StepCodeReasoner aligns code reasoning with verifiable stepwise execution traces via print anchors and bi-level GRPO reinforcement learning, reaching SOTA results on CRUXEval (91.1%) and LiveCodeBench (86.5%) for a 7B model.
hub
Understanding R1-Zero-Like Training: A Critical Perspective
80 Pith papers cite this work. Polarity classification is still indexing.
80
Pith papers citing it
abstract
DeepSeek-R1-Zero has shown that reinforcement learning (RL) at scale can directly enhance the reasoning capabilities of LLMs without supervised fine-tuning. In this work, we critically examine R1-Zero-like training by analyzing its two core components: base models and RL. We investigate a wide range of base models, including DeepSeek-V3-Base, to understand how pretraining characteristics influence RL performance. Our analysis reveals that DeepSeek-V3-Base already exhibit ''Aha moment'', while Qwen2.5 base models demonstrate strong reasoning capabilities even without prompt templates, suggesting potential pretraining biases. Additionally, we identify an optimization bias in Group Relative Policy Optimization (GRPO), which artificially increases response length (especially for incorrect outputs) during training. To address this, we introduce Dr. GRPO, an unbiased optimization method that improves token efficiency while maintaining reasoning performance. Leveraging these insights, we present a minimalist R1-Zero recipe that achieves 43.3% accuracy on AIME 2024 with a 7B base model, establishing a new state-of-the-art. Our code is available at https://github.com/sail-sg/understand-r1-zero.
hub tools
citation-role summary
background 1
citation-polarity summary
claims ledger
- abstract DeepSeek-R1-Zero has shown that reinforcement learning (RL) at scale can directly enhance the reasoning capabilities of LLMs without supervised fine-tuning. In this work, we critically examine R1-Zero-like training by analyzing its two core components: base models and RL. We investigate a wide range of base models, including DeepSeek-V3-Base, to understand how pretraining characteristics influence RL performance. Our analysis reveals that DeepSeek-V3-Base already exhibit ''Aha moment'', while Qwen2.5 base models demonstrate strong reasoning capabilities even without prompt templates, suggestin
co-cited works
roles
background 1polarities
background 1representative citing papers
Distillation signals align better with ideal updates on incorrect student rollouts than correct ones, with optimal teacher context depending on student capacity and task.
DeepRefine refines agent-compiled knowledge bases via multi-turn abductive diagnosis and RL training with a GBD reward, yielding consistent downstream task gains.
The cancellation hypothesis shows how rollout-level rewards produce token-level credit assignment in critic-free RL through cancellation of opposing signals on shared tokens, with empirical support and batching interventions that enhance performance.
DiagnosticIQ benchmark shows frontier LLMs perform similarly on standard rule-to-action tasks but lose substantial accuracy under distractor expansion and condition inversion, pointing to calibration as the key deployment issue.
The cumulative token IS ratio gives unbiased prefix correction and lower variance than full-sequence ratios for token-level gradients in LLM policy optimization, enabling CTPO to outperform GRPO and GSPO baselines on mathematical reasoning tasks.
HORA adaptively allocates rollouts using hit utility to improve Pass@K over compute-matched GRPO on math reasoning benchmarks while preserving Pass@1.
POPO uses bounded importance sampling on positive rollouts and a siamese policy network to achieve implicit negative gradients and stable optimization, matching or exceeding GRPO on math benchmarks such as 36.67% on AIME 2025.
Weblica scales RL training for visual web agents by building thousands of reproducible environments through HTTP caching for stable replays and LLM synthesis from real sites, yielding an 8B model that beats similar open baselines on navigation benchmarks.
PairAlign learns compact audio token sequences via self-alignment of paired content views using an autoregressive decoder, achieving strong cross-view consistency and edit-distance preservation while reducing token count by 55% on TIMIT.
AffectGPT-RL applies reinforcement learning to optimize non-differentiable emotion wheel metrics in open-vocabulary multimodal emotion recognition, yielding performance gains and state-of-the-art results on basic emotion recognition benchmarks.
OPD for LLMs suffers length inflation and repetition collapse; StableOPD uses reference divergence and rollout mixing to prevent it and improve math reasoning performance by 7.2% on average.
PRIME enables agents to proactively reason in user-centric tasks by iteratively evolving structured memories from interaction trajectories without gradient-based training.
A single LLM improves its own reasoning by self-distilling from privileged verified traces as teacher to its question-only student policy, outperforming off-policy distillation and RL on math benchmarks with better token efficiency.
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.
