Announcing TensorFlow 1.5

TensorFlow 1.5 has been released, introducing several new features aimed at enhancing user experience and performance. A notable addition is the Eager Execution preview, which allows for an imperative, define-by-run programming style, making TensorFlow operations executable directly in Python, thereby simplifying research and development processes. The update also includes a developer preview of TensorFlow Lite, designed for mobile and embedded devices, which enables low-latency execution of trained models on-device without the need for cloud processing. Additionally, TensorFlow 1.5 supports GPU acceleration updates with CUDA 9 and cuDNN 7 for improved performance on Windows and Linux. The documentation site has been updated to provide a more streamlined Getting Started flow, alongside various other enhancements such as updates to Accelerated Linear Algebra (XLA) and RunConfig.