Experience Replay

Conventional federated learning assumes that greater learner participation improves training performance, by leveraging abundant, independently generated local data. However, in federated reinforcement learning (FRL) for unmanned aerial vehicle (UAV) teams in hazardous environments where experience generation is severely constrained by safety considerations, energy limitations, and mission duration, this assumption may break. This work introduces Experience-Constrained Hierarchical Federated Reinforcement Learning (EC-HFRL), a framework in which clusters act as federated learning agents, while multiple intra-cluster learners represent parallel learning resources that reuse a shared experience pool. We show that increasing participation does not necessarily improve learning performance. Instead, learning performance is strongly associated with experience reuse strategy and the dominance of key analytically identified gradient transition experiences within a cluster. In particular, minibatch size primarily determines effective replay exposure, while higher intra-cluster participation increases reuse level. Empirical results demonstrate that the performance regimes are strongly associated with the structure of the learning signal, rather than federated aggregation effects, clarifying the limited and secondary role of learner participation in experience-constrained FRL.

Deploying reinforcement learning policies in the real world requires adapting to time-varying environments. We study this problem in the contextual Markov Decision Process (cMDP) framework, where a family of environments is indexed by a low-dimensional context unknown at test time. The standard approach decomposes the problem: train a so-called "universal policy" which assumes knowledge of the true context, then pair it with a context estimator which approximates context using the observed trajectory. We identify a simple, counterintuitive trick that substantially improves the estimator: randomly delete a fraction of the training buffer after each round. This works because data is collected across multiple rounds using progressively better policies, and older trajectories come from a different distribution than what the estimator will face at deployment time; random deletion creates an implicit exponential decay on older data while preserving diversity without requiring any explicit identification of which samples are stale. This reduces robustness gap by 30% for MLPs and by 6% on average for recurrent networks. Strikingly, it allows a narrow MLP with 5x fewer parameters to outperform a wide MLP trained without deletion. To understand when and why deletion helps, we analyze regularized empirical risk minimization with a mismatch between the train distribution and the distribution at deployment; in this idealized setting, we prove that removing a single uniformly random training point decreases expected test loss in expectation under mild conditions. For ridge regression we make this quantitative: deletion helps when the regularization coefficient is moderate and the signal-to-noise ratio (SNR) is sufficiently low, and, crucially, this SNR threshold gives a direct measure of how large the distribution mismatch between training and deployment must be for deletion to be beneficial.

In reinforcement learning (RL) for robotic manipulation, the Decision Transformer (DT) has emerged as an effective framework for addressing long-horizon tasks. However, DT's performance depends heavily on the coverage of collected experiences. Without an active exploration mechanism, standard DT relies on uniform replay, which leads to poor sample efficiency, limited exploration, and reduced overall effectiveness. At the same time, while excessive exploration can help avoid local optima, it often delays policy convergence and leads to degraded efficiency. To address these limitations, we propose E$^2$DT, a DT-guided k-Determinantal Point Process sampling framework that enables the model to actively shape its own experience selection. Our framework is experience-aware, allowing E$^2$DT to be both efficient, by prioritizing sampling quality, such as high-return, high-uncertainty, and underrepresented trajectories, and effective, by ensuring diversity across trajectory windows to preserve policy optimality. Specifically, DT's internal latent embeddings measure diversity across trajectory windows, while quality is quantified through a composite metric that integrates return-to-go (RTG) quantiles, predictive uncertainty, and stage coverage based on inverse frequency. These two dimensions are integrated into a novel quality-diversity joint kernel that prioritizes the most informative experiences, thereby enabling learning that is both efficient and effective. We evaluate E$^2$DT on challenging robotic manipulation benchmarks in both simulation and real-robot settings. Results show that it consistently outperforms prior methods. These findings demonstrate that coupling policy learning with experience-aware sampling provides a principled path toward robust long-horizon robotic learning.

