Reinforcement learning (RL) is effective for improving code generation but suffers from data scarcity. While experience replay mitigates this, existing approaches rely on static, in-epoch metrics that overlook training dynamics, often introducing low-utility or outdated data. Analyzing RL dynamics via dataset cartography, we observe that “ambiguous” samples, which are vital for model generalization, rapidly migrate to “easy-to-learn” regions, diminishing their training value. To address this, we propose Adaptive Ambiguity Replay (A2R) for RL, a plug-and-play module that prioritizes cross-epoch ambiguous samples. To neutralize the noise from stale experiences, A2R incorporates an adaptive importance mechanism based on policy divergence to weigh replayed rollouts. Extensive experiments on nine LLMs (3B–14B) demonstrate that A2R outperforms state-of-the-art baselines on real-world code editing tasks across both unseen and learned domains. Our results highlight cross-epoch ambiguity as a key factor for effective replay in RL. Code: https://github.com/TsingZ0/verl-A2R

Knowledge Base Question Answering (KBQA) aims to retrieve accurate answers to natural language queries by retrieving and reasoning over large-scale structured knowledge bases (KBs). Advanced semantic parsing-based methods promoted by large language models (LLMs) demonstrate superior performance by transforming questions into structured queries, i.e., logical forms (LFs). However, LFs generated by LLMs could be non-executable due to the inherent semantic hallucination issue of LLMs and the complex graph retrieval characteristics of the KBQA task. To address this challenge, we propose a novel "generate-verify-refine" framework, termed Action-Reflection-Integrated KBQA (ARI-KBQA) for reliable LF generation. ARI-KBQA introduces a dual-module cooperative architecture: First, an action generator is trained to produce initial query paths based on a hop-by-hop reasoning strategy. Then a reflection verifier dynamically validates path feasibility by interacting with the KBs. Consequently, ARI-KBQA filters out invalid LFs and provides semantic correction feedback to the action generator for iteratively refining LFs. Evaluations on standard KBQA benchmarks show that the proposed ARI-KBQA significantly enhances model performance with a reduced search space, especially in complex multi-hop query scenarios.

Reinforcement learning (RL) is widely used for post-training large language models (LLMs) in code editing, where group-relative methods, such as GRPO, are popular due to their critic-free and normalized advantage estimation. However, in real-world code-editing scenarios, reward distributions are often skewed with unpredictable noise, leading to distorted advantage computation and increased rollout outliers. To address this issue, we propose Group Adaptive Policy Optimization (GAPO), which adaptively finds an interval with the highest SNR (Signal to Noise Ratio) per prompt and uses the median of that interval as an adaptive Q to replace the group mean in advantage calculation to reduce noise further. This adaptive Q robustly handles rollout noise while remaining plug-and-play and efficient. We evaluate GAPO on nine instruction-tuned LLMs (3B–14B) using a collected large dataset of 51,844 real-world, history-aware code-editing tasks spanning 10 programming languages. GAPO yields up to 4.35 in-domain (ID) and 5.30 out-of-domain (OOD) exact-match improvements over GRPO and its variant DAPO, while achieving lower clipping ratios and higher GPU throughput. Code: https://github.com/TsingZ0/verl-GAPO

Knowledge base question answering (KBQA) refers to the task of answering natural language questions using large-scale structured knowledge bases (KBs). Existing semantic parsing-based (SP-based) methods achieve superior performance by directly converting questions into structured logical form (LF) queries using fine-tuned large language models (LLMs). However, these methods face the key challenge of difficulty in directly generating LFs for complex graph structures, which often leads to non-executable LFs that negatively impact overall KBQA performance. To address this challenge, we propose KaeDe, a novel generate-then-retrieve method for KBQA. This approach integrates knowledge-aware question decomposition and subsequent progressive LF generation within the generation phase, followed by an unsupervised retrieval phase. Specifically, the original question is decomposed into simplified, topic entity-centric sub-questions and explanations within the KB context. Path-level LFs are derived from these intermediate expressions and then combined into a comprehensive graph-level LF. Finally, the LF is refined through unsupervised entity and relation retrieval. Experimental results demonstrate that our method achieves state-of-the-art (SOTA) performance on WebQuestionSP (WebQSP) and ComplexWebQuestions (CWQ) benchmarks, particularly with fewer model parameters.

Knowledge-based complex reasoning remains a significant challenge for large language models (LLMs) with in-context learning. To tackle this issue, previous studies focus on ensuring behavior fidelity, factuality, or reliability in generated reasoning processes that guide LLMs to produce solutions. However, these studies often neglect the simultaneous optimization on all these three aspects for each thought. The main challenges are the lack of comprehensive assessment mechanisms and the difficulty of efficient thought-level optimization. This paper introduces the Evolution of Thoughts (EoT) framework, which enhances the factuality, fidelity, and reliability of each thought in the reasoning process through a few LLM inferences. We propose a thought assessment method that is sensitive to knowledge and LLM behaviors, using three scorers to evaluate each thought by considering domain context, semantic alignment, and behavior impact. Additionally, we establish a self-reflective evolution mechanism to facilitate each reasoning process generation in a single-forward inference. Extensive experiments demonstrate that, for knowledge-based complex tasks, EoT improves the factuality and fidelity of reasoning processes by approximately 16.5% and 48.8%, respectively, while enhancing LLM reasoning capability by about 6.2%, outperforming advanced approaches.