Self-Repulsive Sampling for Diffusion Language Models
cs.LG
Submitted: 2026-09-30
Updated: 2026-09-30
Code: https://github.com/sean-lamont/odd
Sources
- When More Sampling Hurts: The Modal Ceiling and Correlation Ceiling of Test-Time Scaling
- Are More LLM Calls All You Need? Towards Scaling Laws of Compound Inference Systems
- Evaluating Large Language Models Trained on Code
- Training Verifiers to Solve Math Word Problems
- Particle Guidance: non-I.I.D. Diverse Sampling with Diffusion Models
- Inference-Time Scaling of Diffusion Language Models via Trajectory Refinement
- DiffusionGemma Technical Report
- dVoting: Fast Voting for dLLMs
- Mercury: Ultra-Fast Language Models Based on Diffusion
- Determinantal point processes for machine learning
- Free Lunch for Pass@$k$? Low Cost Diverse Sampling for Diffusion Language Models
- Test-Time Scaling in Diffusion LLMs via Hidden Semi-Autoregressive Experts
- QuasiMoTTo: Quasi-Monte Carlo Test-Time Scaling
- A Survey on Diffusion Language Models
- Stein Variational Gradient Descent: A General Purpose Bayesian Inference Algorithm
- Self-Rewarding Sequential Monte Carlo for Masked Diffusion Language Models
- D5P4: Partition Determinantal Point Process for Diversity in Parallel Discrete Diffusion Decoding
- A Tale of Two Temperatures: Simple, Efficient, and Diverse Sampling from Diffusion Language Models
- Efficient semantic uncertainty quantification in language models via diversity-steered sampling
- A Universal Avoidance Method for Diverse Multi-branch Generation
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks