SCALAR: Learning and Composing Skills through LLM Guided Symbolic Planning and Deep RL Grounding
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911501031309312 |
|---|---|
| author | Zabounidis, Renos Wu, Yue Stepputtis, Simon Kim, Woojun Li, Yuanzhi Mitchell, Tom Sycara, Katia |
| author_facet | Zabounidis, Renos Wu, Yue Stepputtis, Simon Kim, Woojun Li, Yuanzhi Mitchell, Tom Sycara, Katia |
| contents | LM-based agents excel when given high-level action APIs but struggle to ground language into low-level control. Prior work has LLMs generate skills or reward functions for RL, but these one-shot approaches lack feedback to correct specification errors. We introduce SCALAR, a bidirectional framework coupling LLM planning with RL through a learned skill library. The LLM proposes skills with preconditions and effects; RL trains policies for each skill and feeds back execution results to iteratively refine specifications, improving robustness to initial errors. Pivotal Trajectory Analysis corrects LLM priors by analyzing RL trajectories; Frontier Checkpointing optionally saves environment states at skill boundaries to improve sample efficiency. On Craftax, SCALAR achieves 88.2% diamond collection, a 1.9x improvement over the best baseline, and reaches the Gnomish Mines 9.1% of the time where prior methods fail entirely. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_09036 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SCALAR: Learning and Composing Skills through LLM Guided Symbolic Planning and Deep RL Grounding Zabounidis, Renos Wu, Yue Stepputtis, Simon Kim, Woojun Li, Yuanzhi Mitchell, Tom Sycara, Katia Machine Learning LM-based agents excel when given high-level action APIs but struggle to ground language into low-level control. Prior work has LLMs generate skills or reward functions for RL, but these one-shot approaches lack feedback to correct specification errors. We introduce SCALAR, a bidirectional framework coupling LLM planning with RL through a learned skill library. The LLM proposes skills with preconditions and effects; RL trains policies for each skill and feeds back execution results to iteratively refine specifications, improving robustness to initial errors. Pivotal Trajectory Analysis corrects LLM priors by analyzing RL trajectories; Frontier Checkpointing optionally saves environment states at skill boundaries to improve sample efficiency. On Craftax, SCALAR achieves 88.2% diamond collection, a 1.9x improvement over the best baseline, and reaches the Gnomish Mines 9.1% of the time where prior methods fail entirely. |
| title | SCALAR: Learning and Composing Skills through LLM Guided Symbolic Planning and Deep RL Grounding |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2603.09036 |