Max Kaufmann(@Max_A_Kaufmann):Is training against the CoT always bad? RL training can lead to obfuscated CoT making it difficult to 'read an LLMs thoughts'. How can we predict when obfuscation occurs?🤔 Our new @GoogleDeepMind paper introduces a framework to predict this before training starts!

Max Kaufmann

@Max_A_Kaufmann

Maxxing @AnthropicAI. On leave from ML PhD @UofT/@VectorInst | prev. founding team @AISecurityInst, VSR 🎓 @GoogleDeepMind

加入 May 2017

628 正在关注 591 粉丝

Max Kaufmann@Max_A_Kaufmann

2026.04.01 18:07

Is training against the CoT always bad? RL training can lead to obfuscated CoT making it difficult to 'read an LLMs thoughts'. How can we predict when obfuscation occurs?🤔 Our new @GoogleDeepMind paper introduces a framework to predict this before training starts!

显示更多