RL Fine-Tuning

Single-view 3D human reconstruction has achieved remarkable progress through the adoption of multi-view diffusion models, yet the recovered 3D humans often exhibit unnatural poses. This phenomenon becomes pronounced when reconstructing 3D humans with dynamic or challenging poses, which we attribute to the limited scale of available 3D human datasets with diverse poses. To address this limitation, we introduce DrPose, Direct Reward fine-tuning algorithm on Poses, which enables post-training of a multi-view diffusion model on diverse poses without requiring expensive 3D human assets. DrPose trains a model using only human poses paired with single-view images, employing a direct reward fine-tuning to maximize PoseScore, which is our proposed differentiable reward that quantifies consistency between a generated multi-view latent image and a ground-truth human pose. This optimization is conducted on DrPose15K, a novel dataset that was constructed from an existing human motion dataset and a pose-conditioned video generative model. Constructed from abundant human pose sequence data, DrPose15K exhibits a broader pose distribution compared to existing 3D human datasets. We validate our approach through evaluation on conventional benchmark datasets, in-the-wild images, and a newly constructed benchmark, with a particular focus on assessing performance on challenging human poses. Our results demonstrate consistent qualitative and quantitative improvements across all benchmarks. Project page: https://seunguk-do.github.io/drpose.

Optimizing large language models for industrial sales requires balancing long-term commercial objectives (e.g., conversion rate) with immediate linguistic constraints such as fluency and compliance. Conventional reinforcement learning often merges these heterogeneous goals into a single reward, causing high-magnitude session-level rewards to overwhelm subtler turn-level signals, which leads to unstable training or reward hacking. To address this issue, we propose Dual-Horizon Credit Assignment (DuCA), a framework that disentangles optimization across time scales. Its core, Horizon-Independent Advantage Normalization (HIAN), separately normalizes advantages from turn-level and session-level rewards before fusion, ensuring balanced gradient contributions from both immediate and long-term objectives to the policy update. Extensive experiments with a high-fidelity user simulator show DuCA outperforms the state-of-the-art GRPO baseline, achieving a 6.82% relative improvement in conversion rate, reducing inter-sentence repetition by 82.28%, and lowering identity detection rate by 27.35%, indicating a substantial improvement for an industrial sales scenario that effectively balances the dual demands of strategic performance and naturalistic language generation.

Reinforcement learning (RL) for mathematical reasoning can suffer from reward sparsity: for challenging problems, LLM fails to sample any correct trajectories, preventing RL from receiving meaningful positive feedback. At the same time, there often exist human-written reference solutions along with the problem (e.g., problems from AoPS), but directly fine-tuning on these solutions offers no benefit because models often cannot imitate human proofs that lie outside their own reasoning distribution. We introduce Reference-Guided Fine-Tuning (ReGFT), a simple and effective method that utilizes human-written reference solutions to synthesize positive trajectories on hard problems and train on them before RL. For each problem, we provide the model with a partial reference solution and let it generate its own reasoning trace, ensuring the resulting trajectories remain in the model's reasoning space while still benefiting from reference guidance. Fine-tuning on these reference-guided trajectories increases the number of solvable problems and produces a checkpoint that receives more positive rewards during RL. Across three benchmarks (AIME24, AIME25, BeyondAIME), ReGFT consistently improves supervised accuracy, accelerates DAPO training, and raises the final performance plateau of RL. Our results show that ReGFT effectively overcomes reward sparsity and unlocks stronger RL-based mathematical reasoning.

While reinforcement learning (RL) has been central to the recent success of large language models (LLMs), RL optimization is notoriously unstable, especially when compared to supervised fine-tuning (SFT). In this work, we investigate the stability gap between SFT and RL from a gradient-based perspective, and show that the convexity of the SFT loss with respect to model logits plays a key role in enabling stable training. Our theoretical analysis demonstrates that this property induces favorable gradient directionality during optimization. In contrast, Proximal Policy Optimization (PPO), a widely adopted policy gradient algorithm utilizing a clipped surrogate objective, lacks this stabilizing property. Motivated by this observation, we propose Logits Convex Optimization (LCO), a simple yet effective policy optimization framework that aligns the learned policy with an optimal target derived from the original RL objective, thereby emulating the stabilizing effects of logits-level convexity. Extensive experiments across multiple model families show that our LCO framework consistently improves training stability and outperforms conventional RL methods on a broad range of benchmarks.

