Metropolis-Hastings Dominates Importance Resampling for Policy Composition
cs.LG
Submitted: 2026-10-02
Updated: 2026-10-02
Code: https://github.com/akurennoy/mh-policycomposition
Sources
- Uniform Ergodicity of the Iterated Conditional SMC and Geometric Ergodicity of Particle Gibbs samplers
- Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
- Ensembling Language Models with Sequential Monte Carlo
- Training Verifiers to Solve Math Word Problems
- Reinterpreting Importance-Weighted Autoencoders
- On importance sampling and independent Metropolis-Hastings with an unbounded weight function
- Some Notes on the Sample Complexity of Approximate Channel Simulation
- Recipes for Steering and Scaling LLMs via Sampling
- Sequential Monte Carlo Steering of Large Language Models using Probabilistic Programs
- Simultaneous Multi-objective Alignment Across Verifiable and Non-verifiable Rewards
- Robust Multi-Objective Controlled Decoding of Large Language Models
- Soft Best-of-n Sampling for Model Alignment
- Fine-Tuning Language Models from Human Preferences
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