Deep reinforcement learning for quantum circuit optimization faces three bottlenecks: replay buffers that overlook temporal difference (TD) target reliability, curriculum-based architecture search requiring a full quantum-classical evaluation after every edit, and the discard of noiseless trajectories when retraining under hardware noise. We address these limitations by treating replay as a central algorithmic lever. We introduce ReaPER+, an annealed replay rule that transitions from TD-error prioritization to reliability-aware sampling as value estimates mature. ReaPER+ achieves up to 4x higher sample efficiency than fixed PER, ReaPER, and uniform replay, while matching prior on-policy solution quality with up to 32x fewer interactions At 12 qubits, fixed ReaPER reaches the lowest energy error in the fewest steps, while PER and uniform replay find more compact circuits at higher error. On tasks scaling to 20 qubits, ReaPER+ retains its advantage, demonstrating that reliability-aware annealing extends beyond small-system benchmarks. LunarLander-v3 confirms that the ReaPER+ is domain-agnostic, it improves success rates by up to 26.8% over PER and 21.8% over fixed ReaPER, with a 3% AUC gain over both. We further introduce OptCRLQAS, which amortizes quantum-classical evaluations across multiple architectural edits, reducing training wall-clock time by up to 67.5% on 12-qubit without degrading solution quality. Finally, lightweight replay-buffer transfer warm-starts noisy optimization from noiseless trajectories, without weight transfer or $ε$-greedy pretraining, reducing steps to chemical accuracy by 85-90% and final energy error by up to 90% relative to from-scratch learning. Transfer gains increase with system size. Together, these results establish experience storage, sampling, and transfer as decisive levers for sample efficient, noise-aware quantum circuit optimization.

Continual learning enables models to acquire new knowledge over time while retaining previously learned capabilities. However, its application to text-to-3D generation remains unexplored. We present ReConText3D, the first framework for continual text-to-3D generation. We first demonstrate that existing text-to-3D models suffer from catastrophic forgetting under incremental training. ReConText3D enables generative models to incrementally learn new 3D categories from textual descriptions while preserving the ability to synthesize previously seen assets. Our method constructs a compact and diverse replay memory through text-embedding k-Center selection, allowing representative rehearsal of prior knowledge without modifying the underlying architecture. To systematically evaluate continual text-to-3D learning, we introduce Toys4K-CL, a benchmark derived from the Toys4K dataset that provides balanced and semantically diverse class-incremental splits. Extensive experiments on the Toys4K-CL benchmark show that ReConText3D consistently outperforms all baselines across different generative backbones, maintaining high-quality generation for both old and new classes. To the best of our knowledge, this work establishes the first continual learning framework and benchmark for text-to-3D generation, opening a new direction for incremental 3D generative modeling. Project page is available at: https://mauk95.github.io/ReConText3D/.

While Experience Replay - the practice of storing rollouts and reusing them multiple times during training - is a foundational technique in general RL, it remains largely unexplored in LLM post-training due to the prevailing belief that fresh, on-policy data is essential for high performance. In this work, we challenge this assumption. We present a systematic study of replay buffers for LLM post-training, formalizing the optimal design as a trade-off between staleness-induced variance, sample diversity and the high computational cost of generation. We show that strict on-policy sampling is suboptimal when generation is expensive. Empirically, we show that a well-designed replay buffer can drastically reduce inference compute without degrading - and in some cases even improving - final model performance, while preserving policy entropy.

Catastrophic forgetting remains a key challenge in Continual Learning (CL). In replay-based CL with severe memory constraints, performance critically depends on the sample selection strategy for the replay buffer. Most existing approaches construct memory buffers using embeddings learned under supervised objectives. However, class-agnostic, self-supervised representations often encode rich, class-relevant semantics that are overlooked. We propose a new method, Multiple Embedding Replay Selection, MERS, which replaces the buffer selection module with a graph-based approach that integrates both supervised and self-supervised embeddings. Empirical results show consistent improvements over SOTA selection strategies across a range of continual learning algorithms, with particularly strong gains in low-memory regimes. On CIFAR-100 and TinyImageNet, MERS outperforms single-embedding baselines without adding model parameters or increasing replay volume, making it a practical, drop-in enhancement for replay-based continual learning.