Referring Multi-Object Tracking has attracted increasing attention due to its human-friendly interactive characteristics, yet it exhibits limitations in low-visibility conditions, such as nighttime, smoke, and other challenging scenarios. To overcome this limitation, we propose a new RGB-Thermal RMOT task, named RT-RMOT, which aims to fuse RGB appearance features with the illumination robustness of the thermal modality to enable all-day referring multi-object tracking. To promote research on RT-RMOT, we construct the first Referring Multi-Object Tracking dataset under RGB-Thermal modality, named RefRT. It contains 388 language descriptions, 1,250 tracked targets, and 166,147 Language-RGB-Thermal (L-RGB-T) triplets. Furthermore, we propose RTrack, a framework built upon a multimodal large language model (MLLM) that integrates RGB, thermal, and textual features. Since the initial framework still leaves room for improvement, we introduce a Group Sequence Policy Optimization (GSPO) strategy to further exploit the model's potential. To alleviate training instability during RL fine-tuning, we introduce a Clipped Advantage Scaling (CAS) strategy to suppress gradient explosion. In addition, we design Structured Output Reward and Comprehensive Detection Reward to balance exploration and exploitation, thereby improving the completeness and accuracy of target perception. Extensive experiments on the RefRT dataset demonstrate the effectiveness of the proposed RTrack framework.

Vision-Language-Action (VLA) models have demonstrated significant potential for generalist robotic policies; however, they struggle to generalize to long-horizon complex tasks in novel real-world domains due to distribution shifts and the scarcity of high-quality demonstrations. Although reinforcement learning (RL) offers a promising avenue for policy improvement, applying it to real-world VLA fine-tuning faces challenges regarding exploration efficiency, training stability, and sample cost. To address these issues, we propose IG-RFT, a novel Interaction-Guided Reinforced Fine-Tuning system designed for flow-based VLA models. Firstly, to facilitate effective policy optimization, we introduce Interaction-Guided Advantage Weighted Regression (IG-AWR), an RL algorithm that dynamically modulates exploration intensity based on the robot's interaction status. Furthermore, to address the limitations of sparse or task-specific rewards, we design a novel hybrid dense reward function that integrates the trajectory-level reward and the subtask-level reward. Finally, we construct a three-stage RL system comprising SFT, Offline RL, and Human-in-the-Loop RL for fine-tuning VLA models. Extensive real-world experiments on four challenging long-horizon tasks demonstrate that IG-RFT achieves an average success rate of 85.0%, significantly outperforming SFT (18.8%) and standard Offline RL baselines (40.0%). Ablation studies confirm the critical contributions of IG-AWR and hybrid reward shaping. In summary, our work establishes and validates a novel reinforced fine-tuning system for VLA models in real-world robotic manipulation.

Financial market movements are often driven by discrete financial events conveyed through news, whose impacts are heterogeneous, abrupt, and difficult to capture under purely numerical prediction objectives. These limitations have motivated growing interest in using textual information as the primary source of trading signals in learning-based systems. Two key challenges hinder existing approaches: (1) the absence of large-scale, event-centric datasets that jointly model news semantics and statistically grounded market reactions, and (2) the misalignment between language model reasoning and financially valid trading behavior under dynamic market conditions. To address these challenges, we propose Janus-Q, an end-to-end event-driven trading framework that elevates financial news events from auxiliary signals to primary decision units. Janus-Q unifies event-centric data construction and model optimization under a two-stage paradigm. Stage I focuses on event-centric data construction, building a large-scale financial news event dataset comprising 62,400 articles annotated with 10 fine-grained event types, associated stocks, sentiment labels, and event-driven cumulative abnormal return (CAR). Stage II performs decision-oriented fine-tuning, combining supervised learning with reinforcement learning guided by a Hierarchical Gated Reward Model (HGRM), which explicitly captures trade-offs among multiple trading objectives. Extensive experiments demonstrate that Janus-Q achieves more consistent, interpretable, and profitable trading decisions than market indices and LLM baselines, improving the Sharpe Ratio by up to 102.0% while increasing direction accuracy by over 17.5% compared to the strongest competing strategies.

