Infinity-Doc2-5M: Multilingual Document Parsing Training Data

Infinity-Doc2-5M is a training dataset containing 5 million samples for document parsing scenarios. It covers diverse document types including academic papers, research reports, and financial reports, and supports both Chinese and English languages. The dataset was created by infly and was last updated on the platform in May 2026.

Use Cases

• Train document layout recognition models based on the described multiple layout types.
• Develop multilingual information extraction systems based on the Chinese and English language support.
• Fine-tune models for parsing specific document formats like academic papers or financial reports.
• Benchmark document parsing algorithms on a large-scale, diverse corpus.

Strengths

• Contains 5 million samples, providing a large-scale resource for training.
• Covers a wide range of document types, including academic and financial reports.
• Supports both Chinese and English languages, enabling multilingual applications.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

infly

Freshness

Last updated 2026-05-12 02:17:45; freshness should be verified.