论文检索

Applications · Language, Speech and Dialog

Nan Huo, Xiaohan Xu, Jinyang Li, Per Jacobsson, Shipei Lin, Bowen Qin, Binyuan Hui, Xiaolong Li, Ge Qu, Shuzheng Si 等

Large language models (LLMs) have demonstrated remarkable performance on single-turn text-to-SQL tasks, but real-world database applications predominantly require multi-turn interactions to handle ambiguous queries, execution errors, and evolving user requirements. Existing multi-turn benchmarks fall short of capturing this complexity, either by treating conversation histories as static context or by limiting evaluation to narrow, read-only (SELECT-ONLY) operations, thereby potentially failing to reflect the challenges encountered in production-grade database assistant. In this work, we introduce BIRD-INTERACT, a benchmark that restores this missing realism through: (1) a **comprehensive interaction environment** that couples each database with a hierarchical knowledge base, metadata files, and a function-driven user simulator, enabling models to solicit clarifications, retrieve knowledge, and recover from execution errors without human supervision; (2) two **evaluation settings** reflecting real-world interaction settings which contain a pre-defined conversational protocol (c-Interact) and a more open-ended agentic setting (a-Interact) in which the model autonomously decides when to query the user simulator or explore the DB environment; (3) a **challenging task suite** that covers the full CRUD spectrum for both business-intelligence and operational use cases, guarded by executable test cases. Each task features ambiguous and follow-up sub-tasks, requiring LLMs to engage in dynamic interaction. The suite is organized into two sets: a full set (BIRD-INTERACT-FULL) of 600 tasks which unfold up to 11,796 dynamic interactions for a comprehensive overview of performance and a lite set (BIRD-INTERACT-LITE) of 300 tasks, with simplified databases for detailed behavioral analysis of interactions, and fast development of methods. Our empirical results highlight the difficulty of BIRD-INTERACT: the most recent flagship model GPT-5 completes only 8.67% of tasks in the c-Interact setting and 17.00% in the a-Interact setting on the full task suite. Further analysis via memory grafting and Interaction Test-time Scaling (ITS) validates the importance of effective interaction for achieving success in dynamic text-to-SQL tasks.

Applications · Language, Speech and Dialog

Jessy Lin, Vincent-Pierre Berges, Xilun Chen, Scott Yih, Gargi Ghosh, Barlas Oguz

LLMs are known to store vast amounts of knowledge in their parametric memory. However, learning and recalling facts from this memory is known to be unreliable, depending largely on the prevalence of particular facts in the training data and other factors which are poorly understood. Practitioners are lacking tools which will allow them to ensure that the models learn a given body of knowledge reliably and consistently. To this end, we propose Active Reading: a framework where we train models to study a given set of material with self-generated learning strategies. First, we demonstrate models trained with Active Reading on expert domains absorb significantly more knowledge than vanilla finetuning and other data augmentations. We train expert 8B models that achieve 66% on a Wikipedia-grounded subset of SimpleQA (+313% relative over vanilla finetuning) and 26% on FinanceBench (+160% relative over vanilla finetuning) by applying Active Reading to the source documents for each benchmark. Finally, we show that Active Reading can be utilized at pre-training scale to build more factual models. As a demonstration of this, we release WikiExpert-8B, a Wikipedia-expert model trained on 1 trillion generated tokens, which outcompetes models with hundreds of billions of parameters on factual QA.

Applications · Language, Speech and Dialog

Philippe Laban, Hiroaki Hayashi, Yingbo Zhou, Jennifer Neville

Large Language Models (LLMs) are conversational interfaces. As such, LLMs have the potential to assist their users not only when they can fully specify the task at hand, but also to help them define, explore, and refine what they need through multi-turn conversational exchange. Although analysis of LLM conversation logs has confirmed that underspecification occurs frequently in user instructions, LLM evaluation has predominantly focused on the single-turn, fully-specified instruction setting. In this work, we perform large-scale simulation experiments to compare LLM performance in single- and multi-turn settings. Our experiments confirm that all the top open- and closed-weight LLMs we test exhibit significantly lower performance in multi-turn conversations than single-turn, with an average drop of 39% across six generation tasks. Analysis of 200,000+ simulated conversations decomposes the performance degradation into two components: a minor loss in aptitude and a significant increase in unreliability. We find that LLMs often make assumptions in early turns and prematurely attempt to generate final solutions, on which they overly rely. In simpler terms, we discover that when LLMs take a wrong turn in a conversation, they get lost and do not recover.

