CapRL++ applies reinforcement learning with verifiable rewards to dense image and video captioning by scoring captions via the accuracy of a vision-free LLM answering MCQs from the caption alone.
hub
Crossing the reward bridge: Expanding RL with verifiable rewards across diverse domains
24 Pith papers cite this work. Polarity classification is still indexing.
24
Pith papers citing it
hub tools
citation-role summary
background 4
citation-polarity summary
roles
background 4polarities
background 4representative citing papers
This survey introduces the Generate-Filter-Control-Replay (GFCR) taxonomy to structure rollout pipelines for RL-based post-training of reasoning LLMs.
A cross-evaluation setup with native-speaker SME rubrics benchmarks LLMs on underrepresented Arabic dialects and quantifies automated judge bias and cultural-reasoning gaps.
DR-MV3D decomposes MV3D-VQA into global map construction, question-conditioned view planning, and egocentric grounding, supervised by global consistency and local trajectory rewards optimized via GRPO.
ADR generates novel verifiable code tasks via atomic decomposition and recombination, outperforming heuristic baselines in originality, difficulty, and downstream RLVR gains across coding domains.
ReBel uses belief-consistency supervision and belief-aware grouping to improve credit assignment in long-horizon RL for LLM agents, achieving up to 20.4 percentage points higher success and 2.1x better sample efficiency than GRPO on ALFWorld and WebShop.
V-tableR1 uses a critic VLM for dense step-level feedback and a new PGPO algorithm to shift multimodal table reasoning from pattern matching to verifiable logical steps, achieving SOTA accuracy with a 4B open-source model.
Introduces MemHome benchmark and RL with multi-dimensional rewards for memory-driven smart home device control.
SpeciaRL applies a dynamic verifier-based reward in reinforcement learning to steer reasoning LMMs toward correct and specific predictions on fine-grained open-world image classification tasks.
MARL-Rad trains region-specific and global agents with reinforcement learning on clinical rewards to produce more accurate radiology reports than prior methods on MIMIC-CXR and IU X-ray datasets.
CURE-MED pairs a new 13-language medical reasoning benchmark with curriculum RL to raise logical correctness to 70% and language consistency to 95% at 32B scale while outperforming baselines.
UCAS refines RLVR advantage signals with a logit-space self-confidence proxy for response-level modulation and asymmetric token-level penalties based on raw logit certainty to boost exploration and reduce entropy collapse.
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.
RaR uses aggregated rubric feedback as rewards in on-policy RL, delivering up to 31% relative gains on HealthBench and 7% on GPQA-Diamond versus direct Likert LLM-as-judge baselines.
PointVG-R is a new MLLM that reaches SOTA on pointing localization by 15.86 mIoU points via a geometric reasoning pipeline, EgoPoint-CoT dataset, SFT, RL, and variance-based reward weighting.
GeoMin uses geometric distribution modeling on labeled data to assess self-reward reliability, enabling better performance in semi-supervised RLVR with only 10% of typical annotations.
CARE-RL combines PA-GRM for task-adaptive rewards on open-ended tasks and DACSP for modulating RL updates using historical capability directions, reporting higher total average scores than baselines on Qwen models.
CAST adds non-privileged self-teacher scoring and bidirectional advantage flipping to GRPO so that zero-variance groups still produce verifier-signed token gradients.
M2A uses null-space model merging to combine mathematical and agentic reasoning in LLMs, raising SWE-Bench Verified performance from 44.0% to 51.2% on Qwen3-8B without retraining.
DVPO learns token-level value distributions and uses asymmetric risk regularization to contract lower tails while expanding upper tails, outperforming PPO and GRPO under noisy supervision in dialogue, math, and QA tasks.
Direct Reasoning Optimization applies token-level Reasoning Reflection Reward (R3) focused on high-variance tokens and rubric-gating constraints to improve sample-efficient RL training of LLMs on unverifiable tasks.
BV-Blend blends prompt-local and semantic-cluster historical reward statistics via SEM-derived weights to stabilize critic-free RL advantage estimation.
SPG-Layout combines statistical object priors with hierarchical large-object-first placement to produce physically plausible text-driven 3D scenes in non-Manhattan rooms and outperforms baselines on a new 500-scene benchmark.
citing papers explorer
-
CapRL++: Unified Reinforcement Learning with Verifiable Rewards for Dense Image and Video Captioning
CapRL++ applies reinforcement learning with verifiable rewards to dense image and video captioning by scoring captions via the accuracy of a vision-free LLM answering MCQs from the caption alone.
-
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.
-
Benchmarking Frontier LLMs on Arabic Cultural and Sociolinguistic Knowledge: A Cross-Evaluation Framework with Human SME Ground Truth
A cross-evaluation setup with native-speaker SME rubrics benchmarks LLMs on underrepresented Arabic dialects and quantifies automated judge bias and cultural-reasoning gaps.
-
Dense Reward for Multi-View 3D Reasoning with Global Maps and Local Views
DR-MV3D decomposes MV3D-VQA into global map construction, question-conditioned view planning, and egocentric grounding, supervised by global consistency and local trajectory rewards optimized via GRPO.