Post training via GRPO has demonstrated remarkable effectiveness in improving the generation quality of flow-matching models. However, GRPO suffers from inherently low sample efficiency due to its on-policy training paradigm. To address this limitation, we present OP-GRPO, the first Off-Policy GRPO framework tailored for flow-matching models. First, we actively select high-quality trajectories and adaptively incorporate them into a replay buffer for reuse in subsequent training iterations. Second, to mitigate the distribution shift introduced by off-policy samples, we propose a sequence-level importance sampling correction that preserves the integrity of GRPO's clipping mechanism while ensuring stable policy updates. Third, we theoretically and empirically show that late denoising steps yield ill-conditioned off-policy ratios, and mitigate this by truncating trajectories at late steps. Across image and video generation benchmarks, OP-GRPO achieves comparable or superior performance to Flow-GRPO with only 34.2% of the training steps on average, yielding substantial gains in training efficiency while maintaining generation quality.

Recently, reinforcement learning~(RL) has become an important approach for improving the capabilities of large language models~(LLMs). In particular, reinforcement learning from verifiable rewards~(RLVR) has emerged as a promising paradigm for reasoning tasks. However, existing RL-based training still remains only a rough approximation to human learning. Human learners leverage both external and internal experience to guide exploration and gradually internalize useful trajectories into stable knowledge. Motivated by this gap, we ask: how can LLMs better utilize and internalize experience during RLVR training? To answer this question, we propose Dual Guidance Optimization~(DGO), a unified framework that leverages external and internal experience to improve training effectiveness. Specifically, DGO first constructs an experience bank from previously explored trajectories. The policy then performs exploration under the joint guidance of the experience bank and the model's internal knowledge. The resulting trajectories are further used to refine the experience bank and optimize model parameters, forming a closed loop of experience utilization and internalization. Experiments show that DGO consistently outperforms baseline methods, suggesting that better utilization and internalization of experience lead to more effective reasoning.

LLM-agent training pipelines routinely discard failed trajectories even though GPT-4o achieves only 14-20% on WebArena and below 55% pass@1 on ToolBench; even specialised systems at 50-65% leave the majority of trajectories unused. We introduce AgentHER, which recovers this lost signal by adapting Hindsight Experience Replay (HER) to natural-language agent trajectories: a trajectory that fails goal A is often a correct demonstration for an achievable alternative goal B. AgentHER realises this through a four-stage pipeline (failure classification, outcome extraction, LLM-guided relabeling with confidence gating, and data packaging) that converts discarded failures into SFT, DPO, and ShareGPT training data. On WebArena and ToolBench under a strict task-disjoint held-out protocol, AgentHER improves over success-only SFT by +7.6-11.4% across four model families (GPT-4o, Qwen2.5-72B/7B, LLaMA-3.1-8B), achieves 2x sample efficiency, and beats the strongest experience-centric baseline (Agent Workflow Memory) by +3.0-6.2%. Two robustness mechanisms, failure-severity weighting and cross-model multi-judge verification (gpt-4o-mini paired with Qwen2.5-72B-Instruct), reduce label noise from 5.9% to 2.9% and raise human-rated relabeling precision to 97.1% on WebArena and 96.0% on ToolBench. A full system-cost audit shows the entire relabeling pipeline costs 2.98 and 26 wall-clock minutes for 3,000 trajectories, i.e. 1.4 x 10^-3 per accepted pair. Code: https://github.com/alphadl/AgentHER

While Reinforcement Learning (RL) enhances Large Language Model reasoning, on-policy algorithms like GRPO are sample-inefficient as they discard past rollouts. Existing experience replay methods address this by reusing accurate samples for direct policy updates, but this can incur additional storage costs and concentrate the policy on a narrow set of solutions. We argue that historical data should prioritize sustaining local policy plasticity rather than simply reinforcing accuracy. We define local policy plasticity as a persistent ability to remain backtrackable and switchable with respect to recent successful trajectories: throughout training, the policy maintains non-negligible support on recently reachable alternatives, so it can smoothly reallocate probability mass when needed without being locked into a single dominant path. To this end, we propose Dynamic Jensen-Shannon Replay (DyJR), a simple yet effective regularization framework using a dynamic reference distribution from recent trajectories. DyJR introduces two innovations: (1) A Time-Sensitive Dynamic Buffer that uses FIFO and adaptive sizing to retain only temporally proximal samples, synchronizing with model evolution; and (2) Jensen-Shannon Divergence Regularization, which replaces direct replay policy-gradient updates with a regularization signal that discourages premature over-commitment to a single Rank-1 trajectory. Experiments on mathematical reasoning, Text-to-SQL, and coding benchmarks demonstrate that DyJR improves over GRPO as well as baselines such as RLEP and Ex-GRPO, while maintaining training efficiency comparable to the original GRPO. Furthermore, Rank-k token probability evolution and semantic-mode analyses show reduced concentration and greater retention of alternative continuations. Matched admission-objective comparisons and repeated runs provide additional evidence for the roles of DyJR's components.

