mattn/go-tflite

Go binding for TensorFlow Lite

What it solves

It provides a way for developers to use TensorFlow Lite models within the Go programming language, filling the gap for those who want to run lightweight machine learning models in Go applications without needing to write C++ or Python.

How it works

The project acts as a Go binding for the TensorFlow Lite C API. It allows Go programs to load .tflite model files, configure interpreter options, allocate tensors, and perform inference by passing input data to the model and retrieving the output results.

Who it’s for

Go developers who need to integrate pre-trained TensorFlow Lite models into their software, including those targeting edge devices or using hardware accelerators like the Edge TPU.

Highlights

  • TFLite Integration: Direct bindings to the TensorFlow Lite C API for running inference in Go.
  • Edge TPU Support: Compatible with the Edge TPU delegate for hardware-accelerated machine learning.
  • Browser-based Testing: Provides a Binder link to test the library with a Jupyter notebook (e.g., for an iris classifier).

Related

  • Project
    Neargye/hello_tf_c_api

    A cross-platform collection of C++ examples and utilities for using the TensorFlow C API to run model inference in native applications.

  • Project
    tensorflow/tflite-micro

    A port of TensorFlow Lite designed to run machine learning models on microcontrollers and DSPs with limited memory.

  • Project
    go-skynet/go-llama.cpp

    Go bindings for llama.cpp that enable high-performance local LLM inference within Go applications using the GGUF format.

  • Project
    tensorflow/tensorflow

    An end-to-end open source platform for machine learning that provides a flexible ecosystem of tools and libraries for building and deploying ML-powered applications.

  • Project
    tensorflow/java

    TensorFlow‑Java provides Maven‑publishable Java/JVM bindings for TensorFlow, including low‑level JNI wrappers (`tensorflow‑core`), a high‑level neural‑network API (`tensorflow‑framework`), and an independent ndarray library. It supports Linux (x86_64, arm64) and macOS (Apple Silicon) binaries, with optional GPU builds, and integrates with standard Java build tools.