HalfCheetah-v-5

'HalfCheetah-v-5' is a benchmark environment used to evaluate the performance and stability of reinforcement learning algorithms, specifically focusing on the challenges associated with training agents on sequentially correlated transitions.

Papers using HalfCheetah-v-5 (4)

Not All Transitions Matter: Evidence from PPO2026Visualizing Latent Phase Structures in Locomotion Policies: A Multi-Environment Study with Temporal Feature Extension2026Uncovering Latent Phase Structures and Branching Logic in Locomotion Policies: A Case Study on HalfCheetah2026Hindsight Preference Replay Improves Preference-Conditioned Multi-Objective Reinforcement Learning2026