John's Unnamed Lab

AI & ML interests

Safety, Interpretability, Control, Representations, Understanding

Recent Activity

LorenaYannnnn  updated a dataset about 20 hours ago

hewittlab/monitor_cot_for_hack-dataset-length

LorenaYannnnn  updated a model 1 day ago

hewittlab/monitor_cot_for_hack-reward_description-length-explicit_intent-nemotron-3-nano-30b-a3b

LorenaYannnnn  updated a collection 1 day ago

Reward Misspecification Degrades CoT Monitorability

View all activity

Team members 2