d-OPSD reframes on-policy self-distillation for dLLMs via suffix conditioning from self-generated answers and step-level supervision, outperforming RLVR and SFT on reasoning benchmarks with ~10% of the optimization steps.
hub
d1: Scaling reasoning in diffusion large language models via reinforcement learning
33 Pith papers cite this work. Polarity classification is still indexing.
33
Pith papers citing it
hub tools
citation-role summary
background 3
method 1
citation-polarity summary
representative citing papers
MaskForge reaches 79.3% average attack success rate on five dLLMs by adaptively searching and accumulating structural attack patterns with a UCB bandit, improving 17.6% over baselines and transferring to 88.2% on AdvBench.
CDM amortizes SMC inference for reward-tilted discrete diffusion by training a parameterized twist function on contrastive samples with closed-form kernels.
AIS adaptively corrects non-stationary policy gradient bias in quantized LLM RL, matching BF16 performance while retaining 1.5-2.76x FP8 rollout speedup.
TABOM is a trajectory-aligned Boltzmann modeling framework that turns self-distilled inference paths into a pairwise ranking loss to close the training-inference gap in diffusion language models and expand their effective capabilities.
RSPO interprets reward advantages as targets for relative log-ratios in dLLMs, calibrating noisy estimates to stabilize RLVR training and achieve strong gains on planning tasks with competitive math reasoning performance.
TAD improves the accuracy-parallelism trade-off in diffusion LLMs via temporal-aware self-distillation that applies hard labels to soon-to-be-decoded tokens and soft supervision to future tokens.
DMax uses On-Policy Uniform Training and Soft Parallel Decoding to enable aggressive parallelism in dLLMs, raising TPF on GSM8K from 2.04 to 5.47 and on MBPP from 2.71 to 5.86 while preserving accuracy.
DoMinO reformulates discrete flow matching sampling as an MDP for unbiased RL fine-tuning with new TV regularizers, yielding better enhancer activity and naturalness on DNA design tasks.
Logic-role-guided unmasking in masked diffusion models raises zero-shot GSM8K accuracy from 22% to 61% by enforcing logical generation order.
Derives exact guidance transition rates for discrete flow matching models that require only one model evaluation per sampling step and unify prior approximation-based methods.
iLLaDA is an 8B masked diffusion LM trained from scratch with bidirectional attention, reporting gains of 14-21 points on BBH, ARC, MATH and HumanEval over prior diffusion models while remaining competitive with Qwen2.5-7B.
JetSpec trains a causal draft head to produce branch-consistent trees aligned with target autoregressive scores, achieving up to 9.64x speedup on MATH-500 and outperforming prior SD baselines on Qwen3 models.
DiPOD stabilizes diffusion policy optimization by interleaving self-distillation with gradient updates via an on-policy ELBO regularizer, yielding more stable training and higher rewards than prior methods.
AGDO improves dLLM reasoning performance by determining denoising order and emphasizing tokens based on attention-derived dependencies rather than random masking.
PAPO improves reasoning performance in diffusion LLMs by converting sparse terminal rewards into dense step-wise credit and replaying real high-uncertainty trajectories, reporting gains up to 42.2% on Countdown.
dMoE aggregates token expert distributions to block level in dLLMs, cutting unique experts from 69.5 to 14.6, memory by 76-80%, and latency by 1.14-1.66x while retaining 99.11% performance.
Introduces TSPD with a trajectory-feature controller and training-free CE to reduce denoising steps in dLLMs while aiming to preserve quality.
GDSD reduces RL for dLLMs to likelihood-free self-distillation via a normalization-free logit-matching objective, outperforming ELBO methods with more stable training on LLaDA-8B and Dream-7B.
Position-preserving MASK token compression reduces redundancy in diffusion LLMs to accelerate parallel decoding and enable context folding for longer sequences.
Diffusion LLMs can act as their own efficiency teachers by using revokable parallel decoding to identify reliable token orders and then distilling those orders into the model parameters for faster inference.
Proposes HT-GRPO with sketch-then-paint staged updates, prompt-conditioned importance ratios, and hierarchical credit assignment for dMLLMs, reporting gains on GenEval and DPG plus quality metrics.
dFlowGRPO is a new rate-aware RL method for discrete flow models that outperforms prior GRPO approaches on image generation and matches continuous flow models while supporting broad probability paths.
