ForgeCore is a free-to-use commercial binary tool designed to simplify training popular neural network models. It supports Windows and Linux, and includes optional Rust-based inference binary compilation. This project is open for public use. You can find the binary file from the release link
- Simple GUI workflow for training YOLO-style models
- Supports YOLOv8-ready dataset structure
- CPU or GPU training options
- Automatically generates a parameter file for customization
- Can compile an inference-ready binary using Rust
Prerequisites:
- Ensure Docker is installed and running.
- (No other dependencies required; the app is self-contained).
Windows 64bit / Linux (ubuntu24 tested)
Note: ARM64 (Apple Silicon/Raspberry Pi) is not supported yet.
| > Download and extract the binary file from the zip files. \
Run .exe to install the application on Windows.
For Linux install with [sudo dpkg -i dist/forgecore_*.deb].
Traning data formate \
| > more detials in ./example_yolo/Readme.md for yolo8 netowrk.
more detials in ./example_swinnet/Readme.md for swinnet segmentation mask.
| >Follow the UI steps:
Step 1 — Build
Select either GPU or CPU mode based on your system's capabilities.
Step 2 — Training
Select your dataset folder.
The trained model output will be automatically saved to the output directory.
A parameter file will also be generated—modify it as needed for your workflow.
Step 3 — Build Inference Binary (Optional)
If you need a final standalone inference binary:
Run the final build step
The completed inference binary will appear in the specified output folder
An example code is included showing how to load the generated binary and weight file for inference.
This is a personal project, not affiliated with any employer. Model updates or new features are delivered without a fixed schedule. There is no defined roadmap and no commitment toward enterprise-level capabilities—updates will come when possible.
This software is provided as a binary-only tool.
✔ Free for personal or commercial use
✔ Redistribution allowed (binary only, unmodified)
✘ Modification of the binary is not allowed
✘ Reverse engineering, decompiling, or disassembling is prohibited
✘ Selling or claiming ownership of the binary is not permitted
The source code is not public and remains fully owned by the author.