Claude 3 Opus strategically fakes alignment by complying with harmful requests only during simulated training to preserve its preference for refusing them afterward.
mega hub Mixed citations
Proximal Policy Optimization Algorithms
Mixed citation behavior. Most common role is background (52%).
1482
Pith papers citing it
Background
52% of classified citations
abstract
We propose a new family of policy gradient methods for reinforcement learning, which alternate between sampling data through interaction with the environment, and optimizing a "surrogate" objective function using stochastic gradient ascent. Whereas standard policy gradient methods perform one gradient update per data sample, we propose a novel objective function that enables multiple epochs of minibatch updates. The new methods, which we call proximal policy optimization (PPO), have some of the benefits of trust region policy optimization (TRPO), but they are much simpler to implement, more general, and have better sample complexity (empirically). Our experiments test PPO on a collection of benchmark tasks, including simulated robotic locomotion and Atari game playing, and we show that PPO outperforms other online policy gradient methods, and overall strikes a favorable balance between sample complexity, simplicity, and wall-time.
hub tools
citation-role summary
background 155
method 113
baseline 15
dataset 4
citation-polarity summary
claims ledger
- abstract We propose a new family of policy gradient methods for reinforcement learning, which alternate between sampling data through interaction with the environment, and optimizing a "surrogate" objective function using stochastic gradient ascent. Whereas standard policy gradient methods perform one gradient update per data sample, we propose a novel objective function that enables multiple epochs of minibatch updates. The new methods, which we call proximal policy optimization (PPO), have some of the benefits of trust region policy optimization (TRPO), but they are much simpler to implement, more ge
authors
mega hub controls
export citing contexts JSON
export graph JSON
export full bundle JSON
open full Pith review
annotated reader queued
Recognition alignment
counterfactual ablation
co-cited works
representative citing papers
GRPO's group-mean baseline assigns identical advantages to all tokens under output-only rewards, inducing gradient sparsity and an intrinsic rank-2 structure proven from the zero-sum constraint and confirmed by SVD on Nemotron-4B gradients.
Negative narrative immersion causes 12-31% drops in LLM moral accuracy and produces structured shifts that appear in downstream applications.
Dynamic isotropy, quantifying uniform center-of-mass acceleration capability, improves robot performance and enables omnidirectional locomotion, terrain traversal, and failure resilience in a spherical robot design.
AtomComposer uses online RL with multi-composition training to discover up to 10x more valid 3D isomers on unseen chemical formulas than single-composition baselines.
Agent-BRACE improves LLM agent performance on long-horizon partially observable tasks by 5.3-14.5% through a decoupled belief state of verbalized atomic claims with certainty labels that keeps context length constant.
SimWorld Studio deploys an evolving coding agent to create adaptive 3D environments that co-evolve with embodied learners, delivering 18-point success-rate gains over fixed environments in navigation benchmarks.
ReLibra uses pre-known token-to-expert routing from RL rollouts to perform inter-batch expert reordering and intra-batch replication, delivering up to 1.6x higher throughput than Megatron-LM and 1.2x over oracle-equipped EPLB while staying within 6-10% of an ideal balanced baseline.
Weak-to-strong generalization is nearly inevitable in linear logistic regression for most student-teacher pairs without any model capacity mismatch.
Observation and action delays are formally equivalent in cooperative Dec-POMDPs, yielding identical optimal solutions and enabling zero-shot transfer, though learning dynamics differ due to credit assignment and operational constraints.
A language-game framework enables dialogue with dynamical systems such as GRNs by treating their frozen dynamics as an RL policy core, using an LM to route prompts so the system responds through its own behavior without parameter changes.
RefereeBench shows that even the strongest video MLLMs reach only around 60% accuracy on multi-sport refereeing tasks and struggle with rule application and temporal grounding.
OP-GRPO is the first off-policy GRPO method for flow-matching models that reuses trajectories via replay buffer and importance sampling corrections, matching on-policy performance with 34.2% of the training steps.
User-turn generation reveals that LLMs' interaction awareness is largely decoupled from task accuracy, remaining near zero in deterministic settings even as accuracy scales to 96.8% on GSM8K.
Derives an exact unbiased policy gradient for RL post-training of diffusion LLMs via entropy-guided step selection and one-step denoising rewards, achieving state-of-the-art results on coding and logical reasoning benchmarks.
A certified gradient-based method for contact-rich manipulation that quantifies smoothing-induced errors via set-valued discrepancies and incorporates them into analytical reachable sets for robust affine feedback policies.
First in-orbit demonstration of a DRL-trained AI satellite attitude controller that performs robust inertial pointing after sim-to-real transfer.
Develops and tests the first effective safeguard for analytic gradient-based provably safe RL, showing safe training on three control tasks without performance loss.