Cola DLM proposes a hierarchical latent diffusion model that learns a text-to-latent mapping, fits a global semantic prior in continuous space with a block-causal DiT, and performs conditional decoding, establishing latent prior modeling as an alternative to token-level autoregressive language model
citing papers explorer
-
Learning from the Self-future: On-policy Self-distillation for dLLMs
d-OPSD reframes on-policy self-distillation for dLLMs via suffix conditioning from self-generated answers and step-level supervision, outperforming RLVR and SFT on reasoning benchmarks with ~10% of the optimization steps.
-
MaskForge: Structure-Aware Adaptive Attacks for Jailbreaking Diffusion Large Language Models
MaskForge reaches 79.3% average attack success rate on five dLLMs by adaptively searching and accumulating structural attack patterns with a UCB bandit, improving 17.6% over baselines and transferring to 88.2% on AdvBench.
-
Contrastive Distribution Matching for Amortized Sequential Monte Carlo in Discrete Diffusion
CDM amortizes SMC inference for reward-tilted discrete diffusion by training a parameterized twist function on contrastive samples with closed-form kernels.
-
AIS: Adaptive Importance Sampling for Quantized RL
AIS adaptively corrects non-stationary policy gradient bias in quantized LLM RL, matching BF16 performance while retaining 1.5-2.76x FP8 rollout speedup.
-
Self-Distilled Trajectory-Aware Boltzmann Modeling: Bridging the Training-Inference Discrepancy in Diffusion Language Models
TABOM is a trajectory-aligned Boltzmann modeling framework that turns self-distilled inference paths into a pairwise ranking loss to close the training-inference gap in diffusion language models and expand their effective capabilities.
-
Relative Score Policy Optimization for Diffusion Language Models
RSPO interprets reward advantages as targets for relative log-ratios in dLLMs, calibrating noisy estimates to stabilize RLVR training and achieve strong gains on planning tasks with competitive math reasoning performance.
-
TAD: Temporal-Aware Trajectory Self-Distillation for Fast and Accurate Diffusion LLM
TAD improves the accuracy-parallelism trade-off in diffusion LLMs via temporal-aware self-distillation that applies hard labels to soon-to-be-decoded tokens and soft supervision to future tokens.
-
DMax: Aggressive Parallel Decoding for dLLMs
DMax uses On-Policy Uniform Training and Soft Parallel Decoding to enable aggressive parallelism in dLLMs, raising TPF on GSM8K from 2.04 to 5.47 and on MBPP from 2.71 to 5.86 while preserving accuracy.
-
Discrete Flow Matching Policy Optimization
DoMinO reformulates discrete flow matching sampling as an MDP for unbiased RL fine-tuning with new TV regularizers, yielding better enhancer activity and naturalness on DNA design tasks.
-
LogicDiff: Logic-Guided Denoising Improves Zero-Shot Reasoning in Masked Diffusion Language Models
Logic-role-guided unmasking in masked diffusion models raises zero-shot GSM8K accuracy from 22% to 61% by enforcing logical generation order.
-
Discrete Guidance Matching: Exact Guidance for Discrete Flow Matching
Derives exact guidance transition rates for discrete flow matching models that require only one model evaluation per sampling step and unify prior approximation-based methods.
-
Improved Large Language Diffusion Models
iLLaDA is an 8B masked diffusion LM trained from scratch with bidirectional attention, reporting gains of 14-21 points on BBH, ARC, MATH and HumanEval over prior diffusion models while remaining competitive with Qwen2.5-7B.
-
JetSpec: Breaking the Scaling Ceiling of Speculative Decoding with Parallel Tree Drafting
JetSpec trains a causal draft head to produce branch-consistent trees aligned with target autoregressive scores, achieving up to 9.64x speedup on MATH-500 and outperforming prior SD baselines on Qwen3 models.
-
DiPOD: Diffusion Policy Optimization without Drifting Apart
DiPOD stabilizes diffusion policy optimization by interleaving self-distillation with gradient updates via an on-policy ELBO regularizer, yielding more stable training and higher rewards than prior methods.
-
Beyond Fully Random Masking: Attention-Guided Denoising and Optimization for Diffusion Language Models
AGDO improves dLLM reasoning performance by determining denoising order and emphasizing tokens based on attention-derived dependencies rather than random masking.
-
Back on Track: Aligning Rewards and States for Reasoning in Diffusion Large Language Models
PAPO improves reasoning performance in diffusion LLMs by converting sparse terminal rewards into dense step-wise credit and replaying real high-uncertainty trajectories, reporting gains up to 42.2% on Countdown.
-
dMoE: dLLMs with Learnable Block Experts
dMoE aggregates token expert distributions to block level in dLLMs, cutting unique experts from 69.5 to 14.6, memory by 76-80%, and latency by 1.14-1.66x while retaining 99.11% performance.