Reinforcement Learning (RL) has significantly improved large language model reasoning, but existing RL fine-tuning methods rely heavily on heuristic techniques such as entropy regularization and reweighting to maintain stability. In practice, they often suffer from late-stage performance collapse, leading to degraded reasoning quality and unstable training. We identify a key factor behind this instability: a small fraction of tokens, termed spurious tokens (around 0.01%), which contribute little to the reasoning outcome but receive disproportionately amplified gradient updates due to inheriting the full sequence-level reward. We present a unified framework for evaluating token-level optimization impacts across spurious risk, gradient norms, and entropy changes. Building on the analysis of token characteristics that severely disrupt optimization, we propose the Silencing Spurious Tokens (S2T) mechanism to efficiently suppress their gradient perturbations. Incorporating this mechanism into a group-based objective, we propose Spurious-Token-Aware Policy Optimization (STAPO), which promotes stable and effective large-scale model refinement. Across six mathematical benchmarks and three model scales, STAPO consistently demonstrates superior entropy stability and achieves average accuracy improvements over baselines including GRPO, 20-Entropy, and JustRL under two widely-adopted evaluation settings.

Large Language Models (LLMs) are changing the coding paradigm, known as vibe coding, yet synthesizing algorithmically sophisticated and robust code still remains a critical challenge. Incentivizing the deep reasoning capabilities of LLMs is essential to overcoming this hurdle. Reinforcement Fine-Tuning (RFT) has emerged as a promising strategy to address this need. However, most existing approaches overlook the heterogeneous difficulty and granularity inherent in test cases, leading to an imbalanced distribution of reward signals and consequently biased gradient updates during training. To address this, we propose Test-driven and cApability-adaptive cuRriculum reinfOrcement fine-Tuning (TAROT). TAROT systematically constructs, for each problem, a four-tier test suite (basic, intermediate, complex, edge), providing a controlled difficulty landscape for curriculum design and evaluation. Crucially, TAROT decouples curriculum progression from raw reward scores, enabling capability-conditioned evaluation and principled selection from a portfolio of curriculum policies rather than incidental test-case difficulty composition. This design fosters stable optimization and more efficient competency acquisition. Extensive experimental results reveal that the optimal curriculum for RFT in code generation is closely tied to a model's inherent capability, with less capable models achieving greater gains with an easy-to-hard progression, whereas more competent models excel under a hard-first curriculum. TAROT provides a reproducible method that adaptively tailors curriculum design to a model's capability, thereby consistently improving the functional correctness and robustness of the generated code. All code and data are released to foster reproducibility and advance community research at https://github.com/deep-diver/TAROT.

Reinforcement learning has emerged as a powerful paradigm for unlocking reasoning capabilities in language models. However, relying on sparse rewards makes this process highly sample-inefficient, as models must navigate vast search spaces with minimal feedback. While classic curriculum learning aims to mitigate this by ordering data based on complexity, prior works have primarily targeted small datasets and do not directly transfer to the large-scale settings typical of modern language model training. Furthermore, the right ordering for a specific model is often unclear. To address this, we propose Goldilocks, an adaptive data-selection strategy that uses a Selector network to predict the standard deviation of rewards across the model's rollouts for each candidate question. The Selector prioritizes questions with high predicted reward variability, corresponding to questions that are neither too easy nor too hard for the model's current capabilities (Goldilocks principle), while training the model with GRPO. By leveraging the model's performance on seen samples, the Selector continuously adapts to the model's evolving abilities. Across the OpenMathReasoning and Polaris datasets, Goldilocks consistently improves over standard GRPO, requiring up to 78% fewer optimization steps to reach the corresponding GRPO performance.

Group Relative Policy Optimization (GRPO) remains the dominant critic-free approach for fine-tuning LLMs and VLMs, but its compatibility with constrained policy optimization (e.g. for safety-critical domains) has not been carefully examined. In this work, we introduce Constrained GRPO, a Lagrangian-based extension of GRPO for constrained policy optimization. We show that the standard practice of scalarizing rewards before normalization introduces a critical Lagrangian-specific failure mode: GRPO's within-group normalization makes constrained optimization highly sensitive to how multi-component learning signals are aggregated. We show that scalarizing rewards before normalization introduces shared-denominator coupling, so that changing one multiplier alters not only the emphasis on its corresponding constraint, but also the relative weighting of the reward and other constraints. We address this with a simple but crucial modification: scalarizing standardized advantages rather than rewards. This yields a better-conditioned update by addressing the coupling induced by reward scalarization, resulting in better-behaved multiplier dynamics and more stable constraint enforcement in practice. Empirically, across a controlled gridworld, a real-world autonomous driving benchmark, and a mathematical reasoning task, Constrained GRPO consistently achieves better adherence to specified constraints while maintaining or improving task performance.

