DeepMath-103K is a new 103K-problem mathematical dataset with high difficulty, rigorous decontamination, and verifiable answers to support RL training of language-model reasoning.
hub
Big-math: A large-scale, high-quality math dataset for reinforcement learning in language models
21 Pith papers cite this work. Polarity classification is still indexing.
21
Pith papers citing it
hub tools
citation-role summary
background 2
dataset 2
citation-polarity summary
representative citing papers
GRPO, Dr. GRPO, and DAPO are three settings of one dial on the group standard deviation of binary rewards, unified by the group-standard-deviation identity where disagreement equals update magnitude.
ATLAS traces RLVR data to 20 atomic sources, most datasets are variants, and DAPO++ curated with SCA improves RLVR performance while Q predicts training effectiveness.
LLM tutors leak answers under adversarial student attacks, but a fine-tuned jailbreak agent and simple defenses can benchmark and improve robustness.
ProofGrid is a new benchmark for LLM reasoning that uses machine-checkable proofs in minimal formal notation, revealing progress on basic tasks but major gaps in complex combinatorial and synthesis reasoning.
EngiBench shows LLMs accuracy drops with task complexity, degrades under perturbations, and stays below human performance on open-ended engineering problems.
CALIBER elicits and supervises pre-reasoning confidence with prompt-level success probability and post-reasoning confidence with answer-level correctness, cutting ECE by 52.5% on BigMathDigits for a 7B model while remaining competitive on accuracy.
EvoTrainer co-evolves LLM policies and training harnesses via empirical feedback to match or exceed human-engineered RL on math reasoning, code generation, and long-horizon software engineering.
The Tutoring Effectiveness Index (TEI) uses four signals from LLM conversations to select math tutoring responses, raising student improvement rates from 59.0% to 81.9% at N=8 on a frozen DeepSeek-R1-8B model without training or judges.
REFT improves Pass@1/8/64 in RLVR by uniform first-token sampling from top-N candidates across 0.5B-7B models and multiple difficulty levels.
Training-free prompt optimization methods, including five new education-focused ones, surpass the strongest RL-trained baseline across five conditions on two OOD suites while showing distinct teaching behavior patterns.
MONA integrates Nesterov acceleration into Muon's orthogonalization framework, reporting better convergence than Muon and AdamW on MoE models up to 68B parameters trained on 1T tokens and SOTA fine-tuning results.
TokenHD uses a scalable data synthesis engine and importance-weighted training to create token-level hallucination detectors that work on free-form text and scale from 0.6B to 8B parameters, outperforming larger reasoning models.
OPT-BENCH trains LLMs on NP-hard optimization via quality-aware RLVR, achieving 93.1% success rate and 46.6% quality ratio on Qwen2.5-7B while outperforming GPT-4o and transferring gains to other domains.
Voltage pulses create nonvolatile, multi-state metastable CDW phases in bulk EuTe4 at room temperature via out-of-plane phase switching in its moiré superstructure.
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.
GRIFT detects reward hacking in verifiable reasoning by using compressed gradients of CoT traces, outperforming text-based baselines by over 25% and improving rejection fine-tuning performance.
SFT on divergent branch-heavy CoT from DeepSeek-R1 yields worse generalization than convergent CoT from gpt-oss despite lower loss, but filtering frequent branches improves average performance by 3.6% on five reasoning benchmarks.
SPHINX generates synthetic visual puzzles for benchmarking LVLMs, where GPT-5 scores 51.1% and RLVR training improves both in-domain and external visual reasoning performance.
FEST improves RLVR sample efficiency on math and coding benchmarks by combining supervised signals, on-policy signals, and decaying weights on just 128 randomly chosen demonstrations, matching full-dataset baselines.
A survey compiling RL methods, challenges, data resources, and applications for enhancing reasoning in large language models and large reasoning models since DeepSeek-R1.
citing papers explorer
-
DeepMath-103K: A Large-Scale, Challenging, Decontaminated, and Verifiable Mathematical Dataset for Advancing Reasoning
DeepMath-103K is a new 103K-problem mathematical dataset with high difficulty, rigorous decontamination, and verifiable answers to support RL training of language-model reasoning.
-
GRPO, Dr. GRPO, and DAPO Are Three Operations on One Number: The Group-Standard-Deviation Identity
GRPO, Dr. GRPO, and DAPO are three settings of one dial on the group standard deviation of binary rewards, unified by the group-standard-deviation identity where disagreement equals update magnitude.
-
RLVR Datasets and Where to Find Them: Tracing Data Lineage for Better Training Data
ATLAS traces RLVR data to 20 atomic sources, most datasets are variants, and DAPO++ curated with SCA improves RLVR performance while Q predicts training effectiveness.