Applications · Language, Speech and Dialog

Tarek Naous, Philippe Laban, Wei Xu, Jennifer Neville

Conversations with LMs involve two participants: a human user leading the conversation, and an LM assistant responding to the user's request. To satisfy this specific role, LMs are post-trained to be helpful assistants -- optimized to produce exhaustive and well-structured responses, free of ambiguity and grammar errors. User utterances, on the other hand, are rarely perfected, with each user phrasing requests in unique ways, sometimes putting in partial effort at each turn and refining on the fly. To evaluate LM performance in realistic settings, prior work simulated users in multi-turn conversations, often by prompting an LM originally trained to be a helpful assistant to act as a user. However, we show that assistant LMs make for poor user simulators, with the surprising finding that better assistants yield worse simulators. Instead, we introduce purpose-built User Language Models (User LMs) - models post-trained to simulate human users in multi-turn conversations. Through various evaluations, we show how User LMs align better with human behavior and achieve better simulation robustness than existing simulation methods. When leveraging User LMs to simulate coding and math conversations, the performance of a strong assistant (GPT-4o) drops from 74.6% to 57.4%, confirming that more realistic simulation environments lead to assistant struggles as they fail to cope with the nuances of users in multi-turn setups.

Applications · Language, Speech and Dialog

Lu Ma, Hao Liang, Meiyi Qiang, Lexiang Tang, Xiaochen Ma, Zhen Wong, Junbo Niu, Chengyu Shen, Runming He, Yanhao Li 等

Recent advances in large language model (LLM) reasoning have shown that reasoning ability can emerge through reinforcement learning (RL). However, despite these successes, RL in its current form remains insufficient to induce capabilities that exceed the limitations of the base model, as it is primarily optimized based on existing knowledge of the model. To address this limitation, we employ supervised fine-tuning (SFT) to learn what RL cannot, which enables the incorporation of new knowledge and reasoning patterns by leveraging high-quality demonstration data. We analyze the training dynamics of RL and SFT for LLM reasoning and find that RL excels at improving performance on questions within the model's original capabilities, while SFT is more effective at enabling progress on questions beyond the current scope of the model. Motivated by the complementary strengths of RL and SFT, we introduce \textbf{ReLIFT} (\textbf{Re}inforcement \textbf{L}earning \textbf{I}nterleaved with Online \textbf{F}ine-\textbf{T}uning), a novel training strategy. ReLIFT employs RL for general training, but interleaves it with targeted SFT on challenging questions for which high-quality solutions are collected online. By alternating between RL and SFT, ReLIFT addresses model weaknesses as they emerge. Empirically, ReLIFT outperforms previous RLVR methods by an average of +6.7 points across a suite of six benchmarks (five math reasoning and one out-of-distribution). More importantly, ReLIFT surpasses baselines such as individual RL, individual SFT, and various hybrid approaches while reducing the required training time. These results provide compelling evidence that ReLIFT is a powerful and resource-efficient paradigm for developing capable reasoning models. The code is available at \href{https://github.com/TheRoadQaQ/ReLIFT}{here}.

