TensorFlow Machine Learning Framework

TensorFlow is a free and open-source machine learning framework developed by the Google Brain team, released as open source in November 2015. Created by Martin Abadi, Ashish Agarwal, Paul Barham, and others at Google, TensorFlow is one of the most widely used ML frameworks in the world, powering applications at Google (Search, Translate, Photos) and countless organizations. Key features: computation graphs: TensorFlow computations are expressed as dataflow graphs where nodes represent operations and edges represent multidimensional data arrays (tensors). This enables automatic differentiation, distributed execution, and deployment to various hardware. Keras: since version 2.0, Keras is the official high-level API for TensorFlow, providing a user-friendly interface for building neural networks with sequential, functional, and subclassing APIs. Eager execution: since version 2.0, TensorFlow defaults to eager execution (imperative style), with graph mode available via tf.function for production performance. Hardware acceleration: supports CPU, GPU (CUDA for NVIDIA, ROCm for AMD), and TPU (Tensor Processing Unit). Automatic mixed precision for faster training. Model deployment: TensorFlow Serving (production serving), TensorFlow Lite (mobile/IoT), TensorFlow.js (browser), TFX (end-to-end platform). Distributed training: tf.distribute.Strategy for multi-GPU and multi-host training. Dataset API: tf.data for efficient input pipelines. tfhub: pre-trained models. C++ and Python. Apache-2.0.