v1.6.2

datalab-to/markerv1.6.2Mar 26, 2025by VikParuchuri

AI Summary

Adds OpenAI-like LLM service support and fixes PDF extraction to include form fields.

Key Highlights

Added OpenAI-like LLM service
Flatten PDF before imaging to pull in form fields
Improved quality of LLM form and math block results

New Features

OpenAI LLM service
Form field extraction improvements
Improved LLM form and math block results

Full Release Notes

## Bugfixes

- Flatten before imaging PDF, so form fields get pulled in if available
- Add OpenAI LLM service
- Improve quality of LLM form and math block results
- Fix bug around misidentification of some document types
- Benchmark updates

## What's Changed
* Add Page Flattening when extracting PDF Provider Images by @tarun-menta in https://github.com/VikParuchuri/marker/pull/627
* Add OpenAI-like service by @denisshepelin in https://github.com/VikParuchuri/marker/pull/592
* Updates + Fixes by @tarun-menta in https://github.com/VikParuchuri/marker/pull/631
* Dev by @tarun-menta in https://github.com/VikParuchuri/marker/pull/632

## New Contributors
* @denisshepelin made their first contribution in https://github.com/VikParuchuri/marker/pull/592

**Full Changelog**: https://github.com/VikParuchuri/marker/compare/v1.6.1...v1.6.2