multimodal vlm projects

drex 062225 exp (document retrieval and extraction expert) is a specialized fine-tuned version of docscopeocr, optimized for document retrieval, content extraction. built on top of the qwen2.5-vl.

fine-tuned version of qwen2-vl-2b, tailored for tasks that involve messy handwriting recognition, optical character recognition (ocr), and latex formatting.

virex 062225 exp (video information retrieval and extraction expert) model is a fine-tuned version of qwen2.5-vl, optimized for advanced video understanding, sense of reasoning.

a fine-tuned version of qwen2-vl-2b, tailored for tasks involving improved optical character recognition (ocr) and latex formatting compared to the previous version.

post-trained version of qwen2.5-vl-7b-instruct, optimized for document-level optical character recognition (ocr), and accurate image-to-text conversion with mathematical latex formatting.

multimodal ocr space for experimenting with the qwen2 vl ocr 2b, tailored for tasks involving improved optical character recognition (ocr) and latex formatting compared to the previous version. runs on nvidia h200.

fine-tuned version'3 of qwen2-vl-2b-instruct, specifically optimized for messy handwriting recognition, optical character recognition (ocr) and math problem solving with latex formatting.

post-trained version of qwen2-vl-7b, optimized for document-level optical character recognition (ocr), and accurate image-to-text conversion with mathematical latex formatting.