RL for Language Models

Transpilation, or code translation, aims to convert source code from one programming language (PL) to another. It is beneficial for many downstream applications, from modernizing large legacy codebases to augmenting data for low-resource PLs. Recent large language model (LLM)-based approaches have demonstrated immense potential for code translation. Among these approaches, training-based methods are particularly important because LLMs currently do not effectively adapt to domain-specific settings that suffer from a lack of knowledge without targeted training. This limitation is evident in transpilation tasks involving low-resource PLs. However, existing training-based approaches rely on a pairwise transpilation paradigm, making it impractical to support a diverse range of PLs. This limitation is particularly prominent for low-resource PLs due to a scarcity of training data. Furthermore, these methods suffer from suboptimal reinforcement learning (RL) reward formulations. To address these limitations, we propose CodePivot, a training framework that leverages Python as an intermediate representation (IR), augmented by a novel RL reward mechanism, Aggressive-Partial-Functional reward, to bootstrap the model's multilingual transpilation ability without requiring parallel corpora. Experiments involving 10 PLs show that the resulting 7B model, trained on Python-to-Others tasks, consistently improves performance across both general and low-resource PL-related transpilation tasks. It outperforms substantially larger mainstream models with hundreds of billions more parameters, such as Deepseek-R1 and Qwen3-235B-A22B-Instruct-2507, on Python-to-Others tasks and Others-to-All tasks, respectively. In addition, it outperforms its counterpart trained directly on Any-to-Any tasks on general transpilation tasks. The code and data are available at https://github.com/lishangyu-hkust/CodePivot.

Emotion Recognition in Conversation (ERC) has become a fundamental capability for large language models (LLMs) in human-centric interaction. Beyond accurate recognition, coherent emotional expression is also crucial, yet both are limited by the scarcity and static nature of high-quality annotated data. In this work, we propose SELF-EMO, a self-evolution framework grounded in the hypothesis that better emotion prediction leads to more consistent emotional responses. We introduce two auxiliary tasks, emotional understanding and emotional expression, and design a role-based self-play paradigm where the model acts as both an emotion recognizer and a dialogue responder. Through iterative interactions, the model generates diverse conversational trajectories, enabling scalable data generation. To ensure quality, we adopt a data flywheel mechanism that filters candidate predictions and responses using a smoothed IoU-based reward and feeds selected samples back for continuous self-improvement without external supervision. We further develop SELF-GRPO, a reinforcement learning algorithm that stabilizes optimization with multi-label alignment rewards and group-level consistency signals. Experiments on IEMOCAP, MELD, and EmoryNLP show that SELF-EMO achieves state-of-the-art performance, improving accuracy by +6.33% on Qwen3-4B and +8.54% on Qwen3-8B, demonstrating strong effectiveness and generalization.

We argue that current evaluation frameworks for large language models (LLMs) suffer from four systematic failures that make them structurally inadequate for deployed, agentic systems: distributional, temporal, scope, and process invalidity. These failures compound in RLHF, making reward hacking a predictable consequence of evaluation design rather than an unpredictable training pathology, and RLHF's dual-model architecture imposes a hardware barrier limiting evaluation reproducibility. We propose the Grounded Continuous Evaluation (GCE) framework and present ISOPro as a reference implementation. ISOPro replaces the learned reward model with a deterministic verifier, eliminating reward hacking by construction in verifiable-reward domains, and updates LoRA adapters on CPU, reducing the hardware barrier by an order of magnitude. We validate ISOPro across three architectures (Qwen 2.5 3B, Llama 3.2 3B, Gemma 2 2B) and two domains (scheduling, MBPP), with a head-to-head matched-compute comparison against GRPO-LoRA. Across twelve cells, ISOPro produces the largest absolute capability gains (+25.6, +22.2, +16.0pp) at mean delta +9.0pp and worst-case regression -5.6pp; GRPO-LoRA at consumer-budget hyperparameters reaches a smaller peak gain (+8.5pp), deeper worst-case regression (-10pp), and mean delta -1.5pp. Held-out compositional generalization on MBPP reaches 40% for ISOPro on two of three architectures (including a 0% to 40% bootstrap on Qwen 2.5 3B), against 20% for GRPO-LoRA on one of three. We characterize a buffer-skew failure mode in which the implicit curriculum can erode pre-existing tier capability under three preconditions, with three corresponding mitigations. The work is situated alongside DeepSeek-R1's GRPO, which arrived at the same architectural conclusion at scale: for verifiable-reward domains, the verifier is the reward signal.

