Preprint2025
olmOCR
Unknown
olmOCR is an open-source Python toolkit that converts PDFs into linearized plain text while preserving structured content using a fine-tuned 7B VLM.
0Feb 1, 2025Fine Tuning
AI Research
AI & ML Research Papers
A comprehensive index of artificial intelligence and machine-learning research with AI-generated summaries, citation metrics, and direct links to papers and code.
All Papers
9 papersPreprint2023
Nougat
Unknown
Nougat uses a Visual Transformer to convert scientific document images into markup language via OCR.
0Aug 1, 2023Transformers
Preprint2021
Donut
Unknown
Donut is an OCR-free encoder-decoder Transformer that directly processes images and prompts to generate text, eliminating the need for optical character recognition.
0Nov 1, 2021TransformersPrompting
Preprint2024
H2OVL-Mississippi
Unknown
H2OVL-Mississippi introduces small, efficient vision-language models optimized for OCR and document analysis while maintaining strong general VLM performance.
0Oct 1, 2024BenchmarksComputer Vision
arXiv
Preprint2024
Idefics2
Unknown
Idefics2 improves upon Idefics1 with enhanced OCR, simplified architecture, and better pre-trained backbones using open datasets and task-oriented fine-tuning.
0Apr 1, 2024Fine Tuning
Preprint2024
LLaVA 1.6
Unknown
LLaVA 1.6 enhances multimodal AI with dynamic high-resolution processing, improved OCR, and scalable LLM integration.
0Jan 1, 2024Reasoning
Preprint2022
BLOOM
Unknown
BLOOM is a 176B-parameter open-access decoder-only transformer collaboratively developed to democratize large language model technology.
0Nov 1, 2022Transformers
Preprint2024
Neural operators for accelerating scientific simulations and design
Unknown
Neural operators use deep learning to accelerate scientific simulations and democratize access to computational science.
0Jan 1, 2024
Preprint2026
Persian Pixel: A large-scale synthetic OCR dataset for Persian language
Pouria Mahdi, Haq Nawaz Malik
Persian Pixel is a large-scale synthetic OCR dataset with over 343,000 image-text pairs for Persian, generated via a rendering pipeline that models script complexities and applies stochastic degradations to bridge the synthetic-to-real gap.
0Jul 22, 2026TransformersFine Tuning
arXiv