Applications · Language, Speech and Dialog

Lechen Zhang, Tolga Ergen, Lajanugen Logeswaran, Moontae Lee, David Jurgens

Large Language Models (LLMs) have shown impressive capabilities in many scenarios, but their performance depends, in part, on the choice of prompt. Past research has focused on optimizing prompts specific to a task. However, much less attention has been given to optimizing the general instructions included in a prompt, known as a system prompt. To address this gap, we propose SPRIG, an edit-based genetic algorithm that iteratively constructs prompts from prespecified components to maximize the model's performance in general scenarios. We evaluate the performance of system prompts on a collection of 47 different types of tasks to ensure generalizability. Our study finds that a single optimized system prompt performs on par with task prompts optimized for each individual task. Moreover, combining system and task-level optimizations leads to further improvement, which showcases their complementary nature. Experiments also reveal that the optimized system prompts generalize effectively across model families, parameter sizes, and languages. This study provides insights into the role of system-level instructions in maximizing LLM potential.

Applications · Language, Speech and Dialog

Clara Meister

This post is my attempt to write down the UnigramLM tokenization algorithm cleanly and explicitly because, well, I still haven't found such a derivation and I think understanding the theory behind the method could help us make it better. I'll formalize the generative model around which the algorithm is based, derive the EM updates, explain why pruning is needed (and how it's done), and point out the spots where the practical implementation defined by the SentencePiece library diverges from the pretty mathematical models.

Applications · Language, Speech and Dialog

Pierre-Carl Langlais, Pavel Chizhov, Catherine Arnett, Carlos Hinostroza, Mattia Nee, Eliot Jones, Irène Girard, David Mach, Anastasia Stasenko, Ivan Yamshchikov

Large Language Models (LLMs) are pre-trained on large data from different sources and domains. These datasets often contain trillions of tokens, including large portions of copyrighted or proprietary content, which raises questions about the legal use of such models. This underscores the need for truly open pre-training data that complies with data security regulations. In this paper, we introduce Common Corpus, the largest open dataset for LLM pre-training. The data assembled in Common Corpus are either uncopyrighted or under permissive licenses and amount to about two trillion tokens. The dataset contains a wide variety of languages, ranging from the high-resource European languages to some low-resource languages rarely represented in pre-training datasets. In addition, it includes a large amount of code data. The diversity of data sources in terms of covered domains and time periods opens up the paths for both research and entrepreneurial needs in diverse areas of knowledge. In this paper, we present the detailed provenance of data assembling and the details of dataset filtering and curation. We train two small language models on Common Corpus and find that they perform comparably to other models of their size, indicating that our dataset is suitable for multilingual pretraining. Common Corpus represents a key contribution to the ecosystem for open science research on Large Language Models.

Applications · Language, Speech and Dialog

Yuxiao Qu, Anikait Singh, Yoonho Lee, Amrith Setlur, Russ Salakhutdinov, Chelsea Finn, Aviral Kumar

Reasoning requires going beyond pattern matching or memorization of solutions to identify and implement algorithmic procedures that can be used to deduce answers to hard problems. Doing so requires reusing primitives, intermediate results, or procedures across multiple problems. While RL post-training on long chains of thought ultimately aims to uncover this kind of algorithmic behavior, the depth-first and brute-force nature of reasoning traces learned by these models suggests that this is far from a fulfilled promise. To address more effective reasoning, we introduce reasoning abstractions: concise natural language descriptions of procedural and factual knowledge that guide the model toward learning successful reasoning. We train models to be capable of proposing several useful abstractions given a problem, followed by RL training that incentivizes building a solution while using the information provided by these abstractions. This results in a two-player RL training paradigm, abbreviated as RLAD, that jointly trains an abstraction generator and an abstraction-conditioned solution generator. This setup effectively enables structured exploration, decouples learning signals of abstraction proposal and solution generation, and improves generalization to harder problems. We also show that spending more test-time compute into generating abstractions is more beneficial for performance than generating more solutions at large inference-time budgets, illustrating the role of abstractions in guiding global exploration.

