#pre-trained model

This project explores the LCCC and MMChat datasets and pretrained models using the Chinese GPT architecture. It provides code for pretraining and fine-tuning with HuggingFace's Transformers library, facilitating robust Chinese dialogue generation. Updates and resources are included to support sentiment analysis and natural language generation.

CodeGeeX2 is a multilingual code generation model based on the ChatGLM2 framework, achieving enhanced performance with 6 billion parameters. It supports languages including Python, C++, and Java, offering improvements over its predecessor. Key features include faster inference, support for up to 8192 sequence length, and deployment requiring only 6GB GPU memory. The updated CodeGeeX plugin facilitates over 100 languages and offers contextual and cross-file completion. The model weights are available for academic research, with an option for commercial use.

This project provides an efficient solution for real-time super resolution of low-resolution videos utilizing the SR-GAN inspired architecture and pixel shuffle technique. Capable of processing videos up to 720p at 30fps on MPS devices, it offers a pretrained model for image inference and detailed instructions for custom training with editable CLI configurations. The repository welcomes contributions for model enhancement and feature expansion, facilitating advancements in video quality through established machine learning methodologies.

CharacterGLM-6B advances conversational AI through realistic character attributes and behaviors, emphasizing consistency, human-likeness, and engagement. Developed by Lingxin Intelligence and Tsinghua University's CoAI Lab, this model utilizes ChatGLM2 for creating intricate AI personalities. It's intended for academic research to explore complex dialogues and ethical AI development. Discover its capabilities via demos and comprehensive technical documentation.

Explore semantic segmentation with this project using the Unicom model pre-trained on 400 million images. Achieve efficient results with just four training images. Start quickly with 'train_one_gpu.py' or optimize performance with 'train_multi_gpus.py' for multi-GPU support. Follow setup guides for seamless configuration. Connect via QQ group 679897018 for discussions.

Segment Any 3D Gaussians (SAGA) is a platform for 3D object segmentation using Gaussian techniques. It provides tools like a GUI and Jupyter notebooks for both manual and automated segmentation, utilizing pre-trained models and datasets such as 360_v2 and LERF. The platform supports rendering and visualization with clustering methods like HDBSCAN.

Neural-Cherche is a versatile library designed to fine-tune and utilize neural search models like Splade, ColBERT, and SparseEmbed in both offline and online contexts. Compatible with CPU, GPU, and MPS devices, it ensures seamless performance while storing computed embeddings to minimize redundancy. Its straightforward integration with existing data workflows offers a customized search experience. Comprehensive documentation facilitates deployment and training, making it an efficient choice for neural search projects.