Upload README.md with huggingface_hub

Upload app.py with huggingface_hub

Fix: use Qwen2_5_VLForConditionalGeneration, pin transformers>=4.46

Fix VL model loading: use AutoModelForVision2Seq with ignore_mismatched_sizes

Add torchvision dependency for qwen-vl-utils

Replace EasyOCR with Qwen2.5-VL-3B for math-aware image OCR

Fix LaTeX rendering: normalize delimiters for KaTeX, enable latex_delimiters on Markdown output

Launch Nalanda-7B JEE/NEET Question Solver

initial commit