Tag
This paper defines embodied operators as reusable functional modules for embodied intelligence pipelines, presents a taxonomy covering five major categories, and proposes a multi-dimensional benchmark framework for evaluating their deployability and composability.
A new contrastive ablation operator called apostate is introduced that reduces model refusal from 96% to 5% while preserving harmless behavior with only 0.081 KL divergence, tested on Granite 3.3-8B.
OpenAI announced an upgrade to Operator, replacing the GPT-4o-based Computer Using Agent model with an o3-based version that maintains safety guardrails while improving coding capabilities for web automation tasks.