J. Roger Zhao

Calvin has done an excellent job of separating the Mat and Linalg aliases from the Ndarray modules. I think perhaps one thing that could be improved is the positioning/naming of...

Hi @sremedios, sorry for this very late reply. I just run the same MNIST example in an Owl container, and the result looks normal. According to you log, the "unknown...

Yes. Many ndarray operations such as `transpose` are interfaced to the core C code. Currently most core operations support input of float and complex types, but not integer type.

Agree. The optimiser and shape interence tests in TensorFlow might be of help, such as [optimizer_cse_test.cc](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/graph/optimizer_cse_test.cc), [constant_folding_test.cc](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/common_runtime/constant_folding_test.cc), [shape_inference_test.cc](https://github.com/tensorflow/tensorflow/blob/master/tensorflow/core/framework/shape_inference_test.cc), etc.

If I understand correctly, the problem is to get the output of any layer in the network during the inferencing phase. I met similar issues before and have to manually...

Yes the nodes in `network.topo` has to be topologically sorted. Though I don't quite get if there is any ordering issue in `roots`. It's the inputs of the graph, so...

Close for now since the configuration has been updated.

Thanks a lot for this solution! Of course I also need to change the resolver.py file accordingly: ``` from . import caffe_pb2 self.caffepb = caffe_pb2 ```

Thank you all so much for the constructive feedback! Modularity is indeed very important. Like I mentioned in the announcement, currently the plan is to keep the holistic design of...

Thanks for pointing it out. The mentioned doc is not actively maintained now. The `algodiff` code is currently positioned at: https://github.com/owlbarn/owl/tree/main/src/base/algodiff .