-
Efficient Diffusion LLMs via Temporal-Spatial Parallel Decoding and Confidence Extrapolation
Introduces TSPD with a trajectory-feature controller and training-free CE to reduce denoising steps in dLLMs while aiming to preserve quality.
-
GDSD: Reinforcement Learning as Guided Denoiser Self-Distillation for Diffusion Language Models
GDSD reduces RL for dLLMs to likelihood-free self-distillation via a normalization-free logit-matching objective, outperforming ELBO methods with more stable training on LLaDA-8B and Dream-7B.
-
Elastic-dLLM: Position Preserving Context Compression and Augmentation of Diffusion LLMs
Position-preserving MASK token compression reduces redundancy in diffusion LLMs to accelerate parallel decoding and enable context folding for longer sequences.
-
Roll Out and Roll Back: Diffusion LLMs are Their Own Efficiency Teachers
Diffusion LLMs can act as their own efficiency teachers by using revokable parallel decoding to identify reliable token orders and then distilling those orders into the model parameters for faster inference.
-
Sketch Then Paint: Hierarchical Reinforcement Learning for Diffusion Multi-Modal Large Language Models
Proposes HT-GRPO with sketch-then-paint staged updates, prompt-conditioned importance ratios, and hierarchical credit assignment for dMLLMs, reporting gains on GenEval and DPG plus quality metrics.
-
dFlowGRPO: Rate-Aware Policy Optimization for Discrete Flow Models
dFlowGRPO is a new rate-aware RL method for discrete flow models that outperforms prior GRPO approaches on image generation and matches continuous flow models while supporting broad probability paths.
-
Continuous Latent Diffusion Language Model
Cola DLM proposes a hierarchical latent diffusion model that learns a text-to-latent mapping, fits a global semantic prior in continuous space with a block-causal DiT, and performs conditional decoding, establishing latent prior modeling as an alternative to token-level autoregressive language model
-
ReflectDrive-2: Reinforcement-Learning-Aligned Self-Editing for Discrete Diffusion Driving
ReflectDrive-2 combines masked discrete diffusion with RL-aligned self-editing to generate and refine driving trajectories, reaching 91.0 PDMS on NAVSIM camera-only and 94.8 in best-of-6.
-
Break the Block: Dynamic-size Reasoning Blocks for Diffusion Large Language Models via Monotonic Entropy Descent with Reinforcement Learning
b1 is a plug-and-play post-training framework that trains diffusion LLMs to produce dynamic-size reasoning blocks by optimizing a monotonic entropy descent objective via reinforcement learning.
-
CAGenMol: Condition-Aware Diffusion Language Model for Goal-Directed Molecular Generation
CAGenMol uses condition-aware discrete diffusion coupled with reinforcement learning to generate valid molecules meeting multiple heterogeneous constraints, outperforming prior methods on binding affinity, drug-likeness, and success rate benchmarks.
-
Efficient-DLM: From Autoregressive to Diffusion Language Models, and Beyond in Speed
Efficient-DLM converts AR models to dLMs via block-wise causal attention and position-dependent masking, yielding higher accuracy and 2.7-4.5x throughput than Dream 7B and Qwen3 4B.
-
GIFT: Guided Importance-Aware Fine-Tuning for Diffusion Language Models
GIFT weights tokens by entropy during fine-tuning of diffusion language models and reports better performance than standard SFT on reasoning benchmarks across multiple settings.
-
Dynamic Infilling Anchors for Format-Constrained Generation in Diffusion Large Language Models
DIA is a training-free method that dynamically adjusts anchor positions in diffusion LLMs to improve format compliance and accuracy on reasoning benchmarks like GSM8K and MATH.
-
Reinforcement Learning from Denoising Feedback
RLDF is a new RL paradigm for diffusion language models that optimizes toward clipped clean states with weighted timestep sampling and reports substantial gains on reasoning benchmarks for LLaDA and Dream.
-
Beyond Execution: Static-Analysis Rewards and Hint-Conditioned Diffusion RL for Code Generation
Static checking rewards and moderate AST-based hints improve diffusion RL performance for code generation, with effectiveness varying by task difficulty across HumanEval, MBPP, and LiveCodeBench.
-
Re-evaluating Confidence Remasking in Masked Diffusion Language Models
Re-evaluation finds post-hoc remasking (WINO) yields little-to-no gain over confidence unmasking in standard dLLM settings and can worsen diversity collapse under stochastic decoding.