AWorld: Orchestrating the Training Recipe for Agentic AI
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914015030018048 |
|---|---|
| author | Yu, Chengyue Lu, Siyuan Zhuang, Chenyi Wang, Dong Wu, Qintong Li, Zongyue Gan, Runsheng Wang, Chunfeng Hou, Siqi Huang, Gaochi Yan, Wenlong Hong, Lifeng Xue, Aohui Wang, Yanfeng Gu, Jinjie Tsai, David Lin, Tao |
| author_facet | Yu, Chengyue Lu, Siyuan Zhuang, Chenyi Wang, Dong Wu, Qintong Li, Zongyue Gan, Runsheng Wang, Chunfeng Hou, Siqi Huang, Gaochi Yan, Wenlong Hong, Lifeng Xue, Aohui Wang, Yanfeng Gu, Jinjie Tsai, David Lin, Tao |
| contents | The learning from practice paradigm is crucial for developing capable Agentic AI systems, yet it is severely hampered by inefficient experience generation, a bottleneck especially pronounced in complex benchmarks like GAIA. To address this, we introduce AWorld, an open-source system engineered for large-scale agent-environment interaction. By distributing tasks across a cluster, AWorld accelerates experience collection by 14.6x compared to standard single-node, sequential execution. This critical speedup makes extensive reinforcement learning practical and scalable. Leveraging this capability, we trained a Qwen3-32B-based agent that achieves pass@1 accuracy of 32.23% on the GAIA test set, which surpasses GPT-4o (27.91%) and rivals DeepSeek-V3 (31.89%). Our open-source system and the resulting agent provide a practical blueprint for a complete agentic AI training pipeline, from efficient interaction to demonstrable model improvement. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2508_20404 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | AWorld: Orchestrating the Training Recipe for Agentic AI Yu, Chengyue Lu, Siyuan Zhuang, Chenyi Wang, Dong Wu, Qintong Li, Zongyue Gan, Runsheng Wang, Chunfeng Hou, Siqi Huang, Gaochi Yan, Wenlong Hong, Lifeng Xue, Aohui Wang, Yanfeng Gu, Jinjie Tsai, David Lin, Tao Artificial Intelligence The learning from practice paradigm is crucial for developing capable Agentic AI systems, yet it is severely hampered by inefficient experience generation, a bottleneck especially pronounced in complex benchmarks like GAIA. To address this, we introduce AWorld, an open-source system engineered for large-scale agent-environment interaction. By distributing tasks across a cluster, AWorld accelerates experience collection by 14.6x compared to standard single-node, sequential execution. This critical speedup makes extensive reinforcement learning practical and scalable. Leveraging this capability, we trained a Qwen3-32B-based agent that achieves pass@1 accuracy of 32.23% on the GAIA test set, which surpasses GPT-4o (27.91%) and rivals DeepSeek-V3 (31.89%). Our open-source system and the resulting agent provide a practical blueprint for a complete agentic AI training pipeline, from efficient interaction to demonstrable model improvement. |
| title | AWorld: Orchestrating the Training Recipe for Agentic AI |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2508.20404 |