Uncertainty trace profiles from LM reasoning traces predict correct final answers with AUROC up to 0.807 and enable early error detection using only initial tokens.
EqLen is a sample-construction framework that builds equal-length paired segments via dual-track generation and masking for stable group-relative RL in sequences, reframing the length problem as a comparison-unit issue rather than loss scaling.
Freshness-Aware PER augments prioritized experience replay with exponential age decay based on effective sample size to enable successful reuse of trajectories in LLM and VLM reinforcement learning, outperforming on-policy baselines on agentic tasks.
A multi-agent binary reward system with unbiased GRPO post-training on ICLR-320 data outperforms baselines on expert-rated novelty, feasibility, and effectiveness for scientific idea generation.
FoPO trains LLMs for strategic reasoning by combining self-interest with opponent modeling in policy optimization, yielding gains on two new datasets and better out-of-domain generalization than standard baselines.
DEONTICBENCH is a new benchmark of 6,232 deontic reasoning tasks from U.S. legal domains where frontier LLMs reach only ~45% accuracy and symbolic Prolog assistance plus RL training still fail to solve tasks reliably.
TGPO improves on-policy LLM distillation by using teacher predictions conditioned on student rollouts to supply informative guidance when the two distributions diverge.
Policy entropy remains constant in flow-matching models during RLHF due to fixed noise schedules while perceptual diversity collapses from mode-seeking policy gradients, so perceptual entropy constraints are introduced to preserve diversity and improve quality.
HölderPO unifies token aggregation in GRPO via the Hölder mean with dynamic p annealing, reporting 54.9% average math-benchmark accuracy and 93.8% ALFWorld success.
citing papers explorer
-
StepCodeReasoner: Aligning Code Reasoning with Stepwise Execution Traces via Reinforcement Learning
StepCodeReasoner aligns code reasoning with verifiable stepwise execution traces via print anchors and bi-level GRPO reinforcement learning, reaching SOTA results on CRUXEval (91.1%) and LiveCodeBench (86.5%) for a 7B model.
-
Unmasking On-Policy Distillation: Where It Helps, Where It Hurts, and Why
Distillation signals align better with ideal updates on incorrect student rollouts than correct ones, with optimal teacher context depending on student capacity and task.
-
DeepRefine: Agent-Compiled Knowledge Refinement via Reinforcement Learning
DeepRefine refines agent-compiled knowledge bases via multi-turn abductive diagnosis and RL training with a GBD reward, yielding consistent downstream task gains.
-
The Cancellation Hypothesis in Critic-Free RL: From Outcome Rewards to Token Credits
The cancellation hypothesis shows how rollout-level rewards produce token-level credit assignment in critic-free RL through cancellation of opposing signals on shared tokens, with empirical support and batching interventions that enhance performance.
-
DiagnosticIQ: A Benchmark for LLM-Based Industrial Maintenance Action Recommendation from Symbolic Rules
DiagnosticIQ benchmark shows frontier LLMs perform similarly on standard rule-to-action tasks but lose substantial accuracy under distractor expansion and condition inversion, pointing to calibration as the key deployment issue.
-
Rethinking Importance Sampling in LLM Policy Optimization: A Cumulative Token Perspective
The cumulative token IS ratio gives unbiased prefix correction and lower variance than full-sequence ratios for token-level gradients in LLM policy optimization, enabling CTPO to outperform GRPO and GSPO baselines on mathematical reasoning tasks.
-
Where to Spend Rollouts: Hit-Utility Optimal Rollout Allocation for Group-Based RLVR
HORA adaptively allocates rollouts using hit utility to improve Pass@K over compute-matched GRPO on math reasoning benchmarks while preserving Pass@1.
-
Beyond Negative Rollouts: Positive-Only Policy Optimization with Implicit Negative Gradients
POPO uses bounded importance sampling on positive rollouts and a siamese policy network to achieve implicit negative gradients and stable optimization, matching or exceeding GRPO on math benchmarks such as 36.67% on AIME 2025.
-
Weblica: Scalable and Reproducible Training Environments for Visual Web Agents
Weblica scales RL training for visual web agents by building thousands of reproducible environments through HTTP caching for stable replays and LLM synthesis from real sites, yielding an 8B model that beats similar open baselines on navigation benchmarks.
-
PairAlign: A Framework for Sequence Tokenization via Self-Alignment with Applications to Audio Tokenization
PairAlign learns compact audio token sequences via self-alignment of paired content views using an autoregressive decoder, achieving strong cross-view consistency and edit-distance preservation while reducing token count by 55% on TIMIT.
