Hi friends,

I am studying the 5th chapter “pet breeds”, I followed the book and build a **learn** based on resnet34.

the prediction by this **learn** give us a result with shape of (1,37) for each item. i am wondering how it gots the result. like having a weight of (n,37) for the last layer or what? I think if i can read the structure of this **learn**, it will be very helpful. can anybody help to know how to read structure of netlayer? thank you very much.

Did you mean “learn.summary()”?

This gives you something akin to this one

Where my net is just for regression on one value.

Since yours is a classifier for 37 different breeds, I would expect your layer to be something akin to

Softmax […]x37

Alternatively you can also go the learn.model() way which yields something like

for me, so in your case something like

Softmax(in_features=[…], out_features=37)

thank you very much.

the first way ‘learn.summary()’ works for me. it is clear what the model is doing.

however, the second way ‘learn.model()’ raise an error that this.

try just `learn.model`

no parenthesis

2 Likes