MME-VideoOCR: Evaluating OCR-Based Capabilities of Multimodal LLMs in Video Scenarios | Read Paper on Bytez