Traditional on-policy Reinforcement Learning with Verifiable Rewards (RLVR) frameworks suffer from experience waste and reward homogeneity, which directly hinders learning efficiency on difficult samples during large language models post-training. In this paper, we introduce Batch Adaptation Policy Optimization (BAPO), an off-policy RLVR framework to improve the data efficiency in large language models post-training. It dynamically selects training batches by re-evaluating historically difficult samples and reusing high-quality ones, while holding a lower bound guarantee for policy improvement. Extensive experiments further demonstrate that BAPO achieves an average 12.5% improvement over GRPO across mathematics, planning, and visual reasoning tasks. Crucially, BAPO successfully resolves 40.7% of problems that base models consistently fail to solve.

This paper investigates whether sequential context improves goal-conditioned Reinforcement Learning in sparse-reward manipulation tasks. While Hindsight Experience Replay (HER) addresses reward sparsity through goal relabeling, its operation on isolated transitions limits its ability to capture temporal dependencies inherent in joint-space control. We hypothesize that incorporating motion history can enhance policy learning and introduce CONTHER, which integrates a Transformer-based architecture with a modified HER replay buffer. The Transformer encodes sequences of prior states and goals to provide temporal awareness, while the buffer populates experience with artificially successful trajectories. Two architectural variants are analyzed to examine how contextual information should be integrated. In simulated point-reaching tasks with a UR3 manipulator, CONTHER achieves a 38.46% higher average success rate compared to baselines and outperforms the strongest baseline by 28.21%, with faster convergence and more stable learning. The framework is further evaluated on three dynamic tasks requiring complex trajectory following and obstacle avoidance, where temporal context is critical. By operating directly on joint velocities, the approach provides a foundation for transfer to physical systems. The primary contribution is a systematic investigation into fusing sequential context with goal relabeling, offering insights into how temporal awareness benefits policy learning.

Experience replay is widely used to improve learning efficiency in reinforcement learning by leveraging past experiences. However, existing experience replay methods, whether based on uniform or prioritized sampling, often suffer from low efficiency, particularly in real-world scenarios with high-dimensional state spaces. To address this limitation, we propose a novel approach, Efficient Diversity-based Experience Replay (EDER). EDER employs a determinantal point process to model the diversity between samples and prioritizes replay based on the diversity between samples. To further enhance learning efficiency, we incorporate Cholesky decomposition for handling large state spaces in realistic environments. Additionally, rejection sampling is applied to select samples with higher diversity, thereby improving overall learning efficacy. Extensive experiments are conducted on robotic manipulation tasks in MuJoCo, Atari games, and realistic indoor environments in Habitat. The results demonstrate that our approach not only significantly improves learning efficiency but also achieves superior performance in high-dimensional, realistic environments.

Hindsight Experience Replay (HER) enhances sample efficiency in goal-conditioned reinforcement learning (RL) by relabelling failed trajectories with goals that were actually achieved. However, HER assumes access to a goal relabelling function and a predicate function that determines whether a goal has been satisfied. These assumptions break down in instruction-following tasks, where goals are expressed in natural language and differ from the state space. We formalize this as the Hindsight Reinforcement Learning problem, which shows the need to jointly learn these functions alongside the RL policy. To address it, we propose ETHER (Emergent Textual Hindsight Experience Replay), an agent that leverages Emergent Communication. ETHER uses a referential game (RG) to train a speaker and a listener to develop a grounded, artificial language describing environment states. It partially aligns this emergent language with instruction language using co-occurrence patterns between task instructions and RL observations. Experiments on BabyAI's PickupDist task show that ETHER's learned RG speaker and listener can function as the goal relabelling and predicate functions of HER, improving sample efficiency despite imperfect language alignment. Our work bridges Emergent Communication and goal-conditioned RL, opening the door to wider applications of HER.