Reinforcement learning with rule-based, binary-valued reward functions has been shown to significantly enhance the reasoning capabilities of large language models. These models typically rely on REINFORCE-based policy optimization methods, such as REINFORCE with baseline and group relative policy optimization (GRPO). However, a key limitation is that existing methods either ignore reward normalization or use static normalization schemes, which fail to adapt to the evolving policy during training. This mismatch can lead to unstable gradient estimates and hinder training stability. To address this issue, we propose Beta Normalization Policy Optimization (BNPO), a novel method that adaptively normalizes rewards using a Beta distribution with dynamically updated parameters. By aligning normalization with the changing policy distribution, BNPO enables more accurate, lower-variance gradient estimation and promotes stable training dynamics. We provide theoretical analysis showing that BNPO reduces variance and generalizes both REINFORCE and GRPO under binary-valued rewards. In addition, we introduce an advantage decomposition mechanism to extend BNPO to more complex reward settings. Experimental results demonstrate that BNPO achieves state-of-the-art performance among policy optimization methods on reasoning tasks.

Reinforcement learning (RL) offers a principled way to enhance the reasoning capabilities of large language models, yet its effectiveness hinges on training signals that remain informative as models evolve. In practice, RL progress often slows when task difficulty becomes poorly aligned with model capability or when training is dominated by a narrow set of recurring problem patterns.To jointly address these issues, we propose SCALER (Synthetic sCalable Adaptive Learning Environment for Reasoning), a framework that sustains effective learning signals through adaptive environment design.SCALER introduces a scalable synthesis pipeline that converts real-world programming problems into verifiable reasoning environments with controllable difficulty and unbounded instance generation, enabling RL training beyond finite datasets while preserving strong correctness guarantees. Building on this, SCALER further employs an adaptive multi-environment RL strategy that dynamically adjusts instance difficulty and curates the active set of environments to track the model’s capability frontier and maintain distributional diversity. This co-adaptation prevents reward sparsity, mitigates overfitting to narrow task patterns, and supports sustained improvement throughout training. Extensive experiments show that SCALER consistently outperforms other RL baselines across diverse reasoning benchmarks and exhibits more stable, long-horizon training dynamics.

Large Reasoning Models (LRMs) benefit substantially from training on challenging competition-level questions. However, existing automated question synthesis methods lack precise difficulty control, incur high computational costs, and struggle to generate competition-level questions at scale. In this paper, we propose CoDiQ (Controllable Difficult Question Generation), a novel framework enabling fine-grained difficulty control via test-time scaling while ensuring question solvability. Specifically, first, we identify a test-time scaling tendency (extended reasoning token budget boosts difficulty but reduces solvability) and the intrinsic properties defining the upper bound of a model’s ability to generate valid, high-difficulty questions. Then, we develop CoDiQ-Generator from Qwen3-8B, which improves the upper bound of difficult question generation, making it particularly well-suited for challenging question construction. Building on the CoDiQ framework, we build CoDiQ-Corpus (44K competition-grade question sequences). Human evaluations show these questions are significantly more challenging than LiveCodeBench/AIME with over 82% solvability. Training LRMs on CoDiQ-Corpus substantially improves reasoning performance, verifying that scaling controlled-difficulty training questions enhances reasoning capabilities. We open-source CoDiQ-Corpus, CoDiQ-Generator, and implementations to support related research.

Interactive large language model agents are often specialized to a single environment and transfer poorly across others. We study training-free consolidation of such specialists into a single generalist checkpoint. A key difficulty in this setting is that, in multi-turn agents, small merge errors on role-critical outputs can trigger long-horizon failure cascades. We therefore propose Agent-Role Merging (ARM), a neuron-level training-free framework that selects a merged backbone with an answer-free activation-overlap signal and then repairs residual gaps through conflict-aware transplantation of a small set of role-salient neurons. Our analyses further show that role-conditioned tracing yields more benchmark-specific neuron sets, enabling more localized repair. Across Qwen3-8B and Qwen2.5-7B expert pools, ARM produces the strongest balanced merged generalist among compared training-free baselines, improving both average performance and worst-suite robustness while preserving out-of-domain generalization. These results suggest that targeted neuron-level edits offer a practical route to robust multi-turn agent merging without end-to-end retraining.

Complex logical query answering (CLQA) is a challenging task that involves finding answer entities for complex logical queries over incomplete knowledge graphs (KGs). Previous research has explored the use of pre-trained knowledge graph completion (KGC) models, which can predict the missing facts in KGs, to answer complex logical queries. However, KGC models are typically evaluated using ranking evaluation metrics, which may result in values of predictions of KGC models that are not well-calibrated. In this paper, we propose a method for calibrating KGC models, namely CKGC, which enables KGC models to adapt to answering complex logical queries. Notably, CKGC is lightweight and effective. The adaptation function is simple, allowing the model to quickly converge during the adaptation process. The core concept of CKGC is to map the values of predictions of KGC models to the range [0, 1], ensuring that values associated with true facts are close to 1, while values linked to false facts are close to 0. Through experiments on three benchmark datasets, we demonstrate that our proposed calibration method can significantly boost model performance in the CLQA task. Moreover, our approach can enhance the performance of CLQA while preserving the ranking evaluation metrics of KGC models. The code is available at https://github.com/changyi7231/CKGC.