Meet Dots.iferCC: The original version of the original version of 1.7B reaches the SOTA functionality in the Parsing of multilingual documents
dots.um The transformer's open source model designed for multi-language structure makes all the adoption and recognition of content within one of the individual properties, to support more than 100 languages and randomized books.
Architecture
- The combined model: DOTS.O oral oral include the acquisition of a structure and recognition of content in neural-based network based on transformer. This ends the hardware of different discovery and OCR pipes, which allows users to change jobs by correcting the installation promotion.
- Parameters: The model contains 1.7 billion parameters, measuring a computer performance through practical performance.
- Input flexibility: Input can be photo files or PDF documents. The model includes the installation options (such as FITZ_PROPROPSS) to expand the quality in low-settlement or files with many pages.
Skills
- Many languages: The dots.Cuci are trained as Pataset Aspasets Asses more than 100 languages, including the world's largest languages and common documents, which indicates the broader range of languages.
- Deleting content: The model removes the transparent text, tabular data, mathematical formulas (LATEX), and maintain a learning order within the text. Formatting Formats include JSON, Markdown, and HTML, according to the type of content and content type.
- Safeguarded the building: DOTS.ICON FINANCE DOCTS A Document, including table boundaries, formula regions, and images, to ensure that issued data remains reliable in the original document.
Benchmark's operation
The dots of the Legislatures are facing modern AI, which has unshed features below:
| Coat | dots.um | Gemino2.5-pro |
|---|---|---|
| The accuracy of table ts | 88.6% | 85.8% |
| Distance to edit text | 0.032 | 0.055 |
- Tables: OutperFforms Gemino2.5-Pro In Tary Parsing Near.
- Text: Indicates the low range of text editing (which indicates the highest clarification).
- Formules and structure: Parallels or exceeds leading models in formula's recognition and the redesignment of the document.
Shipment and Compilation
- Open Source: It is issued under the MIT License, with the Source, the documents, and previously trained models are found in GitTub. Recompo the PIP Input Instructions, Colla, and DOCKER-based submission.
- API and Scriping: Supports the configuration of flexible tasks with instant templates. The model can be used in conjunction or default pipes for batch pipes.
- Exit Formats: Extracted results are provided in a systematic JVS system, with marking and HTML options where appropriate. Visible documents enables examination of properties found.
Store
Dots.Oour provides a solution to higher accuracy, relating to a multilingual document for receiving formation and recognition of content, open. It is especially suitable for situations that require strong analysis, the language-acnostic processing and discharge of organized information in the pressed or prodpressed areas.
Look GitHub page. Feel free to look our GITHUB page for tutorials, codes and letters of writing. Also, feel free to follow it Sane and don't forget to join ours 100K + ml subreddit Then sign up for Our newspaper.
Michal Sutter is a Master of Science for Science in Data Science from the University of Padova. On the basis of a solid mathematical, machine-study, and data engineering, Excerels in transforming complex information from effective access.