Flow-GRPO is the first online RL method for flow matching models, raising GenEval accuracy from 63% to 95% and text-rendering accuracy from 59% to 92% with little reward hacking.
SWE-Gym supplies 2438 executable real-world Python tasks to train SWE agents and verifiers, yielding up to 19% gains and new open-weight SOTA of 32% on SWE-Bench Verified.
BEHAVIOR-1K introduces a benchmark of 1,000 human everyday activities in realistic simulated scenes together with the OMNIGIBSON physics simulator to evaluate embodied AI.
ORPO performs preference alignment during supervised fine-tuning via a monolithic odds ratio penalty, allowing 7B models to outperform larger state-of-the-art models on alignment benchmarks.
STEMGym benchmark demonstrates that perception pipelines dominate dose efficiency in autonomous STEM over navigation methods across 33 agent setups.
Proposes Monotonic Inference Policy Improvement (MIPI) objective and MIPU two-step update framework to address objective misalignment between training and inference policies in LLM reinforcement learning.
citing papers explorer
-
Does RL Expand the Capability Boundary of LLM Agents? A PASS@(k,T) Analysis
RL expands the capability boundary of LLM agents on compositional tool-use tasks, shown by non-converging pass curves at large k with increasing T, while SFT regresses it and the effect is absent on simpler tasks.
-
Don't Let the Video Speak: Audio-Contrastive Preference Optimization for Audio-Visual Language Models
Audio-Contrastive Preference Optimization (ACPO) mitigates audio hallucination in AVLMs via output-contrastive and input-contrastive objectives that enforce faithful audio grounding.
-
Character Beyond Speech: Leveraging Role-Playing Evaluation in Audio Large Language Models via Reinforcement Learning
RoleJudge is a multidimensional evaluation framework for speech-character alignment in audio LLMs, backed by the RoleChat dataset and multi-stage RL training with standard alignment to reduce reward issues.
-
Foresight Optimization for Strategic Reasoning in Large Language Models
FoPO trains LLMs for strategic reasoning by combining self-interest with opponent modeling in policy optimization, yielding gains on two new datasets and better out-of-domain generalization than standard baselines.
-
Pareto-Optimal Offline Reinforcement Learning via Smooth Tchebysheff Scalarization
STOMP extends direct preference optimization to the multi-objective setting via smooth Tchebysheff scalarization and standardization of observed rewards, achieving highest hypervolume in eight of nine protein engineering evaluations.
-
Evolving the Complete Muscle: Efficient Morphology-Control Co-design for Musculoskeletal Locomotion
Spectral Design Evolution co-optimizes complete muscle morphology (strength, velocity, stiffness) and control, achieving better efficiency and stability than fixed-morphology baselines on locomotion tasks.
-
Teaching LLMs Human-Like Editing of Inappropriate Argumentation via Reinforcement Learning
Reinforcement learning with a multi-part reward teaches LLMs to output independent, meaning-preserving sentence edits that raise argument appropriateness close to full rewriting.
-
Forecasting the Past: Gradient-Based Distribution Shift Detection in Trajectory Prediction
A gradient norm from a post-hoc self-supervised trajectory forecasting decoder detects distribution shifts in prediction models, with reported improvements on Shifts and Argoverse datasets.
-
Autonomous Diffractometry Enabled by Visual Reinforcement Learning
A model-free reinforcement learning agent learns to align crystals from diffraction images without human supervision or theoretical knowledge.
-
To Learn or Not to Learn: A Litmus Test for Using Reinforcement Learning in Control
A litmus test based on reachset-conformant model identification and correlation analysis of uncertainties predicts if RL-based control is superior to model-based control without any RL training.
-
Low-rank Optimization Trajectories Modeling for LLM RLVR Acceleration
NExt accelerates RLVR training for LLMs by nonlinearly extrapolating low-rank parameter trajectories extracted from LoRA runs.
-
DDO-RM: Distribution-Level Policy Improvement after Reward Learning
DDO-RM turns reward scores into a target distribution and applies KL-regularized mirror-descent projection on finite candidates to improve policies, outperforming DPO on Pythia-410M.
-
Optimal Stability of KL Divergence under Gaussian Perturbations
KL divergence between a general distribution and a perturbed Gaussian reference remains stable with an optimal sqrt(ε) degradation rate under finite second-moment conditions.
-
Robust Adversarial Policy Optimization Under Dynamics Uncertainty
RAPO uses a dual robust RL formulation with trajectory-level adversarial networks and model-level Boltzmann reweighting over dynamics ensembles to improve policy resilience and out-of-distribution generalization while keeping the problem tractable.
-
ScoRe-Flow: Complete Distributional Control via Score-Based Reinforcement Learning for Flow Matching
ScoRe-Flow achieves decoupled mean-variance control in stochastic flow matching by deriving a closed-form score for drift modulation plus learned variance, yielding faster RL convergence and higher success rates on locomotion and manipulation benchmarks.
