conftrace_

negative side effect

3 papers

Explore in graph

Also known as

NSE

Co-occurring keywords · papers mentioning both

markov decision process (2) policy constraint (1) policy refinement (1) large language model (1) output refinement (1) lexicographic reward (1) constrained planning (1) in-context reward hacking (1) constrained mdp (1) reward shaping (1)

Papers

Feedback Loops With Language Models Drive In-Context Reward Hacking ICML 2024 Planning and Learning for Non-markovian Negative Side Effects Using Finite State Controllers AAAI 2023 A Multi-Objective Approach to Mitigate Negative Side Effects IJCAI 2020