A verifiable empirical win rate reward combined with gradient masking enables RL training of a 7B model to reach betting-market calibration on NFL win probabilities using only outcome data.
hub Mixed citations
arXiv preprint arXiv:2504.02546 , year=
Mixed citation behavior. Most common role is background (67%).
22
Pith papers citing it
Background
67% of classified citations
hub tools
citation-role summary
background 4
baseline 1
method 1
citation-polarity summary
representative citing papers
TRL extends tandem training to RLVR pipelines, matching GRPO solo reasoning on Qwen3-4B math tasks while improving handoff robustness, reducing distributional drift, and increasing CoT legibility for the junior.
GiGPO adds a hierarchical grouping mechanism to group-based RL so that LLM agents receive both global trajectory and local step-level credit signals, yielding >12% gains on ALFWorld and >9% on WebShop over GRPO while keeping the same rollout and memory footprint.
Theoretical analysis of RLVR update dynamics leads to ACPO, an adaptive clipping method that outperforms DAPO and CISPO on reasoning benchmarks with 3B and 7B models.
SGCD reshapes GRPO token advantages via detached sibling-contrast credit from an external LLM, improving AppWorld and airline tool-use scores while keeping policy gradient as the actor update.
Reinforcement learning after SFT conversion narrows the performance gap between sliding-window attention and full self-attention on math reasoning benchmarks while preserving linear complexity.
HölderPO unifies token-level aggregation in GRPO via the Hölder mean with a tunable p parameter and annealing schedule, delivering 54.9% average accuracy on math benchmarks and 93.8% success on ALFWorld.
FG-ExPO improves GRPO by adaptively scaling the KL penalty with batch accuracy and sampling questions via a Gaussian centered at 0.5 accuracy, delivering up to 13.34 point gains on AIME 2025 pass@32.
Kernel smoothing enables accurate low-variance value and gradient estimates for policy optimization in LLM reasoning under tight sampling constraints per prompt.
DeepSearch embeds MCTS into RLVR training with global frontier selection, entropy guidance, and adaptive replay to achieve 62.95% average accuracy on math reasoning benchmarks while using 5.7x fewer GPU hours than extended training.
Archer introduces response-level entropy normalization and differentiated clipping/KL regularization in RLVR to encourage exploration on reasoning tokens while stabilizing knowledge tokens, yielding gains in pass@1 and pass@K on reasoning benchmarks.
Survey mapping RL techniques onto LLM training and highlighting gaps in value-based, off-policy, and bootstrapping methods.
APMPO boosts average Pass@1 scores on math reasoning benchmarks by 3 points over GRPO by using an adaptive power-mean policy objective and feedback-driven clipping bounds in RLVR training.
FREIA applies free energy principles and adaptive advantage shaping to unsupervised RL, outperforming baselines by 0.5-3.5 Pass@1 points on math reasoning with a 1.5B model.
OpenVLThinkerV2 applies a new Gaussian GRPO training objective with response and entropy shaping to outperform prior open-source and proprietary models on 18 visual reasoning benchmarks.
OneThinker unifies image and video reasoning in one model across 10 tasks via a 600k corpus, CoT-annotated SFT, and EMA-GRPO reinforcement learning, reporting strong results on 31 benchmarks plus some cross-task transfer.
Derives a token-level entropy change approximation revealing four factors, identifies limitations in prior entropy interventions, and proposes STEER which adaptively reweights tokens to mitigate collapse and improve performance on math and coding benchmarks.
AutoDrive-R² adds four-step CoT reasoning with self-reflection to VLA models via SFT on nuScenesR²-6K and GRPO RL under spatial, dynamic, and smoothness rewards, reporting SOTA results on nuScenes and Waymo.
Mujica-MyGo decomposes multi-turn RAG interactions via multi-agent workflows and applies minimalist policy gradient optimization to improve performance on QA benchmarks while avoiding long-context problems.
ASymPO normalizes token losses by average current-policy negative log-probability to restore zero-sum balance in asynchronous LLM RL without behavior information.
EXPO improves GRPO for LLM mathematical reasoning via accuracy-conditioned KL scaling and Gaussian curriculum sampling, delivering gains such as 13.34 points on AIME 2025 pass@32.
A statistical survey of RLHF for LLM alignment that connects preference learning and policy optimization to models like Bradley-Terry-Luce while reviewing methods, extensions, and open challenges.
citing papers explorer
-
Verifiable Rewards for Calibrated Probabilistic Forecasting
A verifiable empirical win rate reward combined with gradient masking enables RL training of a 7B model to reach betting-market calibration on NFL win probabilities using only outcome data.
-
Tandem Reinforcement Learning with Verifiable Rewards
TRL extends tandem training to RLVR pipelines, matching GRPO solo reasoning on Qwen3-4B math tasks while improving handoff robustness, reducing distributional drift, and increasing CoT legibility for the junior.
-
Group-in-Group Policy Optimization for LLM Agent Training
GiGPO adds a hierarchical grouping mechanism to group-based RL so that LLM agents receive both global trajectory and local step-level credit signals, yielding >12% gains on ALFWorld and >9% on WebShop over GRPO while keeping the same rollout and memory footprint.
-
What are Key Factors for Updates in RL for LLM Reasoning?
