Open lgarithm opened 6 years ago
Now we have demonstrated the possibility of implementing well known models with crystalnet, by giving two examples: vgg16 and yolov2.
But we still don't have an example of training the model from scratch.
We need to choose a model that isn't too small, but also not too large, and build a training example for it.
@luomai @zsdonghao any suggestions?
resnet-50 is a good choice.
Now we have demonstrated the possibility of implementing well known models with crystalnet, by giving two examples: vgg16 and yolov2.
But we still don't have an example of training the model from scratch.
We need to choose a model that isn't too small, but also not too large, and build a training example for it.