-
Combinatorial Synthesis: Scaling Code RLVR via Atomic Decomposition and Recombination
ADR generates novel verifiable code tasks via atomic decomposition and recombination, outperforming heuristic baselines in originality, difficulty, and downstream RLVR gains across coding domains.
-
Rewarding Beliefs, Not Actions: Consistency-Guided Credit Assignment for Long-Horizon Agents
ReBel uses belief-consistency supervision and belief-aware grouping to improve credit assignment in long-horizon RL for LLM agents, achieving up to 20.4 percentage points higher success and 2.1x better sample efficiency than GRPO on ALFWorld and WebShop.
-
V-tableR1: Process-Supervised Multimodal Table Reasoning with Critic-Guided Policy Optimization
V-tableR1 uses a critic VLM for dense step-level feedback and a new PGPO algorithm to shift multimodal table reasoning from pattern matching to verifiable logical steps, achieving SOTA accuracy with a 4B open-source model.
-
Trust Your Memory: Verifiable Control of Smart Homes through Reinforcement Learning with Multi-dimensional Rewards
Introduces MemHome benchmark and RL with multi-dimensional rewards for memory-driven smart home device control.
-
Specificity-aware reinforcement learning for fine-grained open-world classification
SpeciaRL applies a dynamic verifier-based reward in reinforcement learning to steer reasoning LMMs toward correct and specific predictions on fine-grained open-world image classification tasks.
-
Multi-Modal Multi-Agent Reinforcement Learning for Radiology Report Generation
MARL-Rad trains region-specific and global agents with reinforcement learning on clinical rewards to produce more accurate radiology reports than prior methods on MIMIC-CXR and IU X-ray datasets.
-
CURE-Med: Curriculum-Informed Reinforcement Learning for Multilingual Medical Reasoning
CURE-MED pairs a new 13-language medical reasoning benchmark with curriculum RL to raise logical correctness to 70% and language consistency to 95% at 32B scale while outperforming baselines.
-
Unlocking Exploration in RLVR: Uncertainty-aware Advantage Shaping for Deeper Reasoning
UCAS refines RLVR advantage signals with a logit-space self-confidence proxy for response-level modulation and asymmetric token-level penalties based on raw logit certainty to boost exploration and reduce entropy collapse.
-
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.
-
Rubrics as Rewards: Reinforcement Learning Beyond Verifiable Domains
RaR uses aggregated rubric feedback as rewards in on-policy RL, delivering up to 31% relative gains on HealthBench and 7% on GPQA-Diamond versus direct Likert LLM-as-judge baselines.
-
PointVG-R: Internalizing Geometric Reasoning in MLLMs for Precise Pointing Localization via Visual Chain of Thought
PointVG-R is a new MLLM that reaches SOTA on pointing localization by 15.86 mIoU points via a geometric reasoning pipeline, EgoPoint-CoT dataset, SFT, RL, and variance-based reward weighting.
-
GeoMin: Data-Efficient Semi-Supervised RLVR via Geometric Distribution Modeling
GeoMin uses geometric distribution modeling on labeled data to assess self-reward reliability, enabling better performance in semi-supervised RLVR with only 10% of typical annotations.
-
CARE-RL: Capability-Aware Reinforcement Learning for Mitigating Cross-Domain Conflicts
CARE-RL combines PA-GRM for task-adaptive rewards on open-ended tasks and DACSP for modulating RL updates using historical capability directions, reporting higher total average scores than baselines on Qwen models.
-
CAST: Non-Privileged Clipped Asymmetric Self-Teaching with Advantage Flipping for GRPO
CAST adds non-privileged self-teacher scoring and bidirectional advantage flipping to GRPO so that zero-variance groups still produce verifier-signed token gradients.
-
M2A: Synergizing Mathematical and Agentic Reasoning in Large Language Models
M2A uses null-space model merging to combine mathematical and agentic reasoning in LLMs, raising SWE-Bench Verified performance from 44.0% to 51.2% on Qwen3-8B without retraining.
-
DVPO: Distributional Value Modeling-based Policy Optimization for LLM Post-Training
DVPO learns token-level value distributions and uses asymmetric risk regularization to contract lower tails while expanding upper tails, outperforming PPO and GRPO under noisy supervision in dialogue, math, and QA tasks.
-
Direct Reasoning Optimization: Token-Level Reasoning Reflectivity Meets Rubric Gates for Unverifiable Tasks
Direct Reasoning Optimization applies token-level Reasoning Reflection Reward (R3) focused on high-variance tokens and rubric-gating constraints to improve sample-efficient RL training of LLMs on unverifiable tasks.
-
BV-Blend: Uncertainty-Weighted Historical Baselines for Stable Critic-Free RL with Verifiable Rewards
BV-Blend blends prompt-local and semantic-cluster historical reward statistics via SEM-derived weights to stabilize critic-free RL advantage estimation.
-
Text-Driven 3D Indoor Scene Synthesis in Non-Manhattan Environments
SPG-Layout combines statistical object priors with hierarchical large-object-first placement to produce physically plausible text-driven 3D scenes in non-Manhattan rooms and outperforms baselines on a new 500-scene benchmark.
- Themis: Training Robust Multilingual Code Reward Models for Flexible Multi-Criteria Scoring