Hello,
I am trying to implement a model that is trained (in a simplified way) as
follows:
Given image1 & image2:
```
out1 = Net(image1)
out2 = Net(image2)
loss = loss_func(out1, out2)
```
However, I only want the back-propagation to be done for one of the images.
What would be the best way to do this?
Right now I am running into the following error when trying to just add one of
those passes under `with mx.autograd.record():`:
```
Error in operator node_1436_backward: [00:58:51]
src/imperative/./imperative_utils.h:774: Check failed:
g.GetAttr<size_t>("storage_type_num_unknown_nodes") == 0U (4 vs. 0)
```
---
[Visit Topic](https://discuss.mxnet.io/t/selective-backpropagation/6347/1) or
reply to this email to respond.
You are receiving this because you enabled mailing list mode.
To unsubscribe from these emails, [click
here](https://discuss.mxnet.io/email/unsubscribe/6ced57751305f3c4e1492f4bbc14f0c483823cd89bf44b9cd6154b7672a9c780).
