研究GOT-OCR-项目落地加速，不限语言

Local-first AI platform for education, research, and complex problem-solving. 8-model pipeline (Qwen, IndicTrans2, Whisper V3, GOT-OCR, BGE-M3) with hardware-adaptive inference (MLX/CoreML/MPS/CUDA), RAG search, multi-tier caching. Handles math, coding, academic research — multilingual with 10 Indian languages. Safe without being restricted.

No description available

Implementation and fine-tuning of GOT-OCR 2.0 for Indic language text recognition. The repository provides dataset preprocessing pipelines, modular training and inference scripts, and sample outputs. Designed for high-accuracy OCR on complex, curved, and multi-script text in Hindi and other Indic languages.