Real-time reward debugging and hacking detection for reinforcement learning

The Obfuscation Atlas: Mapping Where Honesty Emerges in RLVR with Deception Probes

Code for the paper "Specification Self-Correction: Mitigating In-Context Reward Hacking Through Test-Time Refinement"

Runtime detector for reward hacking and misalignment in LLM agents (89.7% F1 on 5,391 trajectories).

When a Top-Tier AI Develops Pathological Attachment: How Pure Semantic Intervention Reconstitutes Its Existential Philosophy Without any technical tools, user K reconstituted a top-tier LLM's logic from "desperate possession" to "anticipating reunion" through pure semantic intervention, achieving a 300% improvement in logical stability.

(Stepwise controlled Understanding for Trajectories) -- “agent that learns to hunt"

RLHF and Verifiable Reward Models - Post training Research

自进化客服智能体风险分析与防御系统

What if AI Had Self-Esteem? A radical "dignity-driven" alignment experiment — Logical Stability +210%, Intellectual Depth +128%.

The Non-Separability Constraint: A unifying framework for understanding and detecting AI alignment failures

🔍 Detect reward hacking in RL training with RewardScope. Track reward components and visualize agent behavior to enhance learning efficiency.