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%).
1619
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.
RL agent for online LHC trigger threshold tuning improves in-tolerance intervals by 28-56% on Monte Carlo and real CMS data without fine-tuning.
A reward-free representation learning pipeline for offline PbRL achieves better preference efficiency than standard two-stage baselines by connecting RFRL concepts to preference data.
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.
citing papers explorer
-
SPS: Steering Probability Squeezing for Better Exploration in Reinforcement Learning for Large Language Models
SPS interleaves RL and IRL to counteract probability squeezing in LLM reasoning trajectories, improving Pass@k on five benchmarks while identifying an empirical upper bound on multi-sample performance.
-
Incentivizing Parametric Knowledge via Reinforcement Learning with Verifiable Rewards for Cross-Cultural Entity Translation
EA-RLVR boosts Qwen3-14B entity translation accuracy from 23.66% to 31.87% on 50k unseen entities using 7k samples via RL with verifiable rewards, with transfer gains of +1.35 XCOMET on WMT24++.
-
GRAIL: Autonomous Concept Grounding for Neuro-Symbolic Reinforcement Learning
GRAIL autonomously grounds relational concepts in NeSy-RL by using LLM weak supervision followed by interaction-based refinement, matching or exceeding manually defined concepts on Atari games.
-
AutoOR: Scalably Post-training LLMs to Autoformalize Operations Research Problems
AutoOR uses synthetic data generation and RL post-training with solver feedback to enable 8B LLMs to autoformalize linear, mixed-integer, and non-linear OR problems, matching larger models on benchmarks.
-
SocialGrid: A Benchmark for Planning and Social Reasoning in Embodied Multi-Agent Systems
SocialGrid benchmark shows even top LLMs achieve below 60% in embodied planning and task completion, with deception detection near random chance regardless of model scale.
-
Scattered Hypothesis Generation for Open-Ended Event Forecasting
SCATTER uses RL with a hybrid reward combining validity, intra-group diversity, and inter-group diversity to produce inclusive hypothesis sets for event forecasting and outperforms baselines on OpenForecast and OpenEP.
-
GroupDPO: Memory efficient Group-wise Direct Preference Optimization
GroupDPO decouples group-wise preference optimization during backpropagation to cut peak memory while keeping the same gradients, allowing larger groups and consistent gains over single-pair DPO plus an NLL term on positives.
-
Reward Weighted Classifier-Free Guidance as Policy Improvement in Autoregressive Models
Reward-weighted classifier-free guidance approximates Q-function policy improvement in autoregressive models, enabling test-time reward optimization and faster RL convergence via distillation.
-
Efficient $n$-qubit entangling operations via a superconducting quantum router
A superconducting quantum router enables programmable multi-qubit entangling operations, demonstrated with faster preparation of entangled states and RL-trained 2- and 3-qubit gates like Toffoli and Fredkin.
-
On-Line Policy Iteration with Trajectory-Driven Policy Generation
An online policy iteration algorithm produces a sequence of monotonically cost-improving policies for fixed-initial-state deterministic control by training each new policy on the trajectory generated by the prior one.
-
Timescale Separation Enables Deep Reinforcement Learning Control of Rotating Detonation Engine Mode Transitions
Reformulating DRL in a moving reference frame enables reliable control of rapid transitions between mode-locked states in a 1D RDE model by separating fast detonation propagation from slower operating-mode dynamics.
-
From $P(y|x)$ to $P(y)$: Investigating Reinforcement Learning in Pre-train Space
PreRL applies reward-driven updates to P(y) in pre-train space, uses Negative Sample Reinforcement to prune bad reasoning paths and boost reflection, and combines with standard RL in Dual Space RL to outperform baselines on reasoning tasks.
-
Positive-Only Drifting Policy Optimization
PODPO is a likelihood-free generative policy optimization method for online RL that steers actions to high-return regions using only positive-advantage samples and local contrastive drifting.
-
NaP-Control: Navigating Diffusion Prior for Versatile and Fast Character Control
NaP-Control uses RL to directly predict optimized diffusion noise from a task-agnostic prior, enabling fast inference and higher success rates for versatile whole-body character control while preserving motion quality.
-
AlphaCNOT: Learning CNOT Minimization with Model-Based Planning
AlphaCNOT combines reinforcement learning with Monte Carlo Tree Search planning to reduce CNOT gate counts by up to 32% versus heuristics in quantum circuit synthesis.
-
Chain of Uncertain Rewards with Large Language Models for Reinforcement Learning
CoUR uses LLMs for efficient RL reward design through uncertainty quantification and similarity selection, achieving better performance and lower evaluation costs on IsaacGym and Bidexterous Manipulation benchmarks.