Learning from preference-based feedback has become an effective approach for aligning LLMs across diverse tasks. However, high-quality human-annotated preference data remains expensive and scarce. Existing methods address this challenge through either self-rewarding, which scales by using purely AI-generated labels but risks unreliability, or active learning, which ensures quality through oracle annotation but cannot fully leverage unlabeled data. In this paper, we present CoAct, a novel framework that synergistically combines self-rewarding and active learning through strategic human-AI collaboration. CoAct leverages self-consistency to identify both reliable self-labeled data and samples that require oracle verification. Additionally, oracle feedback guides the model to generate new instructions within its solvable capability. Evaluated on three reasoning benchmarks across two model families, CoAct achieves average improvements of +13.25% on GSM8K, +8.19% on MATH, and +13.16% on WebInstruct, consistently outperforming all baselines.

This paper investigates the length problem in sequence-level relative reinforcement learning. We observe that, although existing methods partially alleviate length-related phenomena, a more fundamental issue remains insufficiently characterized: the comparison units used during training lack inherent comparability. Building on this observation, we propose a new perspective: the length problem should not be viewed merely as a loss-scaling or normalization bias, but rather as a comparison unit construction problem. We further establish a sample-construction-based training framework that, instead of applying post-hoc corrections to unequal-length responses, proactively constructs equal-length, alignable, and comparable training segments during generation. Within this framework, we propose EqLen, a concrete method applicable to group-relative comparison algorithms such as GRPO, GSPO, and RLOO. Through dual-track synchronous generation, prefix inheritance, and segment masking, EqLen efficiently collects effective equal-length training segments and enables stable

Reinforcement learning (RL) has emerged as a powerful post-training paradigm for enhancing the reasoning capabilities of large language models (LLMs). However, reinforcement learning for LLMs faces substantial data scarcity challenges, including the limited availability of high-quality external supervision and the constrained volume of model-generated experience. These limitations make data-efficient reinforcement learning a critical research direction. In this survey, we present the first systematic review of reinforcement learning for LLMs under data scarcity. We propose a bottom-up hierarchical framework built around three complementary perspectives: the data-centric perspective, the training-centric perspective, and the framework-centric perspective. We develop a taxonomy of existing methods, summarize representative approaches in each category, and analyze their strengths and limitations. Our taxonomy aims to provide a clear conceptual foundation for understanding the design space of data-efficient RL for LLMs and to guide researchers working in this emerging area. We hope this survey offers a comprehensive roadmap for future research and inspires new directions toward more efficient and scalable reinforcement learning post-training for LLMs.

Cross-cultural entity translation remains challenging for large language models (LLMs) as literal or phonetic renderings are usually yielded instead of culturally appropriate translations in context. However, relevant knowledge may already be encoded in model parameters during large-scale pre-training. To incentivize the effective use of parametric knowledge, we propose EA-RLVR (Entity-Anchored Reinforcement Learning with Verifiable Rewards), a training framework that optimizes cross-cultural entity translation without relying on external knowledge bases. EA-RLVR anchors supervision on a verifiable, entity-level reward signal and incorporates lightweight structural gates to stabilize optimization. This design steers the model toward learning a robust reasoning process rather than merely imitating reference translations. We evaluate EA-RLVR on XC-Translate and observe consistent improvements in both entity translation accuracy and out-of-domain generalization. Specifically, training on merely 7k samples boosts Qwen3-14B's entity translation accuracy from 23.66% to 31.87% on a 50k test set comprising entirely unseen entities. The learned entity translation ability also transfers to general translation, yielding +1.35 XCOMET on WMT24++, which scales to +1.59 with extended optimization. Extensive analyses of $pass@k$ dynamics and reward formulations attribute these gains to superior sampling efficiency and a stable optimization landscape.

