Antgo is a machine learning experiment manage platform, which has been integrated deeply with MLTalker. Antgo provides a one-stop model development, deployment, analyze, auto-optimize and manage environment.
(RECOMMENDED) use docker
install from pip
pip install antgo
install from source
- git clone https://github.com/jianzfb/antgo.git
- cd antgo
- pip install -r requirements.txt
- python setup.py build_ext install
Register in MLTalker.
All user experiment records would be managed by MLTalker in user's personal page.
1.step create mvp code(cifar10 classification task)
antgo create mvp --name=cifar10
2.step start training process
python3 ./cifar10/main.py --exp=cifar10 --gpu-id=0 --process=train
3.step check training log
in ./output/cifar10/output/checkpoint
4.step export onnx model
python3 ./cifar10/main.py --exp=cifar10 --checkpoint=./output/cifar10/output/checkpoint/epoch_1500.pth --process=export