做数据处理的朋友可以看看,OpenRouter 这个 API 用一句话描述字段就能从 PDF、图片里抽结构化数据,不用自己写 prompt 和解析逻辑,还能随时换底层模型,适合处理发票、邮件这类批量任务。
#structured data
Zero-training inference on tables removes the usual feature-engineering and fine-tuning cycle, which could make tabular ML accessible to teams without dedicated data scientists. Developers working with single tables or multi-table relational schemas can plug the library into existing PyTorch pipelines and get predictions from pretrained models directly.
For developers working with structured data, LimiX-2 offers a new open-source option that competes with other models in this specialized area.
清华和稳准联合发布了LimiX-2,这个400M参数的模型专门处理结构化数据,在相关国际评测中表现很突出。
This is a new model that outperforms current dataset-specific models and tabular foundation models on TabArena, TALENT, and BCCO benchmarks.
Xiaomi-TabLDM reduces training time by 82% compared to TabFM while maintaining top performance, offering significant efficiency gains for data-intensive industries in ASEAN.
Available now in Claude Connectors Directory, this tool solves document parsing issues that cause hallucinations when feeding complex files directly to Claude.
LlamaParse's spreadsheet extraction offers a cost-effective solution for processing complex financial documents compared to manual methods or OCR tools.
RepairFormer用Transformer自动修复损坏的JSON等文件,保留内容,修复率88%,速度比现有方法快5倍。
LlamaParse 专治合同 PDF 提取难题,能自动识别支付条款、续约条件等关键字段,比传统 OCR 更智能。
LlamaIndex新出Extract,能把杂乱的合同PDF自动变成结构化数据,解析条款和修订不再头疼,企业合同管理直接省下人工翻找的功夫。
a16z 点出了企业软件的下一个战场——语音数据,做 SaaS 或企业工具的团队值得关注这个趋势,提前布局语音分析能力。
企业 AI 团队终于有了一个开箱即用的文档处理框架,做合同、发票、报告等非结构化数据提取的开发者可以直接上手,拖拽式设计降低了门槛,值得一试。