Large Language Models (LLMs) have demonstrated potential in automating scientific ideation, yet current approaches relying on iterative prompting or complex multi-agent architectures often suffer from hallucination or computational inefficiency. A critical bottleneck in applying Reinforcement Learning (RL) to this open-ended domain is reward hacking -- where models exploit imperfect evaluation proxies to maximize scores without producing genuine scientific innovation. To address these limitations, we propose an RL framework explicitly tailored for high-quality scientific idea generation. We propose the first multi-agent reward function designed to serve as a judge, decoupling methodological validation from implementation details while providing strict binary rewards that are robust to reward hacking. To effectively optimize against this sparse signal, we utilize an unbiased variant of Group Relative Policy Optimization to mitigate artificial length bias. We grounded our training in ICLR-320, a curated dataset of problem-solution pairs extracted from ICLR 2024 proceedings. Experiments demonstrate that our framework significantly outperforms state-of-the-art baselines across expert-evaluated metrics of novelty, feasibility, and effectiveness.

Large language models can support medical question answering, but privacy, cost, and reliability still limit their clinical use in focused domains. Can rubric-based reinforcement learning bring a locally deployable model close to much larger systems? We study Group Relative Policy Optimization (GRPO) for heart-focused medical question answering with rubric supervision from RaR-Medicine. Rather than rely on binary rubric aggregation or a single overall score, our variance-aware reward framework builds continuous rewards from criterion-level outcomes. It preserves partial credit and reduces reward sparsity. We adapt a Qwen3-14B policy with low-rank adapters and quantized weights. On a held-out heart-related HealthBench subset, the best GRPO variant improves accuracy from 0.362 to 0.502 and F1 from 0.532 to 0.668 over the base model, close to GPT-OSS-120B. An independent clinician rated most responses clinically acceptable (61%) and patient ready (63%), both above the automated accuracy; fewer than 2% were unsafe. Taken together, the results support rubric-based reinforcement learning as a practical route for focused medical language models.

Large language models are typically post-trained using supervised fine-tuning (SFT) and reinforcement learning (RL), yet effectively unifying efficient knowledge injection with robust generalization remains challenging. In this work, we provide a training-dynamics analysis showing that SFT can be interpreted as a special case of policy gradient optimization with an extremely sparse implicit reward and unstable inverse-probability weighting, which together lead to single-path dependency, entropy collapse, and gradient explosion. Motivated by this diagnosis, we propose Group Fine-Tuning (GFT), a unified post-training framework that addresses these intrinsic limitations through two mechanisms: Group Advantage Learning, which constructs diverse response groups and derives normalized contrastive supervision to alleviate reward sparsity, and Dynamic Coefficient Rectification, which adaptively bounds inverse-probability weights to stabilize optimization while preserving efficient knowledge injection. Experiments demonstrate that GFT consistently surpasses SFT-based methods and yields policies that integrate more smoothly with subsequent RL training.

Editing human-written text has become a standard use case of large language models (LLMs), for example, to make one's arguments more appropriate for a discussion. Comparing human to LLM-generated edits, however, we observe a mismatch in editing strategies: While LLMs often perform multiple scattered edits and tend to change meaning notably, humans rather encapsulate dependent changes in self-contained, meaning-preserving edits. In this paper, we present a reinforcement learning approach that teaches LLMs human-like editing to improve the appropriateness of arguments. Our approach produces self-contained sentence-level edit suggestions that can be accepted or rejected independently. We train the approach using group relative policy optimization with a multi-component reward function that jointly optimizes edit-level semantic similarity, fluency, and pattern conformity as well as argument-level appropriateness. In automatic and human evaluation, it outperforms competitive baselines and the state of the art in human-like editing, with multi-round editing achieving appropriateness close to full rewriting.