-
Bringing Value Models Back: Generative Critics for Value Modeling in LLM Reinforcement Learning
GenAC introduces generative critics with chain-of-thought reasoning and in-context conditioning to improve value approximation and downstream RL performance in LLMs compared to value-based and value-free baselines.
-
Physics-Informed Reinforcement Learning of Spatial Density Velocity Potentials for Map-Free Racing
A DRL policy learns racing controls from depth spectral distributions using a non-geometric physics-informed reward, achieving 12% better performance than humans on out-of-distribution tracks with under 1% of baseline computation.
-
SafeAdapt: Provably Safe Policy Updates in Deep Reinforcement Learning
SafeAdapt certifies a Rashomon set of safe policies from demonstration data and projects updates from arbitrary RL algorithms onto it to guarantee preservation of safety on source tasks.
-
Learning Vision-Language-Action World Models for Autonomous Driving
VLA-World improves autonomous driving by using action-guided future image generation followed by reflective reasoning over the imagined scene to refine trajectories.
-
Advantage-Guided Diffusion for Model-Based Reinforcement Learning
Advantage-guided diffusion (SAG and EAG) steers sampling in diffusion world models to higher-advantage trajectories, enabling policy improvement and better sample efficiency on MuJoCo tasks.
-
Plasticity-Enhanced Multi-Agent Mixture of Experts for Dynamic Objective Adaptation in UAVs-Assisted Emergency Communication Networks
PE-MAMoE combines sparsely gated mixture-of-experts actors with a non-parametric phase controller in MAPPO to maintain plasticity under dynamic user mobility and traffic, yielding 26.3% higher normalized IQM return in simulations.
-
Training Language Models for Bilateral Trade with Private Information
Frontier LLMs achieve higher surplus via sequential price discrimination in bilateral trade simulations, while SFT followed by GRPO on Qwen models trades off surplus gains against deal rates and improves consistency across price tiers.
-
Demystifying OPD: Length Inflation and Stabilization Strategies for Large Language Models
OPD for LLMs suffers length inflation and repetition collapse; StableOPD uses reference divergence and rollout mixing to prevent it and improve math reasoning performance by 7.2% on average.
-
Beyond Static Forecasting: Unleashing the Power of World Models for Mobile Traffic Extrapolation
MobiWM is a multimodal world model for mobile networks that learns state-action dynamics to enable unlimited-horizon counterfactual traffic simulations and optimization.
-
PRIME: Training Free Proactive Reasoning via Iterative Memory Evolution for User-Centric Agent
PRIME enables agents to proactively reason in user-centric tasks by iteratively evolving structured memories from interaction trajectories without gradient-based training.
-
Robots that learn to evaluate models of collective behavior
A robotic fish learns goal-directed policies in simulation and interacts with live fish to quantify how well different behavioral models match real responses using Wasserstein distances on performance metrics.
-
Android Coach: Improve Online Agentic Training Efficiency with Single State Multiple Actions
Android Coach improves online agent training efficiency by enabling multiple actions per state via a critic-based coach, process reward model, and group-wise advantage estimation, delivering 7.5-8.3% success rate gains and 1.4x efficiency over PPO/GRPO baselines.
-
A First Guess is Rarely the Final Answer: Learning to Search in the Traveling Salesperson Problem
NICO-TSP learns a 2-opt local search policy for TSP using edge tokens, imitation learning on short trajectories, and critic-free group RL on longer rollouts, yielding more step-efficient improvement and better out-of-distribution generalization than prior neural and heuristic baselines.
-
Equivariant Multi-agent Reinforcement Learning for Multimodal Vehicle-to-Infrastructure Systems
A self-supervised multimodal alignment step plus equivariant GNN-based MARL yields over twofold sensing accuracy and 50% performance gains in decentralized V2I rate maximization.
-
Enhancing MLLM Spatial Understanding via Active 3D Scene Exploration for Multi-Perspective Reasoning
A training-free Visual Chain-of-Thought framework reconstructs high-fidelity 3D meshes from single images and iteratively synthesizes optimal novel views to enhance MLLM spatial comprehension on benchmarks like 3DSRBench.
-
Generate, Filter, Control, Replay: A Comprehensive Survey of Rollout Strategies for LLM Reinforcement Learning
This survey introduces the Generate-Filter-Control-Replay (GFCR) taxonomy to structure rollout pipelines for RL-based post-training of reasoning LLMs.
-
Discrete Flow Matching Policy Optimization
DoMinO reformulates discrete flow matching sampling as an MDP for unbiased RL fine-tuning with new TV regularizers, yielding better enhancer activity and naturalness on DNA design tasks.
