Salvato in:
| Autori principali: | , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2510.04373 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910028503449600 |
|---|---|
| author | Nekoei, Hadi Jaiswal, Aman Bechard, Patrice Shliazhko, Oleh Ayala, Orlando Marquez Reymond, Mathieu Caccia, Massimo Drouin, Alexandre Chandar, Sarath Lacoste, Alexandre |
| author_facet | Nekoei, Hadi Jaiswal, Aman Bechard, Patrice Shliazhko, Oleh Ayala, Orlando Marquez Reymond, Mathieu Caccia, Massimo Drouin, Alexandre Chandar, Sarath Lacoste, Alexandre |
| contents | Large language model (LLM) agents perform well in sequential decision-making tasks, but improving them on unfamiliar domains often requires costly online interactions or fine-tuning on large expert datasets. These strategies are impractical for closed-source models and expensive for open-source ones, with risks of catastrophic forgetting. Offline trajectories offer reusable knowledge, yet demonstration-based methods struggle because raw traces are long, noisy, and tied to specific tasks. We present Just-in-time Episodic Feedback Hinter (JEF-Hinter), an agentic system that distills offline traces into compact, context-aware hints. A zooming mechanism highlights decisive steps in long trajectories, capturing both strategies and pitfalls. Unlike prior methods, JEF-Hinter leverages both successful and failed trajectories, extracting guidance even when only failure data is available, while supporting parallelized hint generation and benchmark-independent prompting. At inference, a retriever selects relevant hints for the current state, providing targeted guidance with transparency and traceability. Experiments on MiniWoB++, WorkArena-L1, and WebArena-Lite show that JEF-Hinter consistently outperforms strong baselines, including human- and document-based hints. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_04373 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | JEF-Hinter: Leveraging Offline Knowledge for Improving Web Agents Adaptation Nekoei, Hadi Jaiswal, Aman Bechard, Patrice Shliazhko, Oleh Ayala, Orlando Marquez Reymond, Mathieu Caccia, Massimo Drouin, Alexandre Chandar, Sarath Lacoste, Alexandre Artificial Intelligence Large language model (LLM) agents perform well in sequential decision-making tasks, but improving them on unfamiliar domains often requires costly online interactions or fine-tuning on large expert datasets. These strategies are impractical for closed-source models and expensive for open-source ones, with risks of catastrophic forgetting. Offline trajectories offer reusable knowledge, yet demonstration-based methods struggle because raw traces are long, noisy, and tied to specific tasks. We present Just-in-time Episodic Feedback Hinter (JEF-Hinter), an agentic system that distills offline traces into compact, context-aware hints. A zooming mechanism highlights decisive steps in long trajectories, capturing both strategies and pitfalls. Unlike prior methods, JEF-Hinter leverages both successful and failed trajectories, extracting guidance even when only failure data is available, while supporting parallelized hint generation and benchmark-independent prompting. At inference, a retriever selects relevant hints for the current state, providing targeted guidance with transparency and traceability. Experiments on MiniWoB++, WorkArena-L1, and WebArena-Lite show that JEF-Hinter consistently outperforms strong baselines, including human- and document-based hints. |
| title | JEF-Hinter: Leveraging Offline Knowledge for Improving Web Agents Adaptation |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2510.04373 |