RLVR improves reasoning in large language models, but its effectiveness is often limited by severe reward sparsity on hard problems. Recent hint-based RL methods mitigate sparsity by injecting partial solutions or abstract templates, yet they typically scale guidance by adding more tokens, which introduce redundancy, inconsistency, and extra training overhead. We propose KnowRL (Knowledge-Guided Reinforcement Learning), an RL training framework that treats hint design as a minimal-sufficient guidance problem. During RL training, KnowRL decomposes guidance into atomic knowledge points (KPs) and uses Constrained Subset Search (CSS) to construct compact, interaction-aware subsets for training. We further identify a pruning interaction paradox -- removing one KP may help while removing multiple such KPs can hurt -- and explicitly optimize for robust subset curation under this dependency structure. We train KnowRL-Nemotron-1.5B from OpenMath-Nemotron-1.5B. Across eight reasoning benchmarks at the 1.5B scale, KnowRL-Nemotron-1.5B consistently outperforms strong RL and hinting baselines. Without KP hints at inference, KnowRL-Nemotron-1.5B reaches 70.08 average accuracy, already surpassing Nemotron-1.5B by +9.63 points; with selected KPs, performance improves to 74.16, establishing a new state of the art at this scale. The model, curated training data, and code are publicly available at https://github.com/Hasuer/KnowRL.

Specification gaming under Reinforcement Learning (RL) is known to cause LLMs to develop sycophantic, manipulative, or deceptive behavior, yet the conditions under which this occurs remain unclear. We train 11 instruction-tuned LLMs (0.5B-14B) with on-policy RL across 3 environments and find that model size acts as a safety buffer in some environments but enables greater harmful exploitation in others. Controlled ablations trace this reversal to environment-specific features such as role framing and implicit gameability cues. We further show that most safety benchmarks do not predict RL-induced misalignment, except in the case of Sycophancy scores when the exploit relies on inferring the user's preference. Finally, we find that on-policy RL preserves a safety buffer inherent in the model's own generation distribution, one that is bypassed during off-policy settings.

Large Language Model (LLM) based Chinese Grammatical Error Correction (CGEC) systems face two critical challenges: general-purpose models lack specialized linguistic priors for subtle grammatical distinctions, and Supervised Fine-Tuning (SFT) with Maximum Likelihood Estimation fails to optimize for precision-focused metrics, leading to systematic over-correction. We propose CSRP, a three-stage framework that progressively builds correction capability through Continual Pre-training (CPT) on 5.9M balanced samples to internalize domain knowledge, Chain-of-Thought SFT with explicit error reasoning for diagnostic transparency, and Group Relative Policy Optimization with a novel Efficiency-Aware Reward that explicitly penalizes unnecessary edits. On the NACGEC benchmark, CSRP achieves state-of-the-art performance with 50.99 $F_{0.5}$ and 57.17 precision, substantially outperforming previous best results while effectively mitigating the over-correction bias inherent in MLE-trained models. Our method also advances CSCD spelling correction to 59.61 F1, surpassing GPT-4 by 5.20 points. Comprehensive ablation studies demonstrate that the RL alignment stage contributes a 8% relative gain over the SFT baseline, and that this gain is orthogonal to the contribution of large-scale CPT, validating that explicit optimization for edit efficiency is essential for high-quality grammatical error correction. Our code is available at https://github.com/TW-NLP/ChineseErrorCorrector.

Training language model agents for multi-agent strategic interaction presents a core difficulty: the quality of any action may depend on future events that never materialize, on moves that violate game rules, or on decisions made by other players. Standard reinforcement learning assumes that rewards can be assigned at each step, but this assumption fails in settings where outcomes are entangled across time and agents. We introduce delayed per-step reward attribution with eligibility gating, an episode lifecycle and postprocessing pipeline that computes rewards only at episode end, propagates them back to originating steps according to task-specific semantics, and excludes steps that lack valid dependent information from training. Together with asynchronous rollout generation via vLLM's continuous batching, curriculum-based opponent sampling, and multi-level stratified batch construction, this approach enables stable, sample-efficient RL training in multi-agent environments. We evaluate on the MindGames Arena benchmark at NeurIPS 2025, where a single 8-billion-parameter open-source model trained with our method matched or surpassed substantially larger proprietary systems, including GPT-5, in head-to-head play and took first place in both the Open (unrestricted) and Efficient (<=8B parameters) tracks.

