Search
Google recently launched the JavaScript library LiteRT.js, aimed at enhancing the processing speed of machine learning and AI workloads on the browser side. This library introduces the mobile-optimized LiteRT runtime to the web platform, enabling AI models to run efficiently in local browsers without relying on server processing. LiteRT.js supports hardware-accelerated inference within browsers, leveraging WebGPU and WebNN APIs for GPU and specialized hardware optimization. It also offers multi-framework compatibility, allowing direct compilation of PyTorch, JAX, or TensorFlow models. Additionally, it supports seamless integration with the TensorFlow.js architecture, simplifying the model conversion and deployment process, providing a high-performance, low-latency solution for AI applications on the web.
-
C114 Communication Network -
Communication Home