Theoretical analysis of RLVR update dynamics leads to ACPO, an adaptive clipping method that outperforms DAPO and CISPO on reasoning benchmarks with 3B and 7B models.
-
Keep Policy Gradient in Charge: Sibling-Guided Credit Distillation for Long-Horizon Tool-Use Agents
SGCD reshapes GRPO token advantages via detached sibling-contrast credit from an external LLM, improving AppWorld and airline tool-use scores while keeping policy gradient as the actor update.
-
Architecture-Aware Reinforcement Learning Makes Sliding-Window Attention Competitive in Math Reasoning
Reinforcement learning after SFT conversion narrows the performance gap between sliding-window attention and full self-attention on math reasoning benchmarks while preserving linear complexity.
-
Holder Policy Optimisation
HölderPO unifies token-level aggregation in GRPO via the Hölder mean with a tunable p parameter and annealing schedule, delivering 54.9% average accuracy on math benchmarks and 93.8% success on ALFWorld.
-
fg-expo: Frontier-guided exploration-prioritized policy optimization via adaptive kl and gaussian curriculum
FG-ExPO improves GRPO by adaptively scaling the KL penalty with batch accuracy and sampling questions via a Gaussian centered at 0.5 accuracy, delivering up to 13.34 point gains on AIME 2025 pass@32.
-
Kernelized Advantage Estimation: From Nonparametric Statistics to LLM Reasoning
Kernel smoothing enables accurate low-variance value and gradient estimates for policy optimization in LLM reasoning under tight sampling constraints per prompt.
-
DeepSearch: Overcome the Bottleneck of Reinforcement Learning with Verifiable Rewards via Monte Carlo Tree Search
DeepSearch embeds MCTS into RLVR training with global frontier selection, entropy guidance, and adaptive replay to achieve 62.95% average accuracy on math reasoning benchmarks while using 5.7x fewer GPU hours than extended training.
-
Stabilizing Knowledge, Promoting Reasoning: Dual-Token Constraints for RLVR
Archer introduces response-level entropy normalization and differentiated clipping/KL regularization in RLVR to encourage exploration on reasoning tokens while stabilizing knowledge tokens, yielding gains in pass@1 and pass@K on reasoning benchmarks.
-
Modularized Reinforcement Learning on LLMs: From MDP Creation to Exploration and Learning
Survey mapping RL techniques onto LLM training and highlighting gaps in value-based, off-policy, and bootstrapping methods.
-
Adapt to Thrive! Adaptive Power-Mean Policy Optimization for Improved LLM Reasoning
APMPO boosts average Pass@1 scores on math reasoning benchmarks by 3 points over GRPO by using an adaptive power-mean policy objective and feedback-driven clipping bounds in RLVR training.
-
Free Energy-Driven Reinforcement Learning with Adaptive Advantage Shaping for Unsupervised Reasoning in LLMs
FREIA applies free energy principles and adaptive advantage shaping to unsupervised RL, outperforming baselines by 0.5-3.5 Pass@1 points on math reasoning with a 1.5B model.
-
OpenVLThinkerV2: A Generalist Multimodal Reasoning Model for Multi-domain Visual Tasks
OpenVLThinkerV2 applies a new Gaussian GRPO training objective with response and entropy shaping to outperform prior open-source and proprietary models on 18 visual reasoning benchmarks.
-
OneThinker: All-in-one Reasoning Model for Image and Video
OneThinker unifies image and video reasoning in one model across 10 tasks via a 600k corpus, CoT-annotated SFT, and EMA-GRPO reinforcement learning, reporting strong results on 31 benchmarks plus some cross-task transfer.
-
Rethinking Entropy Interventions in RLVR: An Entropy Change Perspective
Derives a token-level entropy change approximation revealing four factors, identifies limitations in prior entropy interventions, and proposes STEER which adaptively reweights tokens to mitigate collapse and improve performance on math and coding benchmarks.
-
AutoDrive-R$^2$: Incentivizing Reasoning and Self-Reflection Capacity for VLA Model in Autonomous Driving
AutoDrive-R² adds four-step CoT reasoning with self-reflection to VLA models via SFT on nuScenesR²-6K and GRPO RL under spatial, dynamic, and smoothness rewards, reporting SOTA results on nuScenes and Waymo.
-
Advancing Multi-Agent RAG Systems with Minimalist Reinforcement Learning
Mujica-MyGo decomposes multi-turn RAG interactions via multi-agent workflows and applies minimalist policy gradient optimization to improve performance on QA benchmarks while avoiding long-context problems.
-
ASymPO: Asymmetric-Scale Policy Optimization for Asynchronous LLM Post-Training Without Behavior Information
ASymPO normalizes token losses by average current-policy negative log-probability to restore zero-sum balance in asynchronous LLM RL without behavior information.
-
expo: Exploration-prioritized policy optimization via adaptive kl regulation and gaussian curriculum sampling
EXPO improves GRPO for LLM mathematical reasoning via accuracy-conditioned KL scaling and Gaussian curriculum sampling, delivering gains such as 13.34 points on AIME 2025 pass@32.
-
Reinforcement Learning from Human Feedback: A Statistical Perspective
A statistical survey of RLHF for LLM alignment that connects preference learning and policy optimization to models like Bradley-Terry-Luce while reviewing methods, extensions, and open challenges.