designed for the financial sector. Fin-R1 is built using a two-stage architecture, leveraging a financial reasoning dataset distilled and processed based on DeepSeek-R1. Through supervised fine-tuning (SFT) and reinforcement learning (RL) training, it demonstrates performance close to DeepSeek-R1 with a parameter size of 7 billion across a range of financial reasoning tasks. It achieves the state-of-the-art (SOTA) in the FinQA and ConvFinQA tasks between those LLMs in our evaluation, surpassing larger models in other tasks as well. Fin-R1 showcases strong reasoning and decision-making capabilities, providing solutions to various problems encountered in the financial domain. Our code is available at https://github.com/SUFE-AIFLM-Lab/Fin-R1.

The success of Reinforcement Learning from Human Feedback (RLHF) critically depends on the quality of the reward model. While this quality is primarily evaluated through accuracy, it remains unclear whether accuracy fully captures what makes a reward model an effective teacher. We address this question from an optimization perspective. First, we prove that regardless of how accurate a reward model is, if it induces low reward variance, then the RLHF objective suffers from a flat landscape. Consequently, even a perfectly accurate reward model can lead to extremely slow optimization, underperforming less accurate models that induce higher reward variance. We additionally show that a reward model that works well for one language model can induce low reward variance, and thus a flat objective landscape, for another. These results establish a fundamental limitation of evaluating reward models solely based on accuracy or independently of the language model they guide. Experiments using models of up to 8B parameters corroborate our theory, demonstrating the interplay between reward variance, accuracy, and reward maximization rate. Overall, our findings highlight that beyond accuracy, a reward model needs to induce sufficient variance for efficient optimization.

Existing AI-generated dance methods primarily train on motion capture data from solo dance performances, but a critical feature of dance in nearly any genre is the interaction of two or more bodies in space. Moreover, many works at the intersection of AI and dance fail to incorporate the ideas and needs of the artists themselves into their development process, yielding models that produce far more useful insights for the AI community than for the dance community. This work addresses both needs of the field by proposing an AI method to model the complex interactions between pairs of dancers and detailing how the technical methodology can be shaped by ongoing co-creation with the artistic stakeholders who curated the movement data. Our model is a probability-and-attention-based Variational Autoencoder that generates a choreographic partner conditioned on an input dance sequence. We construct a custom loss function to enhance the smoothness and coherence of the generated choreography. Our code is open-source, and we also document strategies for other interdisciplinary research teams to facilitate collaboration and strong communication between artists and technologists.

Chain of Thought (CoT) prompting has been shown to significantly improve the performance of large language models (LLMs), particularly in arithmetic and reasoning tasks, by instructing the model to produce intermediate reasoning steps. Despite the remarkable empirical success of CoT and its theoretical advantages in enhancing expressivity, the mechanisms underlying CoT training remain largely unexplored. In this paper, we study the training dynamics of transformers over a CoT objective on an in-context weight prediction task for linear regression. We prove that while a one-layer linear transformer without CoT can only implement a single step of gradient descent (GD) and fails to recover the ground-truth weight vector, a transformer with CoT prompting can learn to perform multi-step GD autoregressively, achieving near-exact recovery. Furthermore, we show that the trained transformer effectively generalizes on the unseen data. With our technique, we also show that looped transformers significantly improve final performance compared to transformers without looping in the in-context learning of linear regression. Empirically, we demonstrate that CoT prompting yields substantial performance improvements.

Large Language Models (LLMs) have demonstrated strong performance across various general Natural Language Processing (NLP) tasks. However, their effectiveness in financial credit assessment applications remains suboptimal, primarily due to the specialized financial expertise required for these tasks. To address this limitation, we propose ZiGong, a Mistral-based model enhanced through multi-task supervised fine-tuning. To specifically combat model hallucination in financial contexts, we introduce a novel data pruning methodology. Our approach utilizes a proxy model to score training samples, subsequently combining filtered data with original datasets for model training. This data refinement strategy effectively reduces hallucinations in LLMs while maintaining reliability in downstream financial applications. Experimental results show our method significantly enhances model robustness and prediction accuracy in real-world financial scenarios.

