Tether's QVAC division launched the world's first LoRA fine-tuning framework supporting Microsoft's BitNet architecture on March 17, significantly reducing memory and computational requirements for AI models and enabling localized training on mobile phones. The technology allows users to conduct personalized learning without uploading data, protecting privacy. Testing shows that modern smartphones can effectively fine-tune billion-parameter models, advancing the application potential of edge computing.