-
AffectGPT-RL: Revealing Roles of Reinforcement Learning in Open-Vocabulary Emotion Recognition
AffectGPT-RL applies reinforcement learning to optimize non-differentiable emotion wheel metrics in open-vocabulary multimodal emotion recognition, yielding performance gains and state-of-the-art results on basic emotion recognition benchmarks.
-
Demystifying OPD: Length Inflation and Stabilization Strategies for Large Language Models
OPD for LLMs suffers length inflation and repetition collapse; StableOPD uses reference divergence and rollout mixing to prevent it and improve math reasoning performance by 7.2% on average.
-
PRIME: Training Free Proactive Reasoning via Iterative Memory Evolution for User-Centric Agent
PRIME enables agents to proactively reason in user-centric tasks by iteratively evolving structured memories from interaction trajectories without gradient-based training.
-
Self-Distilled Reasoner: On-Policy Self-Distillation for Large Language Models
A single LLM improves its own reasoning by self-distilling from privileged verified traces as teacher to its question-only student policy, outperforming off-policy distillation and RL on math benchmarks with better token efficiency.
-
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.
-
Tracing Uncertainty in Language Model "Reasoning"
Uncertainty trace profiles from LM reasoning traces predict correct final answers with AUROC up to 0.807 and enable early error detection using only initial tokens.
-
Rethinking the Comparison Unit in Sequence-Level Reinforcement Learning: An Equal-Length Paired Training Framework from Loss Correction to Sample Construction
EqLen is a sample-construction framework that builds equal-length paired segments via dual-track generation and masking for stable group-relative RL in sequences, reframing the length problem as a comparison-unit issue rather than loss scaling.
-
Freshness-Aware Prioritized Experience Replay for LLM/VLM Reinforcement Learning
Freshness-Aware PER augments prioritized experience replay with exponential age decay based on effective sample size to enable successful reuse of trajectories in LLM and VLM reinforcement learning, outperforming on-policy baselines on agentic tasks.
-
Debate as Reward: A Multi-Agent Reward System for Scientific Ideation via RL Post-Training
A multi-agent binary reward system with unbiased GRPO post-training on ICLR-320 data outperforms baselines on expert-rated novelty, feasibility, and effectiveness for scientific idea generation.
-
Foresight Optimization for Strategic Reasoning in Large Language Models
FoPO trains LLMs for strategic reasoning by combining self-interest with opponent modeling in policy optimization, yielding gains on two new datasets and better out-of-domain generalization than standard baselines.
-
DeonticBench: A Benchmark for Reasoning over Rules
DEONTICBENCH is a new benchmark of 6,232 deontic reasoning tasks from U.S. legal domains where frontier LLMs reach only ~45% accuracy and symbolic Prolog assistance plus RL training still fail to solve tasks reliably.
-
Teacher-Guided Policy Optimization for LLM Distillation
TGPO improves on-policy LLM distillation by using teacher predictions conditioned on student rollouts to supply informative guidance when the two distributions diverge.
-
When Policy Entropy Constraint Fails: Preserving Diversity in Flow-based RLHF via Perceptual Entropy
Policy entropy remains constant in flow-matching models during RLHF due to fixed noise schedules while perceptual diversity collapses from mode-seeking policy gradients, so perceptual entropy constraints are introduced to preserve diversity and improve quality.
-
H\"older Policy Optimisation
HölderPO unifies token aggregation in GRPO via the Hölder mean with dynamic p annealing, reporting 54.9% average math-benchmark accuracy and 93.8% ALFWorld success.
-
Learning to Foresee: Unveiling the Unlocking Efficiency of On-Policy Distillation
On-policy distillation gains efficiency from early foresight in module allocation and low-rank update directions, enabling EffOPD to accelerate training by 3x via adaptive extrapolation without extra modules or tuning.
-
Seir\^enes: Adversarial Self-Play with Evolving Distractions for LLM Reasoning
Seirênes trains LLMs via adversarial self-play to generate and overcome evolving distractions, producing gains of 7-10 points on math reasoning benchmarks and exposing blind spots in larger models.
-
UNIPO: Unified Interactive Visual Explanation for RL Fine-Tuning Policy Optimization
UNIPO is the first unified interactive visualization tool exposing token-level training dynamics of RL fine-tuning algorithms for LLMs through high-level overviews, step inspectors, and side-by-side comparisons.
-
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.
