Nvidia and Apple Show Off “ReDrafter,” a New AI Text Generation Model
Nvidia and Apple collaborated to create “ReDrafter,” an innovative AI method accelerating text generation in Large Language Models (LLMs).
Apple and Nvidia have worked together to create a cool new AI method called “ReDrafter,” which is meant to speed up the process of making text in Large Language Models (LLMs). This partnership brings together Apple’s experience in custom silicon and Nvidia’s top-of-the-line GPU technology to help advance AI.
Recurrent Drafter refers to the shortened term known as ‘ReDrafter’ that helps to enlighten text creation with the help of beam search and tree attention. Apple has always used its own silicon in AI, but to add this way to the TensorRT-LLM framework, it collaborated with Nvidia to optimize the GPUs for LLMs.
To extend compatibility with ReDrafter’s algorithms, Nvidia included additional operators into TensorRT-LLM. The benchmark tests using the app proved that it helps in creating tokens 2.7 faster, the delay was reduced, and the GPU power consumption was also reduced. What this implies is that AI based apps are now more efficient and less have an impact on the environment.
Apple and Nvidia have been rivals in the past, but this relationship shows that they both want to make AI better. It doesn’t look like there will be a long-term partnership, but the success of ReDrafter points to possible short-term partnerships that could shape the future of creative AI.
