ChemCoTBench-V2 is a new rule-verifiable benchmark with 5,620 samples across 18 tasks that evaluates LLM chemical reasoning traces using deterministic chemistry rules and reference traces rather than final answers alone.
hub
Deepseekmath-v2: Towards self-verifiable mathematical reasoning
21 Pith papers cite this work. Polarity classification is still indexing.
21
Pith papers citing it
hub tools
citation-role summary
background 2
citation-polarity summary
roles
background 2polarities
background 2representative citing papers
Pseudo-Formalization decomposes proofs into self-contained natural language modules for independent LLM-based Block Verification, outperforming LLM-as-judge baselines on olympiad and research math benchmarks while releasing ArxivMathGradingBench.
DataPRM is an environment-aware generative process reward model that improves LLM data analysis agents by 7-11% on benchmarks via active verification and reflection-aware ternary rewards.
SageMath-augmented ReAct agents raise solve rates by +9.7 pp on average on a curated 133-problem RealMath subset, with GPT-5.5 reaching 75.2%.
Introduces RSI metric and RSI-S filtering method for adaptive token selection in RLVR, reporting 2-3 point gains over GRPO on AIME/AMC benchmarks.
Unsupervised rewards combining model uncertainty and semantic consistency allow protein language models to self-steer via SRO and BRO algorithms, outperforming DPO and KTO on out-of-distribution prompts while approaching oracle performance.
Reasoning Arena converts non-diverse reward groups in RLVR into relative rewards via adaptive trace tournaments and Bradley-Terry fitting on anchor comparisons, claiming 7.6% average gains and 27-41% faster training on math/coding benchmarks.
ReviewGuard aligns LLM peer reviews with future citations via impact-aligned RL, achieving Spearman ρ=0.776 on rejected-then-published AI/ML papers versus 0.492 for human reviewers and flagging 5.6× more high-impact cases.
Self-trained verification trains verifiers to imitate informed versions of themselves using reference solutions, improving test-time V-R loops and training-time self-improvement with reported gains of 2x on hard math and 14x on scientific reasoning.
DORA's multi-version streaming rollout enables 2-3x higher throughput in asynchronous RL for LLMs while preserving convergence by maintaining policy consistency, data integrity, and bounded staleness.
MathNet compiles 30,676 multilingual Olympiad problems with solutions into a benchmark showing top LLMs score 69–78% while embedding retrievers rarely find mathematically equivalent problems at rank 1.
Introduces MemHome benchmark and RL with multi-dimensional rewards for memory-driven smart home device control.
Frontier AI models score below 10% on Riemann-Bench, a private expert-authored benchmark of research-level mathematics with closed-form, programmatically verified solutions.
ImpRIF improves LLM complex instruction following by synthesizing data from reasoning graphs and training models to reason explicitly along those graphs.
Survey mapping RL techniques onto LLM training and highlighting gaps in value-based, off-policy, and bootstrapping methods.
A new pipeline uses interpretability to characterize concepts in preference data and shape rewards via feature or data interventions during LM post-training.
STAR-PólyaMath introduces a multi-agent framework with meta-strategic supervision and state-machine orchestration that reports state-of-the-art and perfect scores on eight top math competition benchmarks.
A closed-loop system couples LLM-based 3D scene generation with RL optimization and VR user interactions to produce adaptive, immersive environments, claiming SOTA results on the ALFRED benchmark.
Lack of exploration from conditioning on prior answers is the primary reason parallel sampling outperforms sequential sampling in large reasoning models.
A six-dimensional MathVerifier supplies hard negatives and per-sample weights that improve DPO performance on math reasoning for a 1.5B Qwen2.5 model over standard SFT and unweighted DPO.
OmniVerifier-M1 is a generalist visual verifier using symbolic outputs for meta-verification and decoupled RL to outperform joint optimization for robust verification and agentic self-correction.
citing papers explorer
-
From Answers to States: Verifiable Process-Level Evaluation of Chemical Reasoning in Large Language Models
ChemCoTBench-V2 is a new rule-verifiable benchmark with 5,620 samples across 18 tasks that evaluates LLM chemical reasoning traces using deterministic chemistry rules and reference traces rather than final answers alone.
-
Pseudo-Formalization for Automatic Proof Verification
Pseudo-Formalization decomposes proofs into self-contained natural language modules for independent LLM-based Block Verification, outperforming LLM-as-judge baselines on olympiad and research math benchmarks while releasing ArxivMathGradingBench.