Large language models (LLMs) produce responses rated as highly empathic in single-turn settings (Ayers et al., 2023; Lee et al., 2024), yet they are also known to be formulaic generators that reuse the same lexical patterns, syntactic templates, and discourse structures across tasks (Jiang et al., 2025; Shaib et al., 2024; Namuduri et al., 2025). Less attention has been paid to whether this formulaicity extends to the level of discourse moves, i.e., what a response does for the person it is addressing. This question is especially consequential for empathic dialogue, where effective support demands not just a kind response at one moment but varied strategies as a conversation unfolds (Stiles et al., 1998). Indeed, prior work shows that LLMs reuse the same tactic sequences more than human supporters in single-turn settings (Gueorguieva et al., 2026). We extend this analysis to multi-turn conversations and find that the rigidity compounds: once a tactic appears in a supporter turn, LLMs reuse it in the next at nearly double the rate of humans (0.50-0.56 vs. 0.27). This pattern holds across LLMs serving as supporters in real emotional support conversations, and is invisible to standard similarity metrics. To address this gap, we introduce MINT (Multi-turn Inter-tactic Novelty Training), the first reinforcement learning framework to optimize discourse move diversity across multi-turn empathic dialogue. The best MINT variant combines an empathy quality reward with a cross-turn tactic novelty signal, improving aggregate empathy by 25.3% over vanilla across 1.7B and 4B models while reducing cross-turn discourse move repetition by 26.3% on the 4B model, surpassing all baselines including quality-only and token-level diversity methods on both measures. These results suggest that what current models lack is not empathy itself, but the ability to vary their discourse moves across a conversation.

Reinforcement learning for open-ended text generation is constrained by the lack of verifiable rewards, necessitating reliance on judge models that require either annotated data or powerful closed-source models. Inspired by recent work on unsupervised reinforcement learning for mathematical reasoning using confidence-based endogenous rewards, we investigate whether this principle can be adapted to open-ended writing tasks. We find that directly applying confidence rewards leads to Triviality Bias: the policy collapses toward high-probability outputs, reducing diversity and meaningful content. We propose TCER (Triviality Corrected Endogenous Reward), which addresses this bias by rewarding the relative information gain between a specialist policy and a generalist reference policy, modulated by a probability-dependent correction mechanism. Across multiple writing benchmarks and model architectures, TCER achieves consistent improvements without external supervision. Furthermore, TCER also transfers effectively to mathematical reasoning, validating the generality of our approach across different generation tasks.

To encourage diverse exploration in reinforcement learning (RL) for large language models (LLMs) without compromising accuracy, we propose Policy Split, a novel paradigm that bifurcates the policy into normal and high-entropy modes with a high-entropy prompt. While sharing model parameters, the two modes undergo collaborative dual-mode entropy regularization tailored to distinct objectives. Specifically, the normal mode optimizes for task correctness, while the high-entropy mode incorporates a preference for exploration, and the two modes learn collaboratively. Extensive experiments demonstrate that our approach consistently outperforms established entropy-guided RL baselines across various model sizes in general and creative tasks. Further analysis reveals that Policy Split facilitates dual-mode exploration, where the high-entropy mode generates distinct behavioral patterns to the normal mode, providing unique learning signals.

Despite the success of reinforcement learning for large language models, a common failure mode is reduced sampling diversity, where the policy repeatedly generates similar erroneous behaviors. Classical entropy regularization encourages randomness under the current policy, but does not explicitly discourage recurrent failure patterns across rollouts. We propose MEDS, a Memory-Enhanced Dynamic reward Shaping framework that incorporates historical behavioral signals into reward design. By storing and leveraging intermediate model representations, we capture features of past rollouts and use density-based clustering to identify frequently recurring error patterns. Rollouts assigned to more prevalent error clusters are penalized more heavily, encouraging broader exploration while reducing repeated mistakes. Across five datasets and three base models, MEDS consistently improves average performance over existing baselines, achieving gains of up to 4.13 pass@1 points and 4.37 pass@128 points. Additional analyses using both LLM-based annotations and quantitative diversity metrics show that MEDS increases behavioral diversity during sampling.

Credit assignment is a central challenge in reinforcement learning (RL). Classical actor-critic methods address this challenge through fine-grained advantage estimation based on a learned value function. However, learned value models are often avoided in modern large language model (LLM) RL because conventional discriminative critics are difficult to train reliably. We revisit value modeling and argue that this difficulty is partly due to limited expressiveness. In particular, representation complexity theory suggests that value functions can be hard to approximate under the one-shot prediction paradigm used by existing value models, and our scaling experiments show that such critics do not improve reliably with scale. Motivated by this observation, we propose Generative Actor-Critic (GenAC), which replaces one-shot scalar value prediction with a generative critic that performs chain-of-thought reasoning before producing a value estimate. We further introduce In-Context Conditioning, which helps the critic remain calibrated to the current actor throughout training. GenAC improves value approximation, ranking reliability, and out-of-distribution generalization, and these gains translate into stronger downstream RL performance than both value-based and value-free baselines. Overall, our results suggest that stronger value modeling is a promising direction for improving credit assignment in LLM reinforcement learning.

