Experts Discuss Forensics for Open-Source Models

John Schulman and Will DePue explore forensics for open-source models, noting that training data and backdoors can be inferred through analysis. Techniques like measuring logprobs and reverse-engineering reinforcement learning could help trace model distillation.

2026-07-27 ~ 2026-07-27 · 3 related posts

• John Schulman: Forensics for Model Training Data and Backdoors Understudied — johnschulman2 · 2026-07-27
• Will DePue: Reverse RL and Large-Scale Sampling Could Recover Model Distribution — willdepue · 2026-07-27
• John Schulman says distillation forensics can also be measured with logprobs — johnschulman2 · 2026-07-27