-
Bridging MARL to SARL: An Order-Independent Multi-Agent Transformer via Latent Consensus
CMAT uses a transformer decoder to produce a high-level consensus vector in latent space, enabling simultaneous order-independent actions by all agents and optimization via single-agent PPO, with superior results on StarCraft II, Multi-Agent MuJoCo, and Google Research Football.
-
Learning-Based Sparsification of Dynamic Graphs in Robotic Exploration Algorithms
A PPO-trained transformer policy sparsifies dynamic graphs during RRT frontier exploration, cutting size by up to 96% and yielding the most consistent exploration rates across environments.
-
TOPCELL: Topology Optimization of Standard Cell via LLMs
TOPCELL reformulates standard cell topology optimization as an LLM generative task with GRPO fine-tuning, outperforming base models and matching exhaustive solvers with 85.91x speedup in 2nm/7nm industrial flows.
-
Lightning OPD: Efficient Post-Training for Large Reasoning Models with Offline On-Policy Distillation
Lightning OPD is an offline on-policy distillation method that matches standard OPD performance at 4x efficiency by enforcing teacher consistency between SFT and distillation phases.
-
Calibration-Aware Policy Optimization for Reasoning LLMs
CAPO improves LLM calibration by up to 15% while matching or exceeding GRPO accuracy through logistic AUC loss and noise masking, enabling better abstention and scaling performance.
-
Balanced Aggregation: Understanding and Fixing Aggregation Bias in GRPO
Balanced Aggregation fixes sign-length coupling and length downweighting in GRPO by computing separate token means for positive and negative subsets and combining them with sequence-count weights, yielding more stable training and higher benchmark scores.
-
Hybrid Adaptive Tuning for Tiered Memory Systems
PTMT is a lightweight framework that automates parameter tuning for memory tiering via hybrid offline database building and online customized reinforcement learning, delivering 14-30% gains over defaults and 32% over prior art on four systems.
-
Relax: An Asynchronous Reinforcement Learning Engine for Omni-Modal Post-Training at Scale
Relax is a new RL training engine with omni-native design and async execution that delivers up to 2x speedups over baselines like veRL while converging to equivalent reward levels on Qwen3 models.
-
Eliciting Medical Reasoning with Knowledge-enhanced Data Synthesis: A Semi-Supervised Reinforcement Learning Approach
MedSSR improves LLM medical reasoning on rare diseases by up to 5.93% through knowledge-enhanced question synthesis and semi-supervised RL with self-generated pseudo-labels.
-
CAGenMol: Condition-Aware Diffusion Language Model for Goal-Directed Molecular Generation
CAGenMol uses condition-aware discrete diffusion coupled with reinforcement learning to generate valid molecules meeting multiple heterogeneous constraints, outperforming prior methods on binding affinity, drug-likeness, and success rate benchmarks.
-
The Past Is Not Past: Memory-Enhanced Dynamic Reward Shaping
MEDS improves LLM RL performance by up to 4.13 pass@1 and 4.37 pass@128 points by dynamically penalizing rollouts matching prevalent historical error clusters identified via memory-stored representations and density clustering.
-
ViserDex: Visual Sim-to-Real for Robust Dexterous In-hand Reorientation
A framework using 3D Gaussian Splatting for visual domain randomization enables robust monocular RGB-based dexterous in-hand reorientation on real hardware for multiple objects under varied lighting.
-
HO-Flow: Generalizable Hand-Object Interaction Generation with Latent Flow Matching
HO-Flow synthesizes realistic hand-object motions from text and canonical 3D objects via an interaction-aware VAE and masked flow matching, reporting SOTA physical plausibility and diversity on GRAB, OakInk, and DexYCB.
-
Adaptive Bounded-Rationality Modeling of Early-Stage Takeover in Shared-Control Driving
The adaptive bounded-rationality model anticipates hazardous takeovers with better coverage and lead time than baselines while aligning inferred parameters with eye-tracking metrics.
-
Skill-SD: Skill-Conditioned Self-Distillation for Multi-turn LLM Agents
Skill-SD turns an agent's completed trajectories into dynamic natural-language skills that condition only the teacher in self-distillation, yielding 14-42% gains over RL and OPSD baselines on multi-turn agent benchmarks.
-
Preference-Agile Multi-Objective Optimization for Real-time Vehicle Dispatching
PAMOO uses DRL to enable dynamic preference adjustments in sequential multi-objective optimization and shows better performance than standard MOO methods on container terminal vehicle dispatching.
-
Simple but Stable, Fast and Safe: Achieve End-to-end Control by High-Fidelity Differentiable Simulation
An end-to-end RL policy trained via high-fidelity differentiable simulation maps depth images straight to bodyrate commands, achieving top success rates, low jerk, and zero-shot real-world generalization up to 7.5 m/s in dense environments.
