Beam Search as Test-Time Self-Distillation via Counterfactual Contexts
cs.LG, cs.AI
Submitted: 2026-09-29
Updated: 2026-09-29
Terminology
Sources
- Explaining and Improving Contrastive Decoding by Extrapolating the Probabilities of a Huge and Hypothetical LM
- Evaluating Large Language Models Trained on Code
- Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor
- Distilling the Knowledge in a Neural Network
- On Almost Surely Safe Alignment of Large Language Models at Inference-Time
- Scalable Power Sampling: Unlocking Efficient, Training-Free Reasoning for LLMs via Distribution Sharpening
- Reasoning with Sampling: Your Base Model is Smarter Than You Think
- The Model Knows, the Decoder Finds: Future Value Guided Particle Power Sampling
- Direct Preference Optimization: Your Language Model is Secretly a Reward Model
- Self-Distillation Enables Continual Learning
- Thinking by Subtraction: Confidence-Driven Contrastive Decoding for LLM Reasoning
- Self-Consistency Improves Chain of Thought Reasoning in Language Models
- Distilling Morphology-Conditioned Hypernetworks for Efficient Universal Morphology Control
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