ethereon/netscope

Input Layer fail to parse shape for multi blobs

leoshine opened this issue · 1 comments

For Input layer, if only one blob is fine, but failed when output multi blobs.
For example:
layer {
name: "Input"
type: "Input"
top: "label"
top: "data"
input_param {
shape {
dim: 10
dim: 1
}
shape {
dim: 10
dim: 3
dim: 224
dim: 224
}
}
}

And this result in:

image

rlnx commented

@leoshine, how should these shapes be interpreted? Does every shape correspond to the particular top blob in the order they are enumerated?