Wu Shuo has learned that Tether CEO Paolo Ardoino disclosed a new version of QVAC Fabric from the Tether AI team, claiming to launch the world's first cross-platform BitNet LoRA framework, based on Vulkan and Apple Metal backend supporting AMD, Intel, Apple, and mobile GPUs, enabling BitNet's LoRA fine-tuning and inference to run across GPU vendors and operating systems. He stated that the highest 3.8 billion parameter model fine-tuning has been demonstrated on devices such as Pixel 9, S25, and iPhone 16, and can fine-tune 13 billion parameter models on iPhone 16; mobile GPU inference is 2–11 times faster than CPU, with memory usage reduced by up to 90%.

View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pin