Practical overview
When it fits
Consider TensorFlow.NET when a machine-learning workflow needs to live in a C# or F# .NET application, or when adapting TensorFlow code into .NET. Review the branch-to-TensorFlow version mapping and compute-package requirements before choosing dependencies.
Who it is for
- C# and F# developers building machine-learning applications in .NET.
- Teams adapting TensorFlow-oriented model workflows from Python to .NET.
- .NET developers who want a Keras high-level interface through TensorFlow.Keras.
Documented uses
- Develop and train machine-learning models from C# or F#.
- Build model training and inference workflows within .NET applications.
- Use a Keras high-level interface in a .NET project.
- Select CPU or GPU compute support according to the documented operating-system options.
Requirements & limits
- The project provides a .NET Standard binding, with C# as its primary repository language and support described for C# and F#.
- TensorFlow.Keras is released as an independent package rather than being the same package as TensorFlow.NET.
- The README maps master and v0.100.x to TensorFlow 2.10, v0.6x to TensorFlow 2.6, and v0.15-tensorflow1.15 to TensorFlow 1.15.
- GPU packages listed for Windows and Linux require CUDA and cuDNN; the README lists CPU packages for Windows, Linux and macOS.
- The maintainers state that they are unavailable for new features and bug fixes at this time, while welcoming contributions and offering to review pull requests.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Install TensorFlow.NET and select compute support
Add the TensorFlow.NET library and the documented compute-support package for a supported operating system.
Before you start
- A .NET project using NuGet or the NuGet Package Manager Console.
- Choose a listed CPU package for Windows, Linux or macOS, or a listed GPU package for Windows or Linux.
- CUDA and cuDNN are required when choosing a GPU package.
Steps
Install the TensorFlow.NET package through NuGet.
Choose and install one compute-support package: the CPU package for Windows/Linux, the CPU package for macOS, or the GPU package for Windows/Linux.
Install TensorFlow.Keras as a separate package if the project needs the Keras interface.
Expected outcome
The project has the TensorFlow.NET package and a selected compute-support package; TensorFlow.Keras can be added separately.
Guide limitations
- The listed GPU packages require CUDA and cuDNN.
- The README does not provide a single compute package for every operating-system and hardware combination.
Compare options for Model inference and serving
These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.
| Project | Documented purpose | Platform | Access and requirements |
|---|---|---|---|
| Tensorflow.net | Develop and train machine-learning models from C# or F#. | Linux, macOS, Windows | |
| Hyperpose | Customize pose-estimation architectures and datasets. | Docker, Linux, macOS, Windows | |
| Multi Task Nlp | Define multiple tasks and their data and model settings in a YAML file. | Linux, macOS, Windows | The project is an open-source library under the Apache-2.0 license, distributed through its public GitHub repository. |
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 14 October 2026
- Original website title
- SciSharp/TensorFlow.NET · Official homepage title · retained in its source language
Related options
TensorFlowSharp — The repository contrasts it with TensorFlow.NET, describing TensorFlowSharp as providing TensorFlow's low-level C++ API and running models built using Python, while TensorFlow.NET
ML.NET — The repository says ML.NET uses TensorFlow.NET as one of its backends for training and inference.
Explore project →