-
Evaluating Answer Leakage Robustness of LLM Tutors against Adversarial Student Attacks
LLM tutors leak answers under adversarial student attacks, but a fine-tuned jailbreak agent and simple defenses can benchmark and improve robustness.
-
Stress-Testing the Reasoning Competence of LLMs With Proofs Under Minimal Formalism
ProofGrid is a new benchmark for LLM reasoning that uses machine-checkable proofs in minimal formal notation, revealing progress on basic tasks but major gaps in complex combinatorial and synthesis reasoning.
-
EngiBench: A Benchmark for Evaluating Large Language Models on Engineering Problem Solving
EngiBench shows LLMs accuracy drops with task complexity, degrades under perturbations, and stays below human performance on open-ended engineering problems.
-
CALIBER: Calibrating Confidence Before and After Reasoning in Language Models
CALIBER elicits and supervises pre-reasoning confidence with prompt-level success probability and post-reasoning confidence with answer-level correctness, cutting ECE by 52.5% on BigMathDigits for a 7B model while remaining competitive on accuracy.
-
EvoTrainer: Co-Evolving LLM Policies and Training Harnesses for Autonomous Agentic Reinforcement Learning
EvoTrainer co-evolves LLM policies and training harnesses via empirical feedback to match or exceed human-engineered RL on math reasoning, code generation, and long-horizon software engineering.
-
The Tutoring Effectiveness Index: Predicting LLM Math Tutor Quality from Four Conversation Signals
The Tutoring Effectiveness Index (TEI) uses four signals from LLM conversations to select math tutoring responses, raising student improvement rates from 59.0% to 81.9% at N=8 on a frozen DeepSeek-R1-8B model without training or judges.
-
Where Rollouts Begin: Low-Load, High-Leverage First-Token Diversification for RLVR
REFT improves Pass@1/8/64 in RLVR by uniform first-token sampling from top-N candidates across 0.5B-7B models and multiple difficulty levels.
-
LLMs Are Already Good Tutors: Training-Free Prompt Optimization for Pedagogical Math Tutoring
Training-free prompt optimization methods, including five new education-focused ones, surpass the strongest RL-trained baseline across five conditions on two OOD suites while showing distinct teaching behavior patterns.
-
MONA: Muon Optimizer with Nesterov Acceleration for Scalable Language Model Training
MONA integrates Nesterov acceleration into Muon's orthogonalization framework, reporting better convergence than Muon and AdamW on MoE models up to 68B parameters trained on 1T tokens and SOTA fine-tuning results.
-
Scalable Token-Level Hallucination Detection in Large Language Models
TokenHD uses a scalable data synthesis engine and importance-weighted training to create token-level hallucination detectors that work on free-form text and scale from 0.6B to 8B parameters, outperforming larger reasoning models.
-
Forge: Quality-Aware Reinforcement Learning for NP-Hard Optimization in LLMs
OPT-BENCH trains LLMs on NP-hard optimization via quality-aware RLVR, achieving 93.1% success rate and 46.6% quality ratio on Qwen2.5-7B while outperforming GPT-4o and transferring gains to other domains.
-
$R^2$-dLLM: Accelerating Diffusion Large Language Models via Spatio-Temporal Redundancy Reduction
Voltage pulses create nonvolatile, multi-state metastable CDW phases in bulk EuTe4 at room temperature via out-of-plane phase switching in its moiré superstructure.
-
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.
-
Detecting and Suppressing Reward Hacking with Gradient Fingerprints
GRIFT detects reward hacking in verifiable reasoning by using compressed gradients of CoT traces, outperforming text-based baselines by over 25% and improving rejection fine-tuning performance.
-
On the Role of Reasoning Patterns in the Generalization Discrepancy of Long Chain-of-Thought Supervised Fine-Tuning
SFT on divergent branch-heavy CoT from DeepSeek-R1 yields worse generalization than convergent CoT from gpt-oss despite lower loss, but filtering frequent branches improves average performance by 3.6% on five reasoning benchmarks.
-
SPHINX: A Synthetic Environment for Visual Perception and Reasoning
SPHINX generates synthetic visual puzzles for benchmarking LVLMs, where GPT-5 scores 51.1% and RLVR training improves both in-domain and external visual reasoning performance.
-
Boosting Reinforcement Learning with Verifiable Rewards via Randomly Selected Few-Shot Guidance
FEST improves RLVR sample efficiency on math and coding benchmarks by combining supervised signals, on-policy signals, and decaying weights on just 128 randomly chosen demonstrations, matching full-dataset baselines.
-
A Survey of Reinforcement Learning for Large Reasoning Models
A survey compiling RL methods, challenges, data resources, and applications for enhancing reasoning in large language models and large reasoning models since DeepSeek-R1.