Set diffusion factorizes likelihood over arbitrary token sets and uses a set-causal diffusion architecture to support KV caching and any-order decoding, yielding improved speed-quality tradeoffs versus prior diffusion LMs.
Any-order flexible length masked diffusion
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
representative citing papers
Discrete Tilt Matching recasts dLLM fine-tuning as state-level matching of tilted local unmasking posteriors, producing a stable weighted cross-entropy loss that improves Sudoku and Countdown performance when applied to LLaDA-8B-Instruct.
SCSI iteratively refines a self-consistent transport map to invert black-box corruptions and enable generative modeling of clean data.
VoidPadding decouples padding from termination in MDLMs via a new [VOID] token, delivering +17.84 average benchmark points and 55.7% fewer decoding steps on Dream-7B-Instruct.
Morph is a flexible-size 3D molecular generative model using unbalanced optimal transport on geometric graphs that matches fixed-size SOTA performance while enabling out-of-distribution generation.
Fast-dLLM++ generalizes Fast-dLLM decoding to heterogeneous confidence profiles via Fréchet profile selection, delivering up to 37% throughput gains on GSM8K, MATH, HumanEval, and MBPP with LLaDA-8B.
ME-DLM augments parallel masked diffusion models with edit-distance-supervised refinements to raise quality on coding and math benchmarks while using far fewer diffusion steps.
A token-history 'credit' score fused into current logits reduces redundant re-masking in diffusion LLM decoding, yielding up to 5.48× tokens-per-forward speedup with small accuracy gains.
citing papers explorer
-
Set Diffusion: Interpolating Token Orderings Between Autoregression and Diffusion for Fast and Flexible Decoding
Set diffusion factorizes likelihood over arbitrary token sets and uses a set-causal diffusion architecture to support KV caching and any-order decoding, yielding improved speed-quality tradeoffs versus prior diffusion LMs.
-
Discrete Tilt Matching
Discrete Tilt Matching recasts dLLM fine-tuning as state-level matching of tilted local unmasking posteriors, producing a stable weighted cross-entropy loss that improves Sudoku and Countdown performance when applied to LLaDA-8B-Instruct.
-
Generative Modeling from Black-box Corruptions via Self-Consistent Stochastic Interpolants
SCSI iteratively refines a self-consistent transport map to invert black-box corruptions and enable generative modeling of clean data.
-
VoidPadding: Let [VOID] Handle Padding in Masked Diffusion Language Models so that [EOS] Can Focus on Semantic Termination
VoidPadding decouples padding from termination in MDLMs via a new [VOID] token, delivering +17.84 average benchmark points and 55.7% fewer decoding steps on Dream-7B-Instruct.
-
Generative Molecular Morphing for Flexible-Size Design via Unbalanced Optimal Transport
Morph is a flexible-size 3D molecular generative model using unbalanced optimal transport on geometric graphs that matches fixed-size SOTA performance while enabling out-of-distribution generation.
-
Fast-dLLM++: Fr\'{e}chet Profile Decoding for Faster Diffusion LLM Inference
Fast-dLLM++ generalizes Fast-dLLM decoding to heterogeneous confidence profiles via Fréchet profile selection, delivering up to 37% throughput gains on GSM8K, MATH, HumanEval, and MBPP with LLaDA-8B.
-
Edit-Based Refinement for Parallel Masked Diffusion Language Models
ME-DLM augments parallel masked diffusion models with edit-distance-supervised refinements to raise quality on coding and math benchmarks while using far fewer diffusion steps.
-
CreditDecoding: Accelerating Parallel Decoding in Diffusion Large Language Models with Trace Credit
A token-history 'credit' score fused into current logits reduces redundant re-masking in diffusion LLM decoding, yielding up to 5.48× tokens-per-forward speedup with small accuracy gains.