-
Power Reinforcement Post-Training of Text-to-Image Models with Super-Linear Advantage Shaping
Super-Linear Advantage Shaping (SLAS) introduces a non-linear geometric policy update for RL post-training of text-to-image models that reshapes the local policy space via advantage-dependent Fisher-Rao weighting to reduce reward hacking and improve performance over GRPO baselines.
-
G-Zero: Self-Play for Open-Ended Generation from Zero Data
G-Zero uses the Hint-δ intrinsic reward to drive co-evolution between a Proposer and Generator via GRPO and DPO, providing a theoretical suboptimality guarantee for self-improvement from internal dynamics alone.
-
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.
-
Structured Recurrent Mixers for Massively Parallelized Sequence Generation
Structured Recurrent Mixers enable algebraic switching between parallel training and recurrent inference representations, delivering higher efficiency, information capacity, and throughput than other linear-complexity models.
-
AIPO: : Learning to Reason from Active Interaction
AIPO trains LLMs to expand their reasoning capability boundary via active multi-agent interaction with Verify, Knowledge, and Reasoning agents during RLVR, using importance sampling and clipping to handle feedback, then drops the agents at inference.
-
Listwise Policy Optimization: Group-based RLVR as Target-Projection on the LLM Response Simplex
LPO reframes group-based RLVR as explicit target-projection on the LLM response simplex and performs exact divergence minimization to achieve monotonic listwise improvement with bounded gradients.
-
ZAYA1-8B Technical Report
ZAYA1-8B is a reasoning MoE model with 700M active parameters that matches larger models on math and coding benchmarks and reaches 91.9% on AIME'25 via Markovian RSA test-time compute.
-
Binary Rewards and Reinforcement Learning: Fundamental Challenges
Binary rewards make the set of reward-maximizing policies infinite in policy gradients; KL control selects the filtered base model but misspecification drives collapse to concentrated valid outputs instead.
-
Injecting Distributional Awareness into MLLMs via Reinforcement Learning for Deep Imbalanced Regression
A plug-and-play RL method adds batch-level distributional supervision via CCC rewards to reduce regression-to-the-mean in MLLMs on imbalanced regression benchmarks.
-
Odysseus: Scaling VLMs to 100+ Turn Decision-Making in Games via Reinforcement Learning
Odysseus adapts PPO with a turn-level critic and leverages pretrained VLM action priors to train agents achieving at least 3x average game progress over frontier models in long-horizon Super Mario Land.
-
Co-Evolving Policy Distillation
CoPD integrates multiple expert capabilities by running parallel RLVR training with bidirectional online policy distillation among experts, outperforming mixed RLVR and sequential OPD while surpassing domain-specific experts on text-image-video reasoning.
-
From Local Indices to Global Identifiers: Generative Reranking for Recommender Systems via Global Action Space
GloRank reformulates list-wise reranking as token generation over a global item identifier space, using supervised pre-training followed by reinforcement learning to maximize list-wise utility and outperforming baselines on benchmarks and industrial data.
-
Hidden States Know Where Reasoning Diverges: Credit Assignment via Span-Level Wasserstein Distance
Span-level Wasserstein distances between hidden-state distributions of correct and incorrect rollouts provide a self-supervised signal to reweight advantages in GRPO, improving fine-grained credit assignment on math and code tasks.
-
TRN-R1-Zero: Text-rich Network Reasoning via LLMs with Reinforcement Learning Only
TRN-R1-Zero is an RL-only post-training method that lets LLMs perform zero-shot node, edge, and graph reasoning on text-rich networks without supervised data or larger-model distillation.
-
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.
-
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.
-
Asymmetric Advantage Modulation Calibrates Entropy Dynamics in RLVR
AsymGRPO decouples positive and negative advantage modulation in RLVR to separately boost useful entropy and suppress noisy entropy, improving LLM reasoning performance.
-
Can LLMs Learn to Reason Robustly under Noisy Supervision?
Online Label Refinement lets LLMs learn robust reasoning from noisy supervision by correcting labels when majority answers show rising rollout success and stable history, delivering 3-4% gains on math and reasoning benchmarks even at high noise levels.
-
MiniMax-M1: Scaling Test-Time Compute Efficiently with Lightning Attention
MiniMax-M1 is a 456B parameter hybrid-attention MoE model trained with CISPO RL that achieves performance comparable or superior to DeepSeek-R1 and Qwen3-235B on reasoning and software engineering tasks while training in three weeks on 512 GPUs.
-
VLM-R1: A Stable and Generalizable R1-style Large Vision-Language Model
VLM-R1 applies R1-style RL using rule-based rewards on visual tasks with clear ground truth to achieve competitive performance and superior generalization over SFT in vision-language models.