This repository offers an open-source Chinese NLP framework covering the full ML lifecycle: data processing, model training, fine-tuning, and inference. It supports multiple foundation models including GPT-2, CLIP, LLaMA, LLaMA2, ChatGLM-6B, and VisionEncoderDecoderModel for multi-modal tasks. The framework implements multi-GPU distributed training and provides utilities for vocabulary modification, enabling users to train and deploy large models even on consumer-grade hardware.