
TorchNeuron is designed to run PyTorch code on Amazon’s Trainium chips with one configuration change and no rewriting.
Amazon is introducing TorchNeuron, software that lets developers run PyTorch code on its Trainium AI chips with one configuration change and no rewriting, according to the company’s Annapurna Labs chip unit.
Amazon plans to announce the tool at a developer conference in Silicon Valley on Monday. It said TorchNeuron will be released publicly as a free, open-source offering.
The move seeks to reduce Nvidia’s software advantage. Nvidia’s Cuda software helps PyTorch code run on its chips, while other cloud companies are also working to attract developers with improved PyTorch tools.
Annapurna Labs software vice president Colin Brace said more than 30 early customers across robotics, world models and voice AI are using TorchNeuron. He said switching to Amazon’s chips can bring steep cost savings.
This article was produced with the help of AI technology. Based on reporting by the outlets listed above.