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve over 90\% accuracy on popular benchmarks like MMLU, limiting informed measurement of state-of-the-art LLM capabilities. In response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. HLE consists of 3,000 questions across dozens of subjects, including mathematics, humanities, and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable, but cannot be quickly answered via internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a significant gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai.

--Class-incremental fault diagnosis requires a model to adapt to new fault classes while retaining previous knowledge. However, limited research exists for imbalanced and long-tailed data. Extracting discriminative features from few-shot fault data is challenging, and adding new fault classes often demands costly model retraining. T o tackle these issues, we introduce a Supervised Contrastive knowledge distiLlation for class Incremental Fault Diagnosis (SCLIFD) framework proposing supervised contrastive knowledge distillation for improved representation learning capability and less forgetting, a novel prioritized exemplar selection method for sample replay to alleviate catastrophic forgetting, and the Random Forest Classifier to address the class imbalance. Extensive experimentation on simulated and real-world industrial datasets across various imbalance ratios demonstrates the superiority of SCLIFD over existing approaches. Data-driven fault diagnosis techniques have gained significant prominence over the past two decades [1-5]. However, most of them necessitate sufficient training data to achieve reliable modeling performance[6-9]. Unfortunately, fault data is typically limited in comparison to normal data. This is because engineering equipment primarily operates under normal conditions, and the probabilities of faults vary across different working environments. Besides, fault simulation experiments are costly and inevitably deviate to some extent from real industrial environments. These possible reasons consequently contribute to class imbalance and a long-tailed distribution among different conditions [10]. The performance of the model typically suffers as it tends to prioritize the normal class, consequently neglecting fault classes or tail classes.

It is challenging to scale Ising machines for industrial-level problems due to algorithm or hardware limitations. Although higher-order Ising models provide a more compact encoding, they are, however, hard to physically implement. This work proposes a theoretical framework of a higher-order Ising simulator, IsingSim. The Ising spins and gradients in IsingSim are decoupled and self-customizable. We significantly accelerate the simulation speed via a bidirectional approach for differentiating the hyperedge functions. Our proof-of-concept implementation verifies the theoretical framework by simulating the Ising spins with exact and approximate gradients. Experiment results show that our novel framework can be a useful tool for providing design guidelines for higher-order Ising machines.

Performing complex tasks in open environments remains challenging for robots, even when using large language models (LLMs) as the core planner. Many LLM-based planners are inefficient due to their large number of parameters and prone to inaccuracies because they operate in open-loop systems. We think the reason is that only applying LLMs as planners is insufficient. In this work, we propose DaDu-E, a robust closed-loop planning framework for embodied AI robots. Specifically, DaDu-E is equipped with a relatively lightweight LLM, a set of encapsulated robot skill instructions, a robust feedback system, and memory augmentation. Together, these components enable DaDu-E to (i) actively perceive and adapt to dynamic environments, (ii) optimize computational costs while maintaining high performance, and (iii) recover from execution failures using its memory and feedback mechanisms. Extensive experiments on real-world and simulated tasks show that DaDu-E achieves task success rates comparable to embodied AI robots with larger models as planners like COME-Robot, while reducing computational requirements by $6.6 \times$. Users are encouraged to explore our system at: \url{https://rlc-lab.github.io/dadu-e/}.

Transformers have excelled in natural language modeling and one reason behind this success is their exceptional ability to combine contextual informal and global knowledge. However, the theoretical basis remains unclear. In this paper, first we introduce the Sparse Contextual Bigram (SCB), a natural extension of the classical bigram model, where the next token's generation depends on a sparse set of earlier positions determined by the last token. We then analyze the training dynamics and sample complexity of learning SCB using a one-layer linear transformer with a gradient-based algorithm. We show that when trained from scratch, the training process can be split into an initial sample-intensive stage where the correlation is boosted from zero to a nontrivial value, followed by a more sample-efficient stage of further improvement. Additionally, we prove that, provided a nontrivial correlation between the downstream and pretraining tasks, finetuning from a pretrained model allows us to bypass the initial sample-intensive stage. We also empirically demonstrate that our algorithm can outperform SGD in this setting and discuss its relationship with the usual softmax-based transformers.