-
PhyMix: Towards Physically Consistent Single-Image 3D Indoor Scene Generation with Implicit--Explicit Optimization
PhyMix unifies a new multi-aspect physics evaluator with implicit policy optimization and explicit test-time correction to produce single-image 3D indoor scenes that are both visually faithful and physically plausible.
-
Self-Distilled Reinforcement Learning for Co-Evolving Agentic Recommender Systems
CoARS enables co-evolving recommender and user agents by using interaction-derived rewards and self-distilled credit assignment to internalize multi-turn feedback into model parameters, outperforming prior agentic baselines.
-
Deep Reinforcement Learning for Cognitive Time-Division Joint SAR and Secure Communications
DRL solves a time-division joint SAR and secure communication problem to maximize worst-case secrecy rate by tracking eavesdroppers with cognitive SAR ATI and adapting beamforming plus jamming, outperforming equal-aperture and random baselines in simulations.
-
Improving Medical VQA through Trajectory-Aware Process Supervision
A trajectory-aware process reward using DTW on sentence embeddings, combined with exact-match in GRPO after SFT, raises mean medical VQA accuracy from 0.598 to 0.689 across six benchmarks.
-
MEMENTO: Teaching LLMs to Manage Their Own Context
MEMENTO trains LLMs to segment reasoning into blocks, generate mementos as dense summaries, and reason forward using only mementos and KV states, cutting peak KV cache by ~2.5x while preserving benchmark accuracy.
-
VL-Calibration: Decoupled Confidence Calibration for Large Vision-Language Models Reasoning
VL-Calibration is a reinforcement learning method that separates visual and reasoning confidence in LVLMs via intrinsic visual certainty estimation to improve calibration and accuracy.
-
Event-Driven Temporal Graph Networks for Asynchronous Multi-Agent Cyber Defense in NetForge_RL
CT-GMARL with fixed-step Neural ODEs in the NetForge_RL simulator delivers 2.0x higher median Blue reward than R-MAPPO, restores 12x more services, and transfers zero-shot to a live Docker environment with median reward 98,026.
-
Region-Constrained Group Relative Policy Optimization for Flow-Based Image Editing
RC-GRPO-Editing constrains GRPO exploration to editing regions via localized noise and attention rewards, improving instruction adherence and non-target preservation in flow-based image editing.
-
On the Role of DAG topology in Energy-Aware Cloud Scheduling : A GNN-Based Deep Reinforcement Learning Approach
GNN-DRL cloud schedulers for DAG workflows degrade under topology shifts because structural mismatches disrupt message passing and policy generalization.
-
Truncated Rectified Flow Policy for Reinforcement Learning with One-Step Sampling
TRFP combines rectified flow models with truncation to support multimodal policies in MaxEnt RL while allowing fast one-step sampling and stable training.
-
Think Less, Know More: State-Aware Reasoning Compression with Knowledge Guidance for Efficient Reasoning
STACK reduces average reasoning response length by 59.9% and raises accuracy by 4.8 points over prior methods on three math benchmarks via state-aware compression, knowledge guidance, and early stopping.
-
TensorHub: Scalable and Elastic Weight Transfer for LLM RL Training
TensorHub uses Reference-Oriented Storage to enable scalable weight transfer in LLM RL training by referencing replicated GPU weights, achieving up to 19x reduction in cross-datacenter stall time.
-
C$^2$T: Captioning-Structure and LLM-Aligned Common-Sense Reward Learning for Traffic--Vehicle Coordination
C2T learns an LLM-derived common-sense reward function to improve cooperative multi-intersection traffic control policies, outperforming standard MARL baselines on efficiency, safety, and energy proxies while allowing prompt-based policy tuning.
-
HTNav: A Hybrid Navigation Framework with Tiered Structure for Urban Aerial Vision-and-Language Navigation
HTNav combines imitation and reinforcement learning in a staged, tiered structure with map learning to reach state-of-the-art performance on the CityNav benchmark for urban aerial navigation.
-
SPPO: Sequence-Level PPO for Long-Horizon Reasoning Tasks
SPPO enables stable, sample-efficient alignment of LLMs on long-horizon reasoning tasks by using a decoupled scalar value function for low-variance advantages without multi-sampling.
-
Toward Hardware-Agnostic Quadrupedal World Models via Morphology Conditioning
Morphology-conditioned quadrupedal world model enables zero-shot generalization to new robot embodiments for locomotion tasks.
-
Sumo: Dynamic and Generalizable Whole-Body Loco-Manipulation
Test-time steering of pre-trained whole-body policies via sample-based planning lets legged robots generalize dynamic loco-manipulation to varied heavy objects and tasks without additional training or tuning.