Applications · Language, Speech and Dialog

Anikait Singh, Sheryl Hsu, Kyle Hsu, Eric Mitchell, Stefano Ermon, Tatsunori Hashimoto, Archit Sharma, Chelsea Finn

Effective personalization of LLMs is critical for a broad range of user-interfacing applications such as virtual assistants and content curation. Inspired by the strong in-context capabilities of LLMs, we propose few-shot preference optimization (FSPO), an algorithm for LLM personalization that reframes reward modeling as a meta-learning problem. Under FSPO, an LLM learns to quickly infer a personalized reward function for a user via a few labeled preferences. FSPO also utilizes user description rationalization (RAT) to encourage better reward modeling and instruction following, recovering performance with the oracle user description. Since real-world preference data is challenging to collect at scale, we propose careful design choices to construct synthetic preference datasets for personalization, generating over 1M synthetic personalized preferences using publicly available LLMs. To successfully transfer from synthetic data to real users, we find it crucial for the data to exhibit both high diversity and coherent, self-consistent structure. We evaluate FSPO on personalized open-ended generation for up to 1,500 synthetic users across three domains: movie reviews, education, and open-ended question answering. We also run a controlled human study. Overall, FSPO achieves an 87% Alpaca Eval win-rate in generating responses that are personalized to synthetic users and a 70% win-rate with real human users in open-ended question answering.

Applications · Language, Speech and Dialog

Xiao Yu, Baolin Peng, Michel Galley, Hao Cheng, Qianhui Wu, Janardhan Kulkarni, Suman Nath, Zhou Yu, Jianfeng Gao

Reasoning models have recently shown remarkable progress in domains such as math and coding. However, their expert-level abilities in math and coding contrast sharply with their performance in long-horizon, interactive tasks such as web navigation and computer/phone-use. Inspired by literature on human cognition, we argue that current AI agents need ``vicarious trial and error'' - the capacity to mentally simulate alternative futures before acting - in order to enhance their understanding and performance in complex interactive environments. We introduce Dyna-Mind, a two-stage training framework that explicitly teaches (V)LM agents to integrate such simulation into their reasoning. In stage 1, we introduce Reasoning with Simulations (ReSim), which trains the agent to generate structured reasoning traces from expanded search trees built from real experience gathered through environment interactions. ReSim thus grounds the agent's reasoning in faithful world dynamics and equips it with the ability to anticipate future states in its reasoning. In stage 2, we propose Dyna-GRPO, an online reinforcement learning method to further strengthen the agent's simulation and decision-making ability by using both outcome rewards and intermediate states as feedback from real rollouts. Experiments on two synthetic benchmarks (Sokoban and ALFWorld) and one realistic benchmark (AndroidWorld) demonstrate that (1) ReSim effectively infuses simulation ability into AI agents, and (2) Dyna-GRPO leverages outcome and interaction-level signals to learn better policies for long-horizon, planning-intensive tasks. Together, these results highlight the central role of simulation in enabling AI agents to reason, plan, and act more effectively in the ever more challenging environments.

Applications · Language, Speech and Dialog

Jinchuan Tian, Sang-gil Lee, Zhifeng Kong, Sreyan Ghosh, Arushi Goel, Chao-Han Huck Yang, Wenliang Dai, Zihan Liu, Hanrong Ye, Shinji Watanabe 等

Recent advances in the audio language modeling (ALM) domain tackle audio understanding and text-to-audio generation as separate tasks. Very few studies attempt to unify these tasks -- an essential step toward advanced multimodal reasoning. This paper introduces Unified Audio Language Model (UALM), which aims to unify audio understanding, text-to-audio generation, and multimodal reasoning in a single model. To achieve this goal, we first present UALM-Gen, a text-to-audio language model that directly predicts audio tokens and is comparable to state-of-the-art diffusion-based models. We then demonstrate, using proper data blending, training recipes, and inference techniques, that our single UALM model matches the quality of state-of-the-art specialized models in audio understanding, text-to-audio generation, and text reasoning. Furthermore, we present UALM-Reason, a multimodal reasoning model that utilizes both text and audio in the intermediate thinking steps to facilitate complex generation tasks. To our knowledge, this is the first demonstration in audio research of cross-modal generative reasoning, with its effectiveness confirmed by subjective evaluations.

