Ant-v-4

'Ant-v4' is a benchmark in reinforcement learning that evaluates the performance of algorithms on continuous control tasks, specifically focusing on the locomotion of an ant-like agent.

Papers using Ant-v-4 (2)

CAMEL: Continuous Action Masking Enabled by Large Language Models for Reinforcement Learning2025 · 1 citesPrefPoE: Advantage-Guided Preference Fusion for Learning Where to Explore2025