mcikalmerdeka/vlms-ocr-playground

chore: fix application name inside the ui

chore: cleaning up sample data folder as they are included in the examples/ folder already

fix task dropdown initialization to match default model (PaddleOCR)

Fix GLM-OCR: Use git transformers version for AutoProcessor support

change default model to PaddleOCR-VL-1.5

add GPU requirement notices and graceful error handling for CPU deployment

fix DeepSeek-OCR-2 compatibility: pin versions and add flash attention fallback

fix PaddleOCR API integration - auth header and fileType

fix dropdown validation error on model switch

refactor Gradio interface to remove redundant theme specification and clean up Textbox components

update README to change color scheme and refine short description

add hf space yaml config in readme

implement Unified OCR Platform with multiple model support, including DeepSeek-OCR-2, GLM-OCR, and PaddleOCR-VL-1.5. Add main application entry point, configuration, and Gradio interface. Update README and requirements for new dependencies.

add example image files

add several sample images to try

update code for model loading

add deepseek ocr 2 code

initialize repo