Applications · Language, Speech and Dialog

Jinchuan Tian, Sang-gil Lee, Zhifeng Kong, Sreyan Ghosh, Arushi Goel, Chao-Han Huck Yang, Wenliang Dai, Zihan Liu, Hanrong Ye, Shinji Watanabe 等

Recent advances in the audio language modeling (ALM) domain tackle audio understanding and text-to-audio generation as separate tasks. Very few studies attempt to unify these tasks -- an essential step toward advanced multimodal reasoning. This paper introduces Unified Audio Language Model (UALM), which aims to unify audio understanding, text-to-audio generation, and multimodal reasoning in a single model. To achieve this goal, we first present UALM-Gen, a text-to-audio language model that directly predicts audio tokens and is comparable to state-of-the-art diffusion-based models. We then demonstrate, using proper data blending, training recipes, and inference techniques, that our single UALM model matches the quality of state-of-the-art specialized models in audio understanding, text-to-audio generation, and text reasoning. Furthermore, we present UALM-Reason, a multimodal reasoning model that utilizes both text and audio in the intermediate thinking steps to facilitate complex generation tasks. To our knowledge, this is the first demonstration in audio research of cross-modal generative reasoning, with its effectiveness confirmed by subjective evaluations.

Applications · Language, Speech and Dialog

Xu Wang, Chenkai Xu, Yijie Jin, Jiachun Jin, Hao Zhang, Kai Yu, Zhijie Deng

Diffusion Large Language Models (dLLMs) have emerged as a promising alternative to autoregressive (AR) LLMs for text generation, with the potential to decode multiple tokens in a single iteration. However, none of the existing open-source dLLMs have achieved superior inference speed over AR LLMs of similar size. This paper breaks this barrier based on a simple and effective strategy named discrete diffusion forcing (D2F). D2F equips dLLMs with two key capabilities: (1) block-wise autoregressive generation to enable KV cache utilization; (2) prediction of following tokens without requiring completion of prior blocks for inter-block parallel decoding. In this way, the vanilla dLLMs are refurbished into an AR-diffusion hybrid paradigm for efficient inference. D2F can be implemented with an asymmetric distillation process based on pre-trained dLLMs to achieve rapid convergence.We further propose a pipelined parallel decoding algorithm, which enables a trade-off between efficiency and efficacy. Empirically, D2F dLLMs achieve more than $\mathbf{2.5\times}$ inference speed than LLaMA3 and Qwen2.5 on GSM8K. Compared to the vanilla dLLMs like LLaDA and Dream, the acceleration can be more than $\mathbf{50\times}$ while maintaining comparable output quality.

Applications · Language, Speech and Dialog

Zhaoning Yu, Zhaolun Su, Leitian Tao, Haozhu Wang, Aashu Singh, Hanchao Yu, Jianyu Wang, Hongyang Gao, Weizhe Yuan, Jason E Weston 等

Reinforcement learning with human-annotated data has boosted chain-of-thought reasoning in large reasoning models, but these gains come at high costs in labeled data while faltering on harder tasks. A natural next step is experience-driven learning, where models improve without curated labels by adapting to unlabeled data. We introduce REinforcement learning with Self-resTRAINt training (RESTRAIN), a self-penalizing RL framework that converts the absence of gold labels into a useful learning signal. Instead of overcommitting to spurious majority votes, RESTRAIN exploits signals from the model’s entire answer distribution: penalizing overconfident rollouts and low-consistency examples while preserving promising reasoning chains. This self-penalization mechanism integrates seamlessly into policy optimization methods such as GRPO, enabling continual self-improvement without supervision. On challenging reasoning benchmarks, RESTRAIN delivers large gains using only unlabeled data. With Qwen3-4B-Base and OctoThinker Hybrid-8B-Base, it boosts Pass@1 by up to +140.7% on AIME25, +36.2% on MMLU STEM, and +19.6% on GPQA-Diamond, nearly matching gold-label training while using no gold labels. These results demonstrate that RESTRAIN establishes a scalable path toward stronger reasoning without gold labels.

