OpenChat
Alignment Lab AI
Open Source
About
OpenChat is a family of open-source large language models (LLMs) acclaimed for their efficiency and innovative training techniques. These models employ C-RLFT (Conditioned Reinforcement Learning from Human Feedback) for fine-tuning, enabling them to learn from mixed-quality data without explicit preference labels. This approach allows OpenChat models to perform comparably to much larger closed-source models like ChatGPT. The project provides both model weights and API-compatible code, facilitating easy deployment. OpenChat includes various model versions, each designed with different base models and datasets, and is continuously updated with new features and improvements 145.