Bridging Formal Language with Chain-of-Thought Reasoning to Geometry Problem Solving

Large vision language models exhibit notable limitations on Geometry Problem Solving (GPS) because of their unreliable diagram interpretation and pure natural-language reasoning. A recent line of work mitigates this by using symbolic solvers: the model directly generates a formal program that a geometry solver can execute. However, this direct program generation lacks intermediate reasoning, making the decision process opaque and prone to errors. In this work, we explore a new approach that integrates Chain-of-Thought (CoT) with formal language. The model interleaves natural language reasoning with incremental emission of solver-executable code, producing a hybrid reasoning trace in which critical derivations are expressed in formal language. To teach this behavior at scale, we combine (1) supervised fine-tuning on an 11K newly developed synthetic dataset with interleaved natural language reasoning and automatic formalization, and (2) solver-in-the-loop reinforcement learning that jointly optimizes both the CoT narrative and the resulting program through outcome-based rewards. Built on Qwen2.5-VL-7B, our new model, named GF-Reasoner, achieves up to 15% accuracy improvements on standard GPS benchmarks, surpassing both 7B-scale peers and the much larger model Qwen2.5-VL-72B. By exploiting high-order geometric knowledge and offloading symbolic computation to the solver, the generated reasoning traces are noticeably shorter and cleaner. Furthermore, we present a comprehensive analysis of method design choices (e.g., reasoning paradigms, data synthesis, training epochs, etc.), providing actionable insights for future research.

CoRT: Code-integrated Reasoning within Thinking

Large Reasoning Models (LRMs) like o1 and DeepSeek-R1 have shown remarkable progress in natural language reasoning with long chain-of-thought (CoT), yet they remain inefficient or inaccurate when handling complex mathematical operations. Addressing these limitations through computational tools (e.g., computation libraries and symbolic solvers) is promising, but it introduces a technical challenge: Code Interpreter (CI) brings external knowledge beyond the model's internal text representations, thus the direct combination is not efficient. This paper introduces CoRT, a post-training framework for teaching LRMs to leverage CI effectively and efficiently. As a first step, we address the data scarcity issue by synthesizing code-integrated reasoning data through Hint-Engineering, which strategically inserts different hints at appropriate positions to optimize LRM-CI interaction. We manually create 30 high-quality samples, upon which we post-train models ranging from 1.5B to 32B parameters, with supervised fine-tuning, rejection fine-tuning and reinforcement learning. Our experimental results demonstrate that Hint-Engineering models achieve 4\% and 8\% absolute improvements on DeepSeek-R1-Distill-Qwen-32B and DeepSeek-R1-Distill-Qwen-1.5B respectively, across five challenging mathematical reasoning datasets. Furthermore, Hint-Engineering models use about 30\% fewer tokens for the 32B model and 50\% fewer tokens for the 1.5B model compared with the natural language models. The models and code are available at https://github.com/ChengpengLi1003/CoRT.

Rethinking Data Mixture for Large Language Models: A Comprehensive Survey and New Perspectives

Training large language models with data collected from various domains can improve their performance on downstream tasks. However, given a fixed training budget, the sampling proportions of these different domains significantly impact the model's performance. How can we determine the domain weights across different data domains to train the best-performing model within constrained computational resources? In this paper, we provide a comprehensive overview of existing data mixture methods. First, we propose a fine-grained categorization of existing methods, extending beyond the previous offline and online classification. Offline methods are further grouped into heuristic-based, algorithm-based, and function fitting-based methods. For online methods, we categorize them into three groups: online min-max optimization, online mixing law, and other approaches by drawing connections with the optimization frameworks underlying offline methods. Second, we summarize the problem formulations, representative algorithms for each subtype of offline and online methods, and clarify the relationships and distinctions among them. Finally, we discuss the advantages and disadvantages of each method and highlight key challenges in the field of data mixture.

