QWEN-DRIVE

Summary from the TLDR AI newsletter

Qwen-Drive is a project that aims to create a Vision-Language Foundation model for autonomous driving. The project uses a staged training strategy that integrates perception, language, and planning objectives. The model achieves specialized driving competence while retaining broad visual understanding and instruction-following capabilities. A GPU with 24 GB+ of memory is recommended.