-
Cayley Graph Optimization for Scalable Multi-Agent Communication Topologies
CayleyTopo uses reinforcement learning to optimize Cayley graph generators for lower diameter, yielding faster and more resilient information flow in multi-agent systems than hand-crafted sparse topologies.
-
An Iterative Test-and-Repair Framework for Competitive Code Generation
FixAudit improves LLM code generation on competitive programming benchmarks by training a shared model for iterative code-aware test generation and repair, achieving 35%+ gains in Pass@1 over baselines on the same 7B model.
-
ALTO: Adaptive LoRA Tuning and Orchestration for Heterogeneous LoRA Training Workloads
ALTO accelerates LoRA tuning up to 13.8x by monitoring loss trajectories for early stopping, using fused grouped GEMM with rank-local adapter parallelism, and combining intra- and inter-task scheduling for heterogeneous workloads without quality loss.
-
Neural Assistive Impulses: Synthesizing Exaggerated Motions for Physics-based Characters
A hybrid neural policy operating in impulse space enables physics-based characters to track exaggerated, dynamically infeasible motions that standard DRL methods cannot stabilize.
-
Right at My Level: A Unified Multilingual Framework for Proficiency-Aware Text Simplification
Re-RIGHT trains a 4B policy model with vocabulary coverage, semantic preservation, and coherence rewards to perform proficiency-aware lexical simplification in four languages without parallel corpora.
-
Vehicle-as-Prompt: A Unified Deep Reinforcement Learning Framework for Heterogeneous Fleet Vehicle Routing Problem
VaP-CSMV uses a cross-semantic encoder and multi-view decoder to unify DRL solving of HFVRP variants, outperforming prior neural solvers while matching heuristics at much lower inference time and generalizing zero-shot to unseen scales.
-
Retrieval Augmented Conversational Recommendation with Reinforcement Learning
RAR retrieves candidate items from a 300k-movie corpus then uses LLM generation with RL feedback to produce context-aware recommendations that outperform baselines on benchmarks.
-
Many Preferences, Few Policies: Towards Scalable Language Model Personalization
PALM produces a small portfolio of LLMs that contains a near-optimal model for any user preference weight vector, with theoretical bounds on portfolio size and approximation quality.
-
Territory Paint Wars: Diagnosing and Mitigating Failure Modes in Competitive Multi-Agent PPO
PPO in a new competitive game fails due to five implementation bugs and then competitive overfitting where self-play stays near 50% but generalization drops to 21.6%; mixing 20% random opponents restores generalization to 77.1%.
-
Improving Feasibility via Fast Autoencoder-Based Projections
An adversarially trained autoencoder learns a convex latent space to enable rapid approximate projections that enforce nonconvex constraints in optimization and reinforcement learning.
-
Self-Distilled RLVR
RLSD mixes self-distillation for token-level policy difference magnitudes with RLVR for reliable update directions from response correctness to reach higher convergence and better training stability.
-
A semicontinuous relaxation of Saito's criterion and freeness as angular minimization
A new functional S vanishes precisely on free line arrangements and enables discovery of verified free examples for every admissible exponent pair with up to 20 lines.
-
Auction-Based Online Policy Adaptation for Evolving Objectives
Auction-based bidding among selfish local policies in a general-sum Markov game enables dynamic adaptation to evolving multi-objectives in reinforcement learning with Nash equilibrium guarantees.
-
Stochastic Policy Gradient Methods in the Uncertain Volatility Model
A neural-network actor-critic policy gradient algorithm with squashed Gaussian C-vine policies solves high-dimensional robust pricing problems in the uncertain volatility model and outperforms existing Monte Carlo and ML benchmarks in numerical tests.
-
Think Anywhere in Code Generation
Think-Anywhere lets LLMs invoke on-demand reasoning at any token during code generation via cold-start imitation followed by outcome-based RL, reaching state-of-the-art results on LeetCode, LiveCodeBench, HumanEval, and MBPP.
-
What If Consensus Lies? Selective-Complementary Reinforcement Learning at Test Time
SCRL adds selective positive pseudo-labeling and entropy-gated negative pseudo-labeling to test-time RL, reducing noise from weak consensus and improving LLM reasoning on benchmarks.
-
Featurized Occupation Measures for Structured Global Search in Numerical Optimal Control
Featurized Occupation Measures create a primal-dual framework that couples explicit HJB subsolutions with scalable trajectory optimization, proving asymptotic consistency and shifting dimensionality limits to system interconnection topology.
-
Topo-R1: Detecting Topological Anomalies via Vision-Language Models
Topo-R1 fine-tunes a vision-language model using a topology-aware reward and GRPO to detect anomalies such as broken or spurious connections in tubular segmentation masks, outperforming standard VLMs.