用大模型自动生成网页导航数据，实现无人工标注的互联网规模训练。
InSTA: Towards Internet-Scale Training For Agents
- 用LLM自动标注15万网站任务，生成可执行轨迹。
- 自研代理成功率达56.9%，超过235倍大的模型。
- 适合想低成本训练高效网页代理的研究者与开发者。
当前训练网页导航智能体主要依赖人工标注热门网站的任务，但人力成本高且效率低。本文提出InSTA流水线，实现无需人工标注的互联网规模训练：第一阶段，用大语言模型（LLM）为15万网站标注智能体任务；第二阶段，由LLM代理执行任务并生成轨迹；第三阶段，由LLM判断轨迹是否成功。实验显示，该方法在识别有害内容上准确率达97%，成功轨迹判别准确率为82.6%。基于Qwen 3 1.7B的小型模型训练的代理，成功率达56.9%，优于235倍大的Qwen 3 235B模型和235倍大的Llama 4 Maverick，达到Gemini 2.5 Flash性能的94.7%。代码、模型与数据已开源。
原文摘要 · Abstract (English)
The predominant approach for training web navigation agents is to gather human demonstrations for a set of popular websites and hand-written tasks, but it is becoming clear that human data is an inefficient resource. We develop a pipeline to facilitate internet-scale training for agents without laborious human annotations. In the first stage, an LLM annotates 150k sites with agentic tasks. In the next stage, LLM agents complete tasks and produce trajectories. In the final stage, an LLM filters trajectories by judging their success. Language models are powerful data curation tools, identifying harmful content with an accuracy of 97%, judging successful trajectories with an accuracy of 82.6%, and producing effective data. We train agents based on Qwen 3 1.7B that are competitive with frontier LLMs as web agents, while being smaller and faster. Our top agent reaches a success rate of 56.9%, outperforming the data collection policy Qwen 3 235B, a 235 times larger Llama 4 Maverick, and reaching 94.7% of the performance of Gemini 2.5 Flash. We are releasing code, models and data at: https://data-for-agents.github.io.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。