On-policy distillation (OPD) trains a student on its own trajectories with token-level teacher feedback and often outperforms off-policy distillation and reinforcement learning with verifiable rewards. However, we find that its standard advantage-weighted policy gradient suffers from three structural weaknesses, including high-variance updates, vanishing gradients in zero-advantage regions, and exploration bottlenecks for models with limited capability. We therefore propose asymmetric on-policy distillation (AOPD), which replaces ineffective policy gradient with localized divergence minimization in non-positive advantage regions while preserving positive reinforcement learning. Experiments on mathematical reasoning benchmarks show that AOPD consistently outperforms standard OPD, with average gains of 4.09 / 8.34 under strong / weak initialization, respectively. AOPD also maintains higher policy entropy during training and better capability retention during sequential tool-use adaptation. Our source code is available at [repository](https://github.com/stargazer319/AOPD).

On-policy reinforcement learning has become the dominant paradigm for reasoning alignment in large language models, yet its sparse, outcome-level rewards make token-level credit assignment notoriously difficult. On-Policy Distillation (OPD) alleviates this by introducing dense, token-level KL supervision from a teacher model, but typically applies this supervision uniformly across all rollouts, ignoring fundamental differences in signal quality. We propose **S**ignal-**C**alibrated **O**n-**P**olicy Distillation **E**nhancement (**SCOPE**), a dual-path adaptive training framework that routes on-policy rollouts by correctness into two complementary supervision paths. For incorrect trajectories, SCOPE performs teacher-perplexity-weighted KL distillation to prioritize instances where the teacher demonstrates genuine corrective capability, while down-weighting unreliable guidance. For correct trajectories, it applies student-perplexity-weighted Maximum Likelihood Estimation to concentrate reinforcement on low-confidence samples at the capability boundary rather than over-reinforcing already mastered ones. Both paths employ a group-level normalization to adaptively calibrate weight distributions, accounting for the intrinsic difficulty variance across prompts. Extensive experiments on six mathematical reasoning benchmarks show that SCOPE achieves average relative improvements of 11.42% in Avg@32 and 7.30% in Pass@32 over competitive baselines, with extended experiments demonstrating its broader applicability. Code is available at [https://github.com/machine981/SCOPE](https://github.com/machine981/SCOPE).

Reward models (RMs) are central to reinforcement learning from human feedback (RLHF), providing the critical supervision signals that align large language models (LLMs) with human preferences.Generative reward models (GRMs) provide greater interpretability than traditional scalar RMs, but they come with a critical trade-off: pairwise methods are hindered by a training-inference mismatch, while pointwise methods require expensive absolute annotations.To bridge this gap, we propose the Preference-aware Task-adaptive Reward Model (PaTaRM).Unlike prior approaches, PaTaRM enables robust pointwise training using readily available pairwise data via a novel Preference-Aware Reward (PAR) mechanism, eliminating the need for explicit rating labels. Furthermore, it incorporates a task-adaptive rubric system that dynamically generates instance-specific criteria for precise evaluation.Extensive experiments demonstrate that PaTaRM achieves an average relative improvement of 8.7% over the corresponding base models on RewardBench and RMBench across the Qwen3-8B and Qwen3-14B backbones.Crucially, when used as a reward model for downstream RLHF, it yields an average relative improvement of 13.6% over the corresponding base policies on IFEval and InfoBench, validating its effectiveness for policy alignment.Our code, data, and checkpoints are available at https://huggingface.co/AIJian/PaTaRM

Large reasoning models (LRMs) achieve remarkable performance via long reasoning chains, but often incur excessive computational overhead due to redundant reasoning, especially on simple tasks. In this work, we systematically quantify the upper bounds of LRMs under both Long-Thinking and No-Thinking modes, and uncover the phenomenon of “Internal Self-Recovery Mechanism” where models implicitly supplement reasoning during answer generation. Building on this insight, we propose Adaptive Self-Recovery Reasoning (ASRR), a framework that suppresses unnecessary reasoning and enables implicit recovery. By introducing accuracy-aware length reward regulation, ASRR adaptively allocates reasoning effort according to problem difficulty, achieving high efficiency with negligible performance sacrifice. Experiments across multiple benchmarks and models show that, compared with GRPO, ASRR reduces reasoning budget by up to 32.5% (1.5B) and 25.7% (7B) with minimal accuracy loss (1.2% and 0.6% pass@1), and significantly boosts harmless rates on safety benchmarks (up to +21.7%). Our results highlight the potential of ASRR for enabling efficient, adaptive, and safer reasoning in LRMs.

Detecting abnormal events in real-world customer service dialogues is highly challenging due to the complexity of business data and the dynamic nature of customer interactions. Moreover, models must demonstrate strong out-of-domain (OOD) generalization to enable rapid adaptation across different business scenarios and maximize commercial value.In this work, we propose a novel Adaptive Perplexity-Aware Reinforcement Learning (APARL) framework that leverages the advanced reasoning capabilities of large language models for abnormal event detection. APARL introduces a dual-loop dynamic curriculum learning architecture, enabling the model to progressively focus on more challenging samples as its proficiency increases. This design effectively addresses performance bottlenecks and significantly enhances OOD transferability.Extensive evaluations on food delivery dialogue tasks show that our model achieves significantly enhanced adaptability and robustness, attaining the highest F1 score with an average improvement of 17.19%, and an average improvement of 9.59% in OOD transfer tests. This method provides a superior solution for industrial deployment of anomaly detection models, contributing to improved operational efficiency and commercial benefits.

The end-to-end task-oriented dialogue system has achieved great success in recent years. Most of these dialogue systems need to accommodate multi-domain dialogue in real-world scenarios. However, due to the high cost of dialogue data annotation and the scarcity of labeled dialogue data, existing methods are difficult to extend to new domains. Therefore, it is important to use limited data to construct multi-domain dialogue systems. To solve this problem, we propose a novel domain attention module. It use the distributional signatures to construct a multi-domain dialogue system effectively with limited data, which has strong extensibility. We also define a adjacent n-gram pattern to explore potential patterns for dialogue entities. Experimental results show that our approach outperforms the baseline models on most metrics. In the few-shot scenario, we show our method get a great improvement compared with previous methods while keeping smaller model scale.