GRPO-style reinforcement learning (RL)-based LLM fine-tuning algorithms have recently gained popularity. Relying on heuristic trust-region approximations, however, they can lead to brittle optimization behavior, as global importance-ratio clipping and group-wise normalization fail to regulate samples whose importance ratios fall outside the clipping range. We propose Query-Adaptive Trust-Region policy Optimization (QUATRO), which directly enforces trust-region constraints through a principled optimization. This yields a clear and interpretable objective that enables explicit control over policy updates and stable, entropy-controlled optimization, with a stabilizer terms arising intrinsically from the exact trust-region formulation. Empirically verified on diverse mathematical reasoning benchmarks, QUATRO shows stable training under increased policy staleness and aggressive learning rates, maintaining well-controlled entropy throughout training.

We introduce OpenTinker, an open infrastructure for training large language model (LLM) agents with many LoRA-backed policies over shared execution resources. Modern agent workloads mix supervised fine-tuning (SFT), online reinforcement learning (RL), rollout generation, validation, and multi-turn environment interaction. In such workloads, LoRA adapters are not static inference artifacts: they are frequently updated policy states whose optimizer state, rollout snapshot, and training data attribution must remain consistent. OpenTinker centers the runtime around this policy lifecycle. Users define environments, agents, and learning objectives, while the system manages training clients, rollout samplers, checkpoint handles, and policy-version refresh. The same data path supports SFT and RL by converting trajectories into token sequences with explicit masks: context and environment observations condition the model, while generated action tokens carry supervised weights or RL advantages. This design enables multi-LoRA SFT/RL training in which many users, tasks, or agents can share a base model while keeping adapter updates, checkpoints, and rollout snapshots isolated. We describe the system architecture, the adapter lifecycle, the service-backed snapshot handoff used by the current implementation, the backend contract for mixed-adapter rollout kernels, and the training scheduler that isolates adapter-local gradients and optimizer state. Representative validation tasks exercise single-turn, multi-turn, LoRA, and multi-agent agentic training.

Negative learning has never been more important than it is today, particularly in large language model (LLM) fine-tuning, where it underpins machine unlearning, preference- and reward-based optimization, and exploratory data generation. Yet its functional role remains a subject of ongoing debate, with existing work variously characterizing it as forgetting, regularization, or target shaping. However, analytical support focuses primarily on the single-update scale, while empirical results at the trajectory scale are typically obtained under negative-only training and may not extend to composite objectives in which it interacts with other components. To develop a deeper understanding, we introduce a gradient-based endpoint probing framework that can examine negative updates separately throughout training. Our analysis reveals that negative learning plays different roles across algorithms and training phases. For example, in off-policy DPO, it acts more like a regularizer early in training and shows evidence of target shaping only toward out-of-distribution endpoints and only in later phases. In on-policy PPO and GRPO, by contrast, negative updates offset endpoint-aligned positive learning without stable target shaping, consistent with a role in sustaining exploration. Guided by these findings, we develop a family of stage-aware methods that modulate the strength of negative updates throughout training, improving fine-tuning performance across tasks.

Vision-Language-Action (VLA) models such as OpenVLA, Octo, and $π_0$ have shown strong generalization by leveraging large-scale demonstrations, yet their performance is still fundamentally constrained by the quality and coverage of supervised data. Reinforcement learning (RL) provides a promising path for improving and fine-tuning VLAs through online interaction. However, conventional policy gradient methods are computationally infeasible in the context of flow-matching based models due to the intractability of the importance sampling process, which requires explicit computation of policy ratios. To overcome this limitation, we propose Flow Policy Optimization (FPO) algorithm, which reformulates importance sampling by leveraging per-sample changes in the conditional flow-matching objective. Furthermore, FPO achieves stable and scalable online reinforcement fine-tuning of the $π_0$ model by integrating structure-aware credit assignment to enhance gradient efficiency, clipped surrogate objectives to stabilize optimization, multi-step latent exploration to encourage diverse policy updates, and a Q-ensemble mechanism to provide robust value estimation. We evaluate FPO on the LIBERO benchmark and the ALOHA simulation task against supervised, preference-aligned, diffusion-based, autoregressive online RL, and $π_0$-FAST baselines, observing consistent improvements over the imitation prior and strong alternatives with stable learning under sparse rewards. In addition, ablation studies and analyses of the latent space dynamics further highlight the contributions of individual components within FPO, validating the effectiveness of the proposed computational modules and the stable convergence of the conditional flow-matching objective during online RL.