-
Rewarding the Scientific Process: Process-Level Reward Modeling for Agentic Data Analysis
DataPRM is an environment-aware generative process reward model that improves LLM data analysis agents by 7-11% on benchmarks via active verification and reflection-aware ternary rewards.
-
Evaluating SageMath-Augmented LLM Agents for Computational and Experimental Mathematics
SageMath-augmented ReAct agents raise solve rates by +9.7 pp on average on a curated 133-problem RealMath subset, with GPT-5.5 reaching 75.2%.
-
Which Tokens Matter? Adaptive Token Selection for RLVR with the Relative Surprisal Index
Introduces RSI metric and RSI-S filtering method for adaptive token selection in RLVR, reporting 2-3 point gains over GRPO on AIME/AMC benchmarks.
-
Be Your Own Teacher: Steering Protein Language Models via Unsupervised Reward Optimization
Unsupervised rewards combining model uncertainty and semantic consistency allow protein language models to self-steer via SRO and BRO algorithms, outperforming DPO and KTO on out-of-distribution prompts while approaching oracle performance.
-
Reasoning Arena: Trace Tournaments When Verifiable Rewards Fall Short
Reasoning Arena converts non-diverse reward groups in RLVR into relative rewards via adaptive trace tournaments and Bradley-Terry fitting on anchor comparisons, claiming 7.6% average gains and 27-41% faster training on math/coding benchmarks.
-
ReviewGuard: Aligning LLM-Assisted Peer Review with Long-Term Scientific Impact
ReviewGuard aligns LLM peer reviews with future citations via impact-aligned RL, achieving Spearman ρ=0.776 on rejected-then-published AI/ML papers versus 0.492 for human reviewers and flagging 5.6× more high-impact cases.
-
Self-Trained Verification for Training- and Test-Time Self-Improvement
Self-trained verification trains verifiers to imitate informed versions of themselves using reference solutions, improving test-time V-R loops and training-time self-improvement with reported gains of 2x on hard math and 14x on scientific reasoning.
-
DORA: A Scalable Asynchronous Reinforcement Learning System for Language Model Training
DORA's multi-version streaming rollout enables 2-3x higher throughput in asynchronous RL for LLMs while preserving convergence by maintaining policy consistency, data integrity, and bounded staleness.
-
MathNet: a Global Multimodal Benchmark for Mathematical Reasoning and Retrieval
MathNet compiles 30,676 multilingual Olympiad problems with solutions into a benchmark showing top LLMs score 69–78% while embedding retrievers rarely find mathematically equivalent problems at rank 1.
-
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.
-
Riemann-Bench: A Benchmark for Moonshot Mathematics
Frontier AI models score below 10% on Riemann-Bench, a private expert-authored benchmark of research-level mathematics with closed-form, programmatically verified solutions.
-
ImpRIF: Stronger Implicit Reasoning Leads to Better Complex Instruction Following
ImpRIF improves LLM complex instruction following by synthesizing data from reasoning graphs and training models to reason explicitly along those graphs.
-
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.
-
Anatomy of Post-Training: Using Interpretability to Characterize Data and Shape the Learning Signal
A new pipeline uses interpretability to characterize concepts in preference data and shape rewards via feature or data interventions during LM post-training.
-
STAR-P\'olyaMath: Multi-Agent Reasoning under Persistent Meta-Strategic Supervision
STAR-PólyaMath introduces a multi-agent framework with meta-strategic supervision and state-machine orchestration that reports state-of-the-art and perfect scores on eight top math competition benchmarks.
-
Closing the Loop: Unified 3D Scene Generation and Immersive Interaction via LLM-RL Coupling
A closed-loop system couples LLM-based 3D scene generation with RL optimization and VR user interactions to produce adaptive, immersive environments, claiming SOTA results on the ALFRED benchmark.
-
Understanding Performance Gap Between Parallel and Sequential Sampling in Large Reasoning Models
Lack of exploration from conditioning on prior answers is the primary reason parallel sampling outperforms sequential sampling in large reasoning models.
-
Hard Negative Sample-Augmented DPO Post-Training for Small Language Models
A six-dimensional MathVerifier supplies hard negatives and per-sample weights that improve DPO performance on math reasoning for a 1.5B Qwen2.5 model over standard SFT and unweighted DPO.
-
OmniVerifier-M1: Multimodal Meta-Verifier with Explicit Structured Recalibration
OmniVerifier-M1 is a generalist visual verifier using symbolic outputs for meta-verification and decoupled RL to outperform joint optimization for robust verification and agentic self-correction.