Applications · Language, Speech and Dialog

Sungho Park, Jueun Kim, Wook-Shin Han

Real-world Table–Text question answering (QA) tasks require models that can reason across long text and source tables, traversing multiple hops and executing complex operations such as aggregation. Yet existing benchmarks are small, manually curated—and therefore error-prone—and contain shallow questions that seldom demand more than two hops or invoke aggregations, grouping, or other advanced analytical operations expressible in natural-language queries. We present SPARTA, an end-to-end construction framework that automatically generates large-scale Table–Text QA benchmarks with lightweight human validation, requiring only one quarter of the annotation time of HybridQA. The framework first constructs a reference fact database by enriching each source table with grounding tables whose tuples are atomic facts automatically extracted from the accompanying unstructured passages, then synthesizes nested queries whose number of nested predicates matches the desired hop count. To ensure that every SQL statement is executable and that its verbalization yields a fluent, human-sounding question, we propose two novel techniques: provenance-based refinement, which rewrites any syntactically valid query that returns a non-empty result, and realistic-structure enforcement, which confines generation to post-order traversals of the query graph. The resulting pipeline produces thousands of high-fidelity question–answer pairs covering aggregations, grouping, and deep multi-hop reasoning across text and tables. On SPARTA, state-of-the-art models that reach over 70 F1 on HybridQA or over 50 F1 on OTT-QA drop by more than 30 F1 points, exposing fundamental weaknesses in current cross-modal reasoning. We will release the benchmark, construction code, and baseline results to spur progress toward robust, realistic Table–Text QA models.

Applications · Language, Speech and Dialog

Jingcong Liang, Shijun Wan, Xuehai Wu, Yitong Li, Qianglong Chen, Duyu Tang, Siyuan Wang, zhongyu wei

Large Reasoning Models (LRMs) have demonstrated impressive performance on complex tasks, including logical puzzle games that require deriving solutions satisfying all constraints. However, whether they can flexibly apply appropriate rules to varying conditions, particularly when faced with non-canonical game variants, remains an open question. Existing corpora focus on popular puzzles like 9x9 Sudoku, risking overfitting to canonical formats and memorization of solution patterns, which can mask deficiencies in understanding novel rules or adapting strategies to new variants. To address this, we introduce **HardcoreLogic**, a challenging benchmark of over 5,000 puzzles across 10 games, designed to test the robustness of LRMs on the "long-tail" of logical games. HardcoreLogic systematically transforms canonical puzzles through three dimensions: **Increased Complexity (IC)**, **Uncommon Elements (UE)**, and **Unsolvable Puzzles (UP)**, reducing reliance on shortcut memorization. Evaluations on a diverse set of LRMs reveal significant performance drops, even for models achieving top scores on existing benchmarks, indicating heavy reliance on memorized stereotypes. While increased complexity is the dominant source of difficulty, models also struggle with subtle rule variations that do not necessarily increase puzzle difficulty. Our systematic error analysis on solvable and unsolvable puzzles further highlights gaps in genuine reasoning. Overall, HardcoreLogic exposes the limitations of current LRMs and establishes a benchmark for advancing high-level logical reasoning.

Applications · Language, Speech and Dialog

Mengqi Zhang, Zisheng Zhou, Xiaotian Ye, Qiang Liu, Zhaochun Ren, Zhumin Chen, Pengjie Ren

