PAC-CF: Calibrating Irreversible Frontier Pruning in LLM-Guided Search
cs.LG, cs.AI, stat.ML
Submitted: 2026-04-15
Updated: 2026-10-06
Comments: 30 pages, 3 figures. Major revision. Earlier versions circulated under the title PAC-MCTS and reported controlled proof-of-concept experiments. This version introduces Native-Trace conformal calibration, frozen-margin deployment, controller-agnostic integration, and benchmark-based multi-domain evaluation
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- EquivPruner: Boosting Efficiency and Quality in LLM-Based Search via Action Pruning
- Systematic Bias in Large Language Models: Discrepant Response Patterns in Binary vs. Continuous Judgment Tasks
- Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters
- MALinZero: Efficient Low-Dimensional Search for Mastering Complex Multi-Agent Planning
- Cost-Awareness in Tree-Search LLM Planning: A Systematic Study
- Language Agent Tree Search Unifies Reasoning Acting and Planning in Language Models
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: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks