HERAKLES couples a language-model planner to a small, continually retrained skill executor and outperforms three baselines on the 17-goal Crafter benchmark, scaling better to reworded and repeated goals.
Adaptations:V(s 1;z) is not pretrained in our set up but learned during training on collected trajectories
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
HERAKLES: Hierarchical Skill Compilation for Open-ended LLM Agents
HERAKLES couples a language-model planner to a small, continually retrained skill executor and outperforms three baselines on the 17-goal Crafter benchmark, scaling better to reworded and repeated goals.