Give Up to Get More: Priors, Not Policies, Govern Test-Time Compute Allocation
cs.LG
Submitted: 2026-11-04
Updated: 2026-11-04
Code: https://github.com/nepersoned/priors-not-policies
Terminology
Sources
- Sampling Luck Masquerades as Allocation Gain: Auditing Test-Time Budget Allocation for Neural Combinatorial Optimization
- When Does Reallocating a Sampling Budget Pay? Instance-wise Allocation for Neural Routing Solvers
- Online Difficulty Filtering for Reasoning Oriented Reinforcement Learning
- Recent Developments in Pandora's Box Problem: Variants and Applications
- Large Language Monkeys: Scaling Inference Compute with Repeated Sampling
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- Thinking Hard, Not Smart: Reasoning Models Fail to Ration Test-Time Compute Across Questions
- Neural Solver Selection for Combinatorial Optimization
- Efficient Test-Time Scaling via Self-Calibration
- Sample, Mix, or Verify? Allocating Test-Time Compute under Imperfect Verifiers
- Sampling headroom is not selection gain: a compute-value audit of test-time scaling for video world models
- Tulu 3: Pushing Frontiers in Open Language Model Post-Training
- Uncertainty-Aware Budget Allocation for Adaptive Test-Time Reasoning
- Know What You Don't Know: Uncertainty Calibration of Process Reward Models
- HPO-B: A Large-Scale Reproducible Benchmark for Black-Box HPO based on OpenML
- AdaBoN: Adaptive Best-of-N Alignment
- Shrinking the Generation-Verification Gap with Weak Verifiers
- The Limits of Inference Scaling Through Resampling
- Every Rollout Counts: Optimal Resource Allocation for Efficient Test-Time Scaling
Related papers
- A Physics-Regulated Neural Framework for Learning 3D Grain Growth Dynamics
- FlexMoE: Learning One-for-All Multi-Budget Expert Pruning for Flexible MoE Deployment
- OptiMind: Improving Small Language Models for Optimization Modeling through Formulation Guidance and Multi-Turn Learning
- Staying in the Sweet Spot: Boost Reasoning Evolution via Responsive Rollout Difficulty Manipulation
- Examining the Impact of Data and Training Methods on Emergent and Subliminal Misalignment
- Rethinking Black-Box On-Policy Distillation: A Rubric-based Analysis