Diffusion large language models (dLLMs) are promising alternatives to autoregressive large language models (AR-LLMs), as they potentially allow higher inference throughput. Reinforcement learning (RL) is crucial to enabling dLLMs to achieve performance comparable to that of AR-LLMs on important tasks, such as reasoning. However, RL algorithms well-suited to dLLMs' unique characteristics have yet to be developed. This paper proposes Distribution Matching Policy Optimization (DMPO), a principled and theoretically grounded RL fine-tuning method specifically designed to enhance the reasoning capabilities of dLLMs by matching the dLLM policy distribution to the optimal, reward-tilted one through cross-entropy optimization. We identify a key implementation challenge with small training batch sizes and propose several effective solutions based on a novel weight baseline subtraction technique. DMPO exhibits superior performance on multiple reasoning benchmarks without supervised fine-tuning, achieving up to a $39.63$ percentage-point improvement in accuracy over prior non-DMPO RL baselines and $67.97$ percentage points over the base model, underscoring the effectiveness of the distribution-matching framework. Our code is available at https://github.com/yuchen-zhu-zyc/DMPO.

Recent advances in large video-language models (VLMs) rely on extensive fine-tuning techniques that strengthen alignment between textual and visual comprehension. Leading pipelines typically pair supervised fine-tuning (SFT) with reinforcement learning from preference data to enhance video comprehension. However, as VLMs scale in parameter size, so does the cost of gathering enough human feedback. To make fine-tuning more cost-effective, recent frameworks explore reinforcement learning with AI feedback (RLAIF), which replace human preference with AI as a judge. Current RLAIF frameworks rely on a specialized reward model trained with video narratives to create calibrated scalar rewards -- an expensive and restrictive pipeline. We propose Oracle-RLAIF, a novel framework that replaces the trained reward model with a more general Oracle ranker which acts as a drop-in model ranking candidate model responses rather than scoring them. Alongside Oracle-RLAIF, we introduce $GRPO_{rank}$, a novel rank-based loss function based on Group Relative Policy Optimization (GRPO) that directly optimizes ordinal feedback with rank-aware advantages. Empirically, we demonstrate that Oracle-RLAIF consistently outperforms leading VLMs using existing fine-tuning methods when evaluated across various video comprehension benchmarks. Oracle-RLAIF paves the path to creating flexible and data-efficient frameworks for aligning large multi-modal video models with reinforcement learning from rank rather than score.

Offline-to-online reinforcement learning (RL) has emerged as a practical paradigm that leverages offline datasets for pretraining and online interactions for fine-tuning. However, its empirical behavior is highly inconsistent: design choices of online fine-tuning that work well in one setting can fail completely in another. Guided by the stability--plasticity principle, we propose a framework that can explain this inconsistency: We argue that efficient fine-tuning must preserve the utility of the stronger offline prior, whether that is the pretrained policy or the offline dataset, while maintaining sufficient plasticity. This perspective identifies three regimes of online fine-tuning, each requiring distinct stability properties. We validate this framework through a large-scale empirical study, finding that the results strongly align with its predictions in 45 out of 63 cases, with only 3 opposite mismatches. This work provides a framework for guiding design choices in offline-to-online RL based on the relative performance of the offline dataset and the pretrained policy.

Fine-tuning large language models (LLMs) for downstream tasks is an essential stage of modern AI deployment. Reinforcement learning (RL) has emerged as the dominant fine-tuning paradigm, underpinning many state-of-the-art LLMs. In contrast, evolution strategies (ES) has largely been overlooked due to the widespread belief that it does not scale to modern model sizes. This paper overturns this assumption by demonstrating the first successful application of ES to full-parameter fine-tuning of LLMs at the billion-parameter scale, without dimensionality reduction. ES can indeed search over extremely high-dimensional parameter spaces and outperform established RL implementations across multiple axes, including improved tolerance to long-horizon and delayed rewards, robustness across diverse base LLMs, reduced susceptibility to reward hacking, and improved training stability. These findings suggest that ES is not merely a viable alternative to RL, but a fundamentally different and powerful backpropagation-free post-training paradigm that opens a new direction for LLM fine-tuning beyond current RL-based approaches.

