Derives an exact unbiased policy gradient for RL post-training of diffusion LLMs via entropy-guided step selection and one-step denoising rewards, achieving state-of-the-art results on coding and logical reasoning benchmarks.
hub Mixed citations
Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models
Mixed citation behavior. Most common role is background (60%).
52
Pith papers citing it
Background
60% of classified citations
abstract
Diffusion language models offer unique benefits over autoregressive models due to their potential for parallelized generation and controllability, yet they lag in likelihood modeling and are limited to fixed-length generation. In this work, we introduce a class of block diffusion language models that interpolate between discrete denoising diffusion and autoregressive models. Block diffusion overcomes key limitations of both approaches by supporting flexible-length generation and improving inference efficiency with KV caching and parallel token sampling. We propose a recipe for building effective block diffusion models that includes an efficient training algorithm, estimators of gradient variance, and data-driven noise schedules to minimize the variance. Block diffusion sets a new state-of-the-art performance among diffusion models on language modeling benchmarks and enables generation of arbitrary-length sequences. We provide the code, along with the model weights and blog post on the project page: https://m-arriola.com/bd3lms
hub tools
citation-role summary
background 6
method 3
baseline 1
citation-polarity summary
representative citing papers
Introduces the first NPU accelerator for diffusion language models with dLLM-specific ISA, hardware execution model, BAOS KV quantization, and 7nm RTL synthesis.
LLaDA is a scalable diffusion-based language model that matches autoregressive LLMs like LLaMA3 8B on tasks and surpasses GPT-4o on reversal poem completion.
The paper introduces Uni-E, a unified energy for DLMs that accounts for model capacity, dependency and invariance, can be computed exactly, and corrects distribution shifts from dependency and invariance.
Continuous language diffusion works by entering high-margin decoder basins where frozen T5 embeddings recover 93-96% of native decisions and linear readouts reach 97.9% agreement, implying models should be evaluated as representation-decoder systems.
Naive samplers beat published diffusion and flow models on gen-PPL with incoherent output, proving the metric unsound and motivating distributional evaluation suites.
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.
BASTION is a budget-aware speculative decoding framework with adaptive tree-structured block diffusion drafting that reports up to 6.61x speedup and 39% improvement over block-diffusion baselines.
DCDM replaces positional blocks with learnable semantic chunks via differentiable Chunking Attention, yielding consistent gains over block and unstructured diffusion baselines up to 1.5B parameters.
Discrete diffusion models learn data support before frequencies because the exact reverse process decomposes edits into a dominant validity scale and a finer probability coefficient.
DARE reuses up to 87% of attention activations in diffusion LLMs through KV caching and output reuse, delivering 1.2x per-layer latency gains with average performance drops of 1.2-2.0%.
NI Sampling accelerates discrete diffusion language models up to 14.3 times by training a neural indicator to select which tokens to sample at each step using a trajectory-preserving objective.
DiffTSP applies discrete diffusion to knowledge graph triple set prediction, recovering all missing triples simultaneously via edge-masking noise reversal and a structure-aware transformer, achieving SOTA on three datasets.
BARD bridges autoregressive and diffusion VLMs with progressive block merging plus stage-wise intra-diffusion distillation, delivering 3x speedup and new SOTA on open dVLMs using under 4.4M data points.
LangFlow is the first continuous diffusion language model to rival discrete diffusion on perplexity and generative perplexity while exceeding autoregressive baselines on several zero-shot tasks.
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 Stochastic Localization lets a single trained network support an entire family of per-token SNR paths for discrete sequence generation, with masked diffusion as a special case, and improves MAUVE scores when fine-tuning pretrained checkpoints.
Early and late denoising steps in masked diffusion LMs are robust to smaller-model replacement, enabling 17% FLOPs reduction with modest generative quality loss.
PartDiffuser is a semi-autoregressive discrete diffusion framework that generates high-fidelity 3D meshes from point clouds by combining inter-part autoregression with intra-part parallel diffusion using a part-aware DiT architecture.
MBD-LMs raise average tokens per forward pass from 3.47 to 6.19 (and to 9.34 with DMax) via multi-block teacher forcing and optimized parallel decoding while holding or slightly improving accuracy on math and code tasks.
R2LM combines causal attention with a reverse Mamba SSM sidecar to supply right-side context in dLLMs, claiming 2.4x-12.9x throughput gains over bidirectional dLLMs and 1.9x-2.9x over AR baselines while matching or exceeding quality.
TwoTower decouples autoregressive context from diffusion denoising using a frozen AR tower and bidirectional diffusion tower, retaining 98.7% baseline quality with 2.42X throughput on a 30B model trained on 2.1T tokens.
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.
On-policy distillation from a frozen autoregressive teacher to a bidirectional student eliminates train-inference mismatch and enables data-efficient ARLM-to-DLM conversion.
citing papers explorer
-
TextLDM: Language Modeling with Continuous Latent Diffusion
TextLDM applies DiT-style latent diffusion with flow matching to language modeling via a REPA-aligned VAE, outperforming prior diffusion LMs and matching GPT-2 when trained from scratch on OpenWebText2.
-
LLaDA2.0: Scaling Up Diffusion Language Models to 100B
LLaDA2.0 scales discrete diffusion language models to 100B parameters via systematic conversion from autoregressive models using a 3-phase WSD training scheme and releases open-source 16B and 100B MoE variants.
-
Dream 7B: Diffusion Large Language Models
Dream 7B is a 7B diffusion LLM that refines sequences in parallel via denoising and outperforms prior diffusion models on general, mathematical, and coding benchmarks with added flexibility in generation order and quality-speed tradeoffs.