The Distillation Game: Adaptive Evaluations & Efficient Defenses
cs.LG, cs.AI
Submitted: 2026-05-21
Updated: 2026-10-07
Code: https://github.com/ysfalh/distillation-game
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Qwen Technical Report
- Training Verifiers to Solve Math Word Problems
- Information-Preserving Reformulation of Reasoning Traces for Antidistillation
- Towards Distillation-Resistant Large Language Models: An Information-Theoretic Perspective
- The Llama 3 Herd of Models
- DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
- Measuring Mathematical Problem Solving With the MATH Dataset
- DOGe: Defensive Output Generation for LLM Protection Against Knowledge Distillation
- What Does It Mean to Break a Distillation Defense?
- Antidistillation Sampling
- Proximal Policy Optimization Algorithms
- Antidistillation Fingerprinting
- How to Steal Reasoning Without Reasoning Traces
- Model Immunization from a Condition Number Perspective
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