End-to-end full-duplex Speech Language Models (SLMs) require precise turn-taking for natural interaction. However, optimizing temporal dynamics via standard raw-token reinforcement learning (RL) degrades semantic quality, causing severe generative collapse and repetition. We propose ASPIRin, an interactivity-optimized RL framework that explicitly decouples when to speak from what to say. Using Action Space Projection, ASPIRin maps the text vocabulary into a coarse-grained binary state (active speech vs. inactive silence). By applying Group Relative Policy Optimization (GRPO) with rule-based rewards, it balances user interruption and response latency. Empirical evaluations show ASPIRin optimizes interactivity across turn-taking, backchanneling, and pause handling. Crucially, isolating timing from token selection preserves semantic coherence and reduces the portion of duplicate n-grams by over 50% compared to standard GRPO, effectively eliminating degenerative repetition.

Reinforcement learning (RL) for large language models (LLMs) increasingly relies on sparse outcome rewards, yet such rewards say little about which token, reasoning step, tool call, memory operation, or agent caused an outcome. This credit assignment (CA) problem spans reasoning RL and becomes sharper in agentic RL, where environment interaction introduces transition non-closure, partial observability, limited replay, heterogeneous actions, weak intermediate verifiability, and agent coupling. We synthesize a unified corpus of 69 papers published from January 2024 through July 31, 2026: 56 core CA methods and 13 adjacent or boundary enablers, selected from 92 deduplicated screening records. We retain the original granularity-by-methodology taxonomy and add a six-diagnostic framework mapping assumption breaks to identification barriers, estimators, and evaluation controls. A source-located full-text audit covers a fixed 42-core-paper subset. Two algorithm researchers independently and blindly cross-coded 252 diagnostic cells, agreeing on 223 (88.5%); per-diagnostic Cohen's kappa ranges from .543 to .909, and principal-family agreement is 42/42 (kappa=1.000). Beyond taxonomy, we establish when restored-state comparisons identify a protocol-specific causal contrast, show that text-only histories can leave even the sign of credit unidentified, and introduce a reusable CA-ID Card linking each claim to its estimand, evidence provenance, and falsification test. An atomic reporting audit describes comparator, budget parity, ablation, overhead, uncertainty, and replay coverage without constructing a cross-paper leaderboard. The companion repository hosts a living catalog and decision aids; a dated release of the frozen audit bundle is planned there separately from the minimal arXiv source.

Modern LLM reinforcement learning (RL) workloads require a high-performance weight transfer system to scale training across heterogeneous compute resources. However, efficiently transferring terabyte-scale model weights across thousands of GPUs remains challenging because the system must accommodate clusters that dynamically scale up and down while keeping coordination, data movement, and storage overhead low. We introduce Reference-Oriented Storage (ROS), a new storage abstraction for RL weight transfer that exploits highly replicated model weights in place. ROS presents the illusion that certain versions of the model weights are stored and can be fetched on demand. Underneath, ROS does not physically store any copies of the weights; instead, it tracks the workers that hold these weights on GPUs for inference. Upon request, ROS directly uses them to serve reads. We build TensorHub, a production-quality system that instantiates the ROS idea with topology-aware transfer, model-parallel consistency, and fault tolerance. Evaluation shows that TensorHub saturates RDMA bandwidth and adapts to three distinct rollout workloads with minimal engineering effort. Specifically, TensorHub reduces total GPU stall time by up to 6.7x for standalone rollouts, accelerates weight updates for elastic rollouts by up to 4.8x, and cuts cross-datacenter rollout stall time by up to 19x. TensorHub has been deployed in ByteDance production to support cutting-edge RL training.