Advanced reasoning in LLMs on challenging domains like mathematical reasoning can be tackled using verifiable rewards based reinforced fine-tuning (ReFT). In standard ReFT frameworks, a behavior model generates multiple completions with answers per problem, for the answer to be then scored by a reward function. While such RL post-training methods demonstrate significant performance improvements across challenging reasoning domains, the computational cost of generating completions during training with multiple inference steps makes the training cost non-trivial. To address this, we draw inspiration from off-policy RL, and speculative decoding to introduce a novel ReFT framework, dubbed Nested-ReFT, where a subset of layers of the target model acts as the behavior model to generate off-policy completions during training. The behavior model configured with dynamic layer skipping per batch during training decreases the inference cost compared to the standard ReFT frameworks. Our theoretical analysis shows that Nested-ReFT yields unbiased gradient estimates with controlled variance. Our empirical analysis demonstrates improved computational efficiency measured as tokens/sec across multiple math reasoning benchmarks and model sizes. Additionally, we explore three variants of bias mitigation to minimize the off-policyness in the gradient updates that allows for maintaining performance that matches the baseline ReFT performance.

Continual post-training (CPT) is a popular and effective technique for adapting foundation models like multimodal large language models to ever-evolving downstream tasks. While existing research primarily focuses on methods like data replay, model expansion, or parameter regularization, the fundamental role of the learning paradigm remains largely unexplored. This paper presents a comparative analysis of two core post-training paradigms: supervised fine-tuning (SFT) and reinforcement fine-tuning (RFT), investigating their respective impacts on knowledge retention during CPT. Our experiments are conducted across multiple multimodal tasks, utilizing Qwen2.5-VL-7B-Instruct as the base model. The investigation yields two significant findings: (1) When continuously learning on downstream tasks, SFT leads to catastrophic forgetting of previously learned tasks. In contrast, RFT inherently preserves prior knowledge and achieves performance comparable to multi-task training. (2) RFT successfully protects and even enhances the model's general knowledge on standard benchmarks, while SFT degrades general model capabilities severely. Further analysis reveals that this stability is not primarily due to explicit mechanisms like KL penalty or chain-of-thought reasoning. We investigate RFT's learning dynamics and find that its selective update mechanism inherently prevents interference with established knowledge. Based on this insight, we propose a rollout-based instance filtering algorithm (RIF-RFT) that enhances the training efficiency of RFT by focusing on learnable samples. Our comprehensive study demonstrates the superiority of RFT as a robust paradigm for continual post-training.

Large Language Model (LLM)-based Multi-Agent Systems (LaMAS) offer a promising approach to collaborative tasks, but applying reinforcement learning to these systems remains challenging because classical multi-agent reinforcement learning assumptions do not fully capture their dependency structures, heterogeneous roles, and autoregressive outputs. This article provides a tutorial-style overview of the transition from reinforcement learning to reinforcement fine-tuning in single-agent and multi-agent settings, highlighting the differences between classical MARL and LLM-based multi-agent optimization. We introduce Multi-Agent Reinforcement Fine-Tuning (MARFT) and formalize LaMAS optimization as a Flexible Markov Game (Flex-MG). The MARFT framework combines predecessor-conditioned policy optimization, a centralized critic, parameter-efficient adapters, and token-level credit assignment. Our token-level analysis clarifies how within-action discounting affects returns and establishes consistency with action-level returns when the token discount is one. Experiments on fixed two- to four-agent mathematical and coding workflows compare MARFT with and without policy parameter sharing against IPPO. Both variants report higher mean evaluation scores in all 24 benchmark-team-size settings, while the variant without parameter sharing shows the clearest advantage in final training reward. Role-specification ablations examine how prescribed roles affect training. We conclude with application perspectives and open challenges, including dynamic workflows, sample efficiency, and system-level reliability. Code: https://github.com/jwliao-ai/MARFT.

Reinforcement finetuning (RFT) has shown great potential for enhancing the mathematical reasoning capabilities of large language models (LLMs), but it is often sample- and compute-inefficient, requiring extensive training. In this work, we introduce AdaRFT (Adaptive Curriculum Reinforcement Finetuning), a method that significantly improves the efficiency of RFT through adaptive curriculum learning. AdaRFT dynamically adjusts the difficulty of training problems based on the model's recent reward signals, ensuring that the model consistently trains on tasks that are challenging but solvable. This adaptive sampling strategy accelerates learning by maintaining an optimal difficulty range, avoiding wasted computation on problems that are too easy or too hard. AdaRFT requires only a lightweight extension to standard RFT algorithms like Proximal Policy Optimization (PPO), without modifying the reward function or model architecture. Experiments on competition-level math datasets demonstrate that AdaRFT improves convergence efficiency and reasoning performance. Given problem-level difficulty annotations, AdaRFT reduces RFT training time by up to 2 times across data distributions and model scales, offering a more scalable and effective RFT framework.