subgoal-driven-long-horizon-agents

Improve long-horizon task success via subgoal decomposition and dense milestone-based rewards, dramatically outperforming sparse-reward RL and standard baselines.

Independent GitHub snapshot · Automated classification · Original documentation governs