RolmOCR is an open-source document OCR model developed by Reducto AI as a drop-in alternative to olmOCR that delivers faster performance with reduced memory usage. Built on Qwen2.5-VL-7B-Instruct and fine-tuned on the full olmOCR-mix dataset, this model excels at extracting plain text from complex documents including PDFs, images, and various document formats. RolmOCR features streamlined processing by eliminating metadata inputs, enhanced robustness through rotated training data, and maintains high accuracy while significantly reducing processing time and VRAM requirements compared to its predecessors.
Fine-tuningDocs | Rolm OCR can be customized with your data to improve responses. Fireworks uses LoRA to efficiently train and deploy your personalized model |
On-demand DeploymentDocs | On-demand deployments allow you to use Rolm OCR on dedicated GPUs with Fireworks' high-performance serving stack with high reliability and no rate limits. |
