tensorflow/docs
Docs
This repository is the official documentation for TensorFlow, a machine learning framework. It provides comprehensive guides, tutorials, and API references for building, training, and deploying machine learning models. The documentation covers the full lifecycle of machine learning projects, from constructing data pipelines and building neural networks with high-level APIs to customizing training loops and deploying trained models in production, on edge devices, or in browsers.
The documentation includes step-by-step tutorials for a range of tasks, including reinforcement learning, ranking models, and custom training loops. It also offers a dedicated guide for deploying TensorFlow models with TensorFlow Serving and Lite. The repository is complemented by community-maintained translations of the documentation into multiple languages.
Features
- Machine Learning Tutorials - Serves as the official documentation repository for the TensorFlow machine learning framework.
- Automatic Differentiation Engines - Provides the automatic differentiation engine that computes gradients for all neural network training.
- Distributed Training Accelerators - Spreads model training across multiple GPUs, machines, or specialized processors to speed computation.
- Unified Distribution Abstractions - Provides a unified API for distributing model training across multiple devices and machines with minimal code changes.
- Eager Execution Modes - Supports eager execution mode for immediate operation evaluation and imperative-style debugging.
- Parallelized Input Pipelines - Ships a dataset pipeline framework for building efficient, parallelized input pipelines from heterogeneous data sources.
- High-Level Model APIs - Provides high-level Keras APIs for building and training neural networks with minimal code.
- Machine Learning Frameworks - Serves as the official documentation for the TensorFlow machine learning framework.
- Model Serving & Deployment - Documents how to serve trained models in production environments for low-latency inference.
- Model Training - Documents building, training, and evaluating neural networks using Keras and custom loops.
- Neural Network Training Frameworks - Ships a complete neural network training framework with sequential and functional APIs.
- API References - Provides auto-generated API references for all TensorFlow Python APIs.
- Tutorials - Provides step-by-step tutorials for neural networks, reinforcement learning, and ranking models.
- Data Preprocessing Pipelines - Builds input pipelines to clean and transform data before feeding it into machine learning models.
- SavedModel Bundles - Exports trained models as language-agnostic SavedModel bundles containing graph, weights, and serving signatures.
- Graph-Based Computational Execution - Represents computations as directed graphs for lazy or eager execution across heterogeneous hardware.
- TPU-Optimized Kernels - Includes TPU-optimized kernel libraries with hardware-specific operations and memory layouts.
- Custom Training Loops - Documents how to implement custom training loops with gradient tapes and subclassing.
- End-to-End Training Pipelines - Documents how to orchestrate end-to-end ML pipelines for automated model training and deployment.
- In-Browser Model Execution - Documents how to train and execute machine learning models directly inside JavaScript environments.
- Edge AI Model Deployment - Provides a dedicated guide for deploying models on mobile and embedded hardware with TensorFlow Lite.
- Deployment Optimizations - Documents model optimization techniques including pruning and quantization for deployment.
- Model Training Monitoring - Provides TensorBoard integration for tracking and visualizing training metrics and model graphs.
- Reinforcement Learning Simulators - Provides tutorials for training decision-making agents using reinforcement learning algorithms.
- Deployment Guides - Provides dedicated guides for deploying models with TensorFlow Serving and Lite.
- Multi-Format Data Loading - Reads CSV, image, and text data sources into processing pipelines for efficient input handling.
- Tensor Operation Compilers - Compiles tensor operation subgraphs into optimized fused kernels for CPU, GPU, and TPU hardware.
- Model Training Metrics - Ships TensorBoard for logging and displaying training metrics in an interactive dashboard.
- Learning and Reference - TensorFlow documentation.