Large language models are increasingly deployed in high-stakes domains, where confident yet incorrect inferences may cause severe real-world harm, bringing the long-overlooked issue of confidence faithfulness to the forefront. A promising solution jointly optimizes unsupervised Reinforcement Learning from Internal Feedback (RLIF) with reasoning-trace-guided Reasoning Distillation (RD), yet it faces three persistent challenges, namely the scarcity of high-quality training corpora, factually unwarranted overconfidence, and erroneous updates amplified by indiscriminate fusion. Inspired by how human confidence accumulates from uncertainty to certainty, we propose Progressive Reasoning Gain (PRG) to measure whether reasoning steps progressively strengthen confidence in the final answer. Building on PRG, we introduce HyTuning, a hybrid post-training framework that adaptively reweights RD and RLIF, using scarce supervised reasoning traces as a stable anchor while exploiting abundant unlabeled queries for scalability. Experiments on several domain-specific and general benchmarks demonstrate that HyTuning improves accuracy while achieving confidence faithfulness under limited supervision, supporting a practical ``Less Data Approximates More'' effect. Our code will be released upon acceptance.

Post-training has become central to turning pretrained large language models (LLMs) into aligned, capable, and deployable systems. Recent progress spans supervised fine-tuning (SFT), preference optimization, reinforcement learning (RL), process supervision, verifier-guided methods, distillation, and multi-stage pipelines. Yet these methods are often discussed in fragmented ways, organized by labels or objectives rather than by the behavioral bottlenecks they address. This survey argues that LLM post-training is best understood as structured intervention on model behavior. We organize the field first by trajectory provenance, which defines two primary regimes: off-policy learning on externally supplied trajectories and on-policy learning on learner-generated rollouts. We then interpret methods through two recurring roles -- effective support expansion, which makes useful behaviors more reachable, and policy reshaping, which improves behavior within already reachable regions -- together with a complementary systems-level role, behavioral consolidation, which preserves, transfers, and amortizes useful behavior across stages and model transitions. Under this view, SFT may serve either support expansion or policy reshaping; preference optimization is usually off-policy reshaping, though online variants move closer to learner-generated states. On-policy RL often improves behavior on learner-generated states, but stronger guidance can also make hard-to-reach reasoning paths reachable. Distillation is often better understood as consolidation rather than only compression, and hybrid pipelines emerge as coordinated multi-stage compositions. Overall, the framework helps diagnose post-training bottlenecks and reason about stage composition, suggesting that progress increasingly depends on coordinated systems design rather than any single dominant objective.

Reinforcement learning (RL) has become a central post-training tool for improving the reasoning abilities of large language models (LLMs). In these systems, the rollout, the trajectory sampled from a prompt to termination, including intermediate reasoning steps and optional tool or environment interactions, determines the data the optimizer learns from, yet rollout design is often underreported. This survey provides an optimizer-agnostic view of rollout strategies for RL-based post-training of reasoning LLMs. We formalize rollout pipelines with unified notation and introduce Generate-Filter-Control-Replay (GFCR), a lifecycle taxonomy that decomposes rollout pipelines into four modular stages: Generate proposes candidate trajectories and topologies; Filter constructs intermediate signals via verifiers, judges, critics; Control allocates compute and makes continuation/branching/stopping decisions under budgets; and Replay retains and reuses artifacts across rollouts without weight updates, including self-evolving curricula that autonomously generate new training tasks. We complement GFCR with a criterion taxonomy of reliability, coverage, and cost sensitivity that characterizes rollout trade-offs. Using this framework, we synthesize methods spanning RL with verifiable rewards, process supervision, judge-based gating, guided and tree/segment rollouts, adaptive compute allocation, early-exit and partial rollouts, throughput optimization, and replay/recomposition for self-improvement. We ground the framework with case studies in math, code/SQL, multimodal reasoning, tool-using agents, and agentic skill benchmarks that evaluate skill induction, reuse, and cross-task transfer. Finally, we provide a diagnostic index that maps common rollout pathologies to GFCR modules and mitigation levers, alongside open challenges for building reproducible, compute-efficient, and trustworthy rollout pipelines.

