TensorFlow Variables are in-memory buffers containing tensors. Learn how to use them to hold and update model parameters during training.
A step-by-step walk through of the details of using TensorFlow infrastructure to train models at scale, using MNIST handwritten digit recognition as a toy example.
TensorBoard is a useful tool for visualizing the training and evaluation of your model(s). This tutorial describes how to build and run TensorBoard as well as how to add Summary ops to automatically output data to the Events files that TensorBoard uses for display.
This tutorial describes how to use the graph visualizer in TensorBoard to help you understand the dataflow graph and debug it.
This tutorial describes the three main methods of getting data into your TensorFlow program: Feeding, Reading and Preloading.
This tutorial describes how to execute TensorFlow programs using a cluster of TensorFlow servers.
This tutorial describes the various constructs implemented by TensorFlow to facilitate asynchronous and concurrent training.
TensorFlow already has a large suite of node operations from which you can compose in your graph, but here are the details of how to add you own custom Op.
TensorFlow Style Guide is set of style decisions that both developers and users of TensorFlow should follow to increase the readability of their code, reduce the number of errors, and promote consistency.
TensorFlow's documentation is largely generated from its source code. Here is an introduction to the formats we use, a style guide, and instructions on how to build updated documentation from the source.
If you have a sizable custom data set, you may want to consider extending TensorFlow to read your data directly in it‘s native format. Here’s how.
This tutorial describes how to construct and execute models on GPU(s).
When deploying large models on multiple GPUs, or when unrolling complex LSTMs or RNNs, it is often necessary to access the same Variable objects from different locations in the model construction code.
The “Variable Scope” mechanism is designed to facilitate that.
If you‘re developing a tool to load, analyze, or manipulate TensorFlow model files, it’s useful to understand a bit about the format in which they're stored. This guide covers the details of the saved model format.
Training a full object recognition model like Inception takes a long time and a lot of images. This example shows how to use the technique of transfer learning to retrain just the final layer of a fully-trained model to recognize new categories of objects, which is a lot faster and easier than completely retraining a new model.
This tutorial describes how to export everything pertaining to a running model and import it later for various purposes.
This guide shows how you can convert a float model into one using eight-bit quantized parameters and calculations. It also describes how the quantization process works under the hood.