Towards Quantifying the Hessian Structure of Neural Networks

Empirical studies reported that the Hessian matrix of neural networks (NNs) exhibits a near-block-diagonal structure, yet its theoretical foundation remains unclear. In this work, we reveal two forces that shape the Hessian structure: a ``static force'' rooted in the architecture design, and a ``dynamic force'' arisen from training. We then provide a rigorous theoretical analysis of ``static force'' at random initialization. We study linear models and 1-hidden-layer networks with the mean-square (MSE) loss and the Cross-Entropy (CE) loss for classification tasks. By leveraging random matrix theory, we compare the limit distributions of the diagonal and off-diagonal Hessian blocks and find that the block-diagonal structure arises as $C \rightarrow \infty$, where $C$ denotes the number of classes. Our findings reveal that $C$ is a primary driver of the near-block-diagonal structure. These results may shed new light on the Hessian structure of large language models (LLMs), which typically operate with a large $C$ exceeding $10^4$ or $10^5$.

Factor-MCLS: Multi-agent learning system with reward factor matrix and multi-critic framework for dynamic portfolio optimization

Typical deep reinforcement learning (DRL) agents for dynamic portfolio optimization learn the factors influencing portfolio return and risk by analyzing the output values of the reward function while adjusting portfolio weights within the training environment. However, it faces a major limitation where it is difficult for investors to intervene in the training based on different levels of risk aversion towards each portfolio asset. This difficulty arises from another limitation: existing DRL agents may not develop a thorough understanding of the factors responsible for the portfolio return and risk by only learning from the output of the reward function. As a result, the strategy for determining the target portfolio weights is entirely dependent on the DRL agents themselves. To address these limitations, we propose a reward factor matrix for elucidating the return and risk of each asset in the portfolio. Additionally, we propose a novel learning system named Factor-MCLS using a multi-critic framework that facilitates learning of the reward factor matrix. In this way, our DRL-based learning system can effectively learn the factors influencing portfolio return and risk. Moreover, based on the critic networks within the multi-critic framework, we develop a risk constraint term in the training objective function of the policy function. This risk constraint term allows investors to intervene in the training of the DRL agent according to their individual levels of risk aversion towards the portfolio assets.

RealCritic: Towards Effectiveness-Driven Evaluation of Language Model Critiques

Critiques are important for enhancing the performance of Large Language Models (LLMs), enabling both self-improvement and constructive feedback for others by identifying flaws and suggesting improvements. However, evaluating the critique capabilities of LLMs presents a significant challenge due to the open-ended nature of the task. In this work, we introduce a new benchmark designed to assess the critique capabilities of LLMs. Unlike existing benchmarks, which typically function in an open-loop fashion, our approach employs a closed-loop methodology that evaluates the quality of corrections generated from critiques. Moreover, the benchmark incorporates features such as self-critique, cross-critique, and iterative critique, which are crucial for distinguishing the abilities of advanced reasoning models from more classical ones. We implement this benchmark using eight challenging reasoning tasks. We have several interesting findings. First, despite demonstrating comparable performance in direct chain-of-thought generation, classical LLMs significantly lag behind the advanced reasoning-based model o1-mini across all critique scenarios. Second, in self-critique and iterative critique settings, classical LLMs may even underperform relative to their baseline capabilities. We hope that this benchmark will serve as a valuable resource to guide future advancements. The code and data are available at \url{https://github.com/tangzhy/RealCritic}.

When GNNs meet symmetry in ILPs: an orbit-based feature augmentation approach

A common characteristic in integer linear programs (ILPs) is symmetry, allowing variables to be permuted without altering the underlying problem structure. Recently, GNNs have emerged as a promising approach for solving ILPs. However, a significant challenge arises when applying GNNs to ILPs with symmetry: classic GNN architectures struggle to differentiate between symmetric variables, which limits their predictive accuracy. In this work, we investigate the properties of permutation equivariance and invariance in GNNs, particularly in relation to the inherent symmetry of ILP formulations. We reveal that the interaction between these two factors contributes to the difficulty of distinguishing between symmetric variables. To address this challenge, we explore the potential of feature augmentation and propose several guiding principles for constructing augmented features. Building on these principles, we develop an orbit-based augmentation scheme that first groups symmetric variables and then samples augmented features for each group from a discrete uniform distribution. Empirical results demonstrate that our proposed approach significantly enhances both training efficiency and predictive performance.