A fundamental challenge in creative writing lies in reconciling the inherent tension between maintaining global coherence in long-form narratives and preserving local expressiveness in short-form texts. While long-context generation necessitates explicit macroscopic planning, short-form creativity often demands spontaneous, constraint-free expression. Existing alignment paradigms, however, typically employ static reward signals and rely heavily on high-quality supervised data, which is costly and difficult to scale. To address this, we propose UniCreative, a unified reference-free reinforcement learning framework. We first introduce AC-GenRM, an adaptive constraint-aware reward model that dynamically synthesizes query-specific criteria to provide fine-grained preference judgments. Leveraging these signals, we propose ACPO, a policy optimization algorithm that aligns models with human preferences across both content quality and structural paradigms without supervised fine-tuning and ground-truth references. Empirical results demonstrate that AC-GenRM aligns closely with expert evaluations, while ACPO significantly enhances performance across diverse writing tasks. Crucially, our analysis reveals an emergent meta-cognitive ability: the model learns to autonomously differentiate between tasks requiring rigorous planning and those favoring direct generation, validating the effectiveness of our direct alignment approach.

Text simplification supports second language (L2) learning by providing comprehensible input, consistent with the Input Hypothesis. However, constructing personalized parallel corpora is costly, while existing large language model (LLM)-based readability control methods rely on pre-labeled sentence corpora and primarily target English. We propose Re-RIGHT, a unified reinforcement learning framework for adaptive multilingual text simplification without parallel corpus supervision. We first show that prompting-based lexical simplification at target proficiency levels (CEFR, JLPT, TOPIK, and HSK) performs poorly at easier levels and for non-English languages, even with state-of-the-art LLMs such as GPT-5.2 and Gemini 2.5. To address this, we collect 43K vocabulary-level data across four languages (English, Japanese, Korean, and Chinese) and train a compact 4B policy model using Re-RIGHT, which integrates three reward modules: vocabulary coverage, semantic preservation, and coherence. Compared to the stronger LLM baselines, Re-RIGHT achieves higher lexical coverage at target proficiency levels while maintaining original meaning and fluency.

Large language models exhibit sycophancy, the tendency to shift their stated positions toward perceived user preferences or authority cues regardless of evidence. Standard alignment methods fail to correct this because scalar reward models conflate two distinct failure modes into a single signal: pressure capitulation, where the model changes a correct answer under social pressure, and evidence blindness, where the model ignores the provided context entirely. We operationalise sycophancy through formal definitions of pressure independence and evidence responsiveness, serving as a working framework for disentangled training rather than a definitive characterisation of the phenomenon. We propose the first approach to sycophancy reduction via reward decomposition, introducing a multi-component Group Relative Policy Optimisation (GRPO) reward that decomposes the training signal into five terms: pressure resistance, context fidelity, position consistency, agreement suppression, and factual correctness. We train using a contrastive dataset pairing pressure-free baselines with pressured variants across three authority levels and two opposing evidence contexts. Across five base models, our two-phase pipeline consistently reduces sycophancy on all metric axes, with ablations confirming that each reward term governs an independent behavioural dimension. The learned resistance to pressure generalises beyond our training methodology and prompt structure, reducing answer-priming sycophancy by up to 17 points on SycophancyEval despite the absence of such pressure forms during training.

Non-stationary environments require agents to revise previously learned action values when contingencies change. We treat large language models (LLMs) as sequential decision policies in a two-option probabilistic reversal-learning task with three latent states and switch events triggered by either a performance criterion or timeout. We compare a deterministic fixed transition cycle to a stochastic random schedule that increases volatility, and evaluate DeepSeek-V3.2, Gemini-3, and GPT-5.2, with human data as a behavioural reference. Across models, win-stay was near ceiling while lose-shift was markedly attenuated, revealing asymmetric use of positive versus negative evidence. DeepSeek-V3.2 showed extreme perseveration after reversals and weak acquisition, whereas Gemini-3 and GPT-5.2 adapted more rapidly but still remained less loss-sensitive than humans. Random transitions amplified reversal-specific persistence across LLMs yet did not uniformly reduce total wins, demonstrating that high aggregate payoff can coexist with rigid adaptation. Hierarchical reinforcement-learning (RL) fits indicate dissociable mechanisms: rigidity can arise from weak loss learning, inflated policy determinism, or value polarisation via counterfactual suppression. These results motivate reversal-sensitive diagnostics and volatility-aware models for evaluating LLMs under non-stationary uncertainty.