TFLite is a set of tools to help developers run TF models on devices with small binary sizes and low latency. TFLite consists of two main components: the TFLite interpreter (tf.lite.Interpreter) and the TFLite converter (tf.lite.TFLiteConverter). The TFLite interpreter is what actually runs the TFLite model on low-power devices, such as mobile phones, embedded Linux devices, and microcontrollers. The TFLite converter, on the other hand, is run on powerful devices that can be used to train the TF model, and it converts the trained TF model into an efficient form for the interpreter.
TFLite is designed to make it easy to perform machine learning on devices without sending any data over a network connection. This improves latency (since there is no data transfer over networks), more privacy (as no data will ever leave the device), and offline...