GaiYu0

@wddabc Thank you very much for your suggestion! (I should have participated in discussion earlier) The module `Parallel` has not been implemented because in the framework of `model_builder` it is...

Resolving parameter sharing might be easier. I am implementing a initializer that enables user to index initialization configuration by parameter. For example, `initializer[convolution0.weight].std = 0.1` specifies that the weight of...

Also, `model_builder` only supports static graph composition for the time being, i.e. statically declare a graph and use a function to convert the graph to a model. Dry-running is going...