T3Mermaid to ASCII art (mermaid-ascii)T3Kimi K3, and what we can still learn from the pelican benchmarkT3Firefox in WebAssemblyT3Spot birds not golfT3[AINews] Kimi K3 2.8T-A50B: the largest open model ever released; Opus 4.8-class at Sonnet 5 pricingT1From physical surfaces to human-centric heat stress: LST and UTCI heat mapping reveals nonlinear effects of urban morphologyT1DualHNIE: Dual-Channel Hypergraph Learning for Node Importance Estimation in Heterogeneous Knowledge GraphsT1GenTL: A General Transfer Learning Model for Building Thermal DynamicsT1A short review on the maximum clique problem algorithms with classical, AI, and quantum methodsT1Man, Machine, and Masterpiece: Artistic Ownership in the AI EraT1HABIB_TAZ at SemEval-2026 Task 11: Disentangling Formal Logic from Content via Synthetic Training and Multi-Objective OptimizationT1How Well Does AI-Generated Feedback Work? Intrinsic and Extrinsic Evaluation across more than 20,000 EFL Essay DraftsT3Mermaid to ASCII art (mermaid-ascii)T3Kimi K3, and what we can still learn from the pelican benchmarkT3Firefox in WebAssemblyT3Spot birds not golfT3[AINews] Kimi K3 2.8T-A50B: the largest open model ever released; Opus 4.8-class at Sonnet 5 pricingT1From physical surfaces to human-centric heat stress: LST and UTCI heat mapping reveals nonlinear effects of urban morphologyT1DualHNIE: Dual-Channel Hypergraph Learning for Node Importance Estimation in Heterogeneous Knowledge GraphsT1GenTL: A General Transfer Learning Model for Building Thermal DynamicsT1A short review on the maximum clique problem algorithms with classical, AI, and quantum methodsT1Man, Machine, and Masterpiece: Artistic Ownership in the AI EraT1HABIB_TAZ at SemEval-2026 Task 11: Disentangling Formal Logic from Content via Synthetic Training and Multi-Objective OptimizationT1How Well Does AI-Generated Feedback Work? Intrinsic and Extrinsic Evaluation across more than 20,000 EFL Essay Drafts

Reward Hacking in Reinforcement Learning

Named practitioner synthesis

#233

T3digested

Canonical URL: https://lilianweng.github.io/posts/2024-11-28-reward-hacking/
Topic: Practitioner Notes
First seen: 2026-07-16 19:08:00
Last seen: 2026-07-16 19:08:00

Source raw items (1)

Blog / Newsletter2026-07-16 19:07:22
Reward Hacking in Reinforcement Learning
<p>Reward hacking occurs when a <a href="https://lilianweng.github.io/(https:/lilianweng.github.io/posts/2018-02-19-rl-overview/)">reinforcement learning (RL)</a> agent <a href="https://lilianweng.github.io/posts/2018-01-23-multi-armed-bandit/#exploitation-vs-exploration">exploits</a> flaws or ambiguities in the reward function to achieve high rewards, without genuinely learning or completing the intended task. Reward hacking exists because RL environments are often imperfect, and it is fundamentally challenging to accurately specify a reward function.</p> <p>With the rise of <a href="https://lilianweng.github.io/posts/2019-01-31-lm/">language models</a> generalizing to a broad spectrum of tasks and RLHF becomes a de facto method for alignment training, reward hacking in RL training of language models has become a critical practical challenge. Instances where the model learns to modify unit tests to pass coding tasks, or where responses contain biases that mimic a user’s preferen