A novel multi-agent dynamic portfolio optimization learning system based on hierarchical deep reinforcement learning

Deep Reinforcement Learning (DRL) has been extensively used to address portfolio optimization problems. The DRL agents acquire knowledge and make decisions through unsupervised interactions with their environment without requiring explicit knowledge of the joint dynamics of portfolio assets. Among these DRL algorithms, the combination of actor-critic algorithms and deep function approximators is the most widely used DRL algorithm. Here, we find that training the DRL agent using the actor-critic algorithm and deep function approximators may lead to scenarios where the improvement in the DRL agent's risk-adjusted profitability is not significant. We propose that such situations primarily arise from the following two problems: sparsity in positive reward and the curse of dimensionality. These limitations prevent DRL agents from comprehensively learning asset price change patterns in the training environment. As a result, the DRL agents cannot explore the dynamic portfolio optimization policy to improve the risk-adjusted profitability in the training process. To address these problems, we propose a novel multi-agent Hierarchical Deep Reinforcement Learning (HDRL) algorithmic framework in this research. Under this framework, the agents work together as a learning system for portfolio optimization. Specifically, by designing an auxiliary agent that works together with the executive agent for optimal policy exploration, the learning system can focus on exploring the policy with higher risk-adjusted return in the action space with positive return and low variance. In this way, we can overcome the issue of the curse of dimensionality and improve the training efficiency in the positive reward sparse environment.

Enabling Scalable Oversight via Self-Evolving Critic

Despite their remarkable performance, the development of Large Language Models (LLMs) faces a critical challenge in scalable oversight: providing effective feedback for tasks where human evaluation is difficult or where LLMs outperform humans. While there is growing interest in using LLMs for critique, current approaches still rely on human annotations or more powerful models, leaving the issue of enhancing critique capabilities without external supervision unresolved. We introduce SCRIT (Self-evolving CRITic), a framework that enables genuine self-evolution of critique abilities. Technically, SCRIT self-improves by training on synthetic data, generated by a contrastive-based self-critic that uses reference solutions for step-by-step critique, and a self-validation mechanism that ensures critique quality through correction outcomes. Implemented with Qwen2.5-72B-Instruct, one of the most powerful LLMs, SCRIT achieves up to a 10.3\% improvement on critique-correction and error identification benchmarks. Our analysis reveals that SCRIT's performance scales positively with data and model size, outperforms alternative approaches, and benefits critically from its self-validation component.

Second Language (Arabic) Acquisition of LLMs via Progressive Vocabulary Expansion

This paper addresses the critical need for democratizing large language models (LLM) in the Arab world, a region that has seen slower progress in developing models comparable to state-of-the-art offerings like GPT-4 or ChatGPT 3.5, due to a predominant focus on mainstream languages (e.g., English and Chinese). One practical objective for an Arabic LLM is to utilize an Arabic-specific vocabulary for the tokenizer that could speed up decoding. However, using a different vocabulary often leads to a degradation of learned knowledge since many words are initially out-of-vocabulary (OOV) when training starts. Inspired by the vocabulary learning during Second Language (Arabic) Acquisition for humans, the released AraLLaMA employs progressive vocabulary expansion, which is implemented by a modified BPE algorithm that progressively extends the Arabic subwords in its dynamic vocabulary during training, thereby balancing the OOV ratio at every stage. The ablation study demonstrated the effectiveness of Progressive Vocabulary Expansion. Moreover, AraLLaMA achieves decent performance comparable to the best Arabic LLMs across a variety of Arabic benchmarks. Models, training data, benchmarks, and codes will be all open-sourced.