Knowledge Editing has emerged as a promising solution for efficiently updating embedded knowledge in large language models (LLMs). While existing approaches demonstrate effectiveness in integrating new knowledge and preserving the original capabilities of LLMs, they fail to maintain fine-grained irrelevant knowledge, namely facts that share the same subject as edited knowledge but differ in relation and object. This challenge arises because subject representations inherently encode multiple attributes, causing the target and fine-grained irrelevant knowledge to become entangled in the representation space, and thus vulnerable to unintended alterations during editing. To address this, we propose DiKE, a novel approach that Disentangles Knowledge representations for LLM Editing (DiKE). DiKE consists of two key components: a Knowledge Representation Disentanglement (KRD) module that decomposes the subject representation into target-knowledge-related and -unrelated components, and a Disentanglementbased Knowledge Edit (DKE) module that updates only the target-related component while explicitly preserving the unrelated one. We further derive a closedform, rank-one parameter update based on matrix theory to enable efficient and minimally invasive edits. To rigorously evaluate fine-grained irrelevant knowledge preservation, we construct FINE-KED, a new benchmark comprising fine-grained irrelevant knowledge at different levels of relational similarity to the edited knowledge. Extensive experiments across multiple LLMs demonstrate that DiKE substantially improves fine-grained irrelevant knowledge preservation while maintaining competitive general editing performance.

Applications · Language, Speech and Dialog

Yanqi Dai, Yuxiang Ji, Xiao Zhang, Yong Wang, Xiangxiang Chu, Zhiwu Lu

Reinforcement Learning with Verifiable Rewards (RLVR) offers a robust mechanism for enhancing mathematical reasoning in large models. However, we identify a systematic lack of emphasis on more challenging questions in existing methods from both algorithmic and data perspectives, despite their importance for refining underdeveloped capabilities. Algorithmically, widely used Group Relative Policy Optimization (GRPO) suffers from an implicit imbalance where the magnitude of policy updates is lower for harder questions. Data-wise, augmentation approaches primarily rephrase questions to enhance diversity without systematically increasing intrinsic difficulty. To address these issues, we propose a two-dual MathForge framework to improve mathematical reasoning by targeting harder questions from both perspectives, which comprises a Difficulty-Aware Group Policy Optimization (DGPO) algorithm and a Multi-Aspect Question Reformulation (MQR) strategy. Specifically, DGPO first rectifies the implicit imbalance in GRPO via difficulty-balanced group advantage estimation, and further prioritizes harder questions by difficulty-aware question-level weighting. Meanwhile, MQR reformulates questions across multiple aspects to increase difficulty while maintaining the original gold answer. Overall, MathForge forms a synergistic loop: MQR expands the data frontier, and DGPO effectively learns from the augmented data. Extensive experiments show that MathForge significantly outperforms existing methods on various mathematical reasoning tasks. The code and augmented data are all available at https://github.com/AMAP-ML/MathForge.

Applications · Language, Speech and Dialog

Shenzhi Yang, Guangcheng Zhu, Haobo Wang, Xing Zheng, Yingfan MA, Zhongqi Chen, Bowen Song, Weiqiang Wang, Junbo Zhao, Gang Chen

Reinforcement learning with verifiable rewards (RLVR) has proven effective in training large reasoning models (LRMs) by leveraging answer-verifiable signals to guide policy optimization, which, however, suffers from high annotation costs. To alleviate this problem, recent work has explored unsupervised RLVR methods that derive rewards solely from the model’s internal consistency, such as through entropy and majority voting. While seemingly promising, these methods often suffer from model collapse in the later stages of training, which may arise from the reinforcement of incorrect reasoning patterns in the absence of external supervision. In this work, we investigate a novel semi-supervised RLVR paradigm that utilizes a small labeled set to **guide** RLVR training on unlabeled samples. Our key insight is that supervised rewards are essential for stabilizing consistency-based training on unlabeled samples, ensuring that only reasoning patterns verified on labeled instances are incorporated into RL training. Technically, we propose an effective policy optimization algorithm TraPO that filters out reliable unlabeled samples by matching their learning trajectory similarity to labeled ones. Building on this, TraPO achieves remarkable data efficiency and strong generalization on nine advanced benchmarks. With only 1K labeled and 3K unlabeled samples, TraPO reaches 42.6% average accuracy, surpassing the best unsupervised method trained on 45K unlabeled samples (38.3%). Notably, when using 4K labeled and 12K unlabeled samples, TraPO even **outperforms the fully supervised model** trained on the full 45K labeled samples on all benchmarks, while using only **10%** of the labeled data. The code is available via https://github.com/ShenzhiYang2000/TRAPO.