Confused about the inconsistency with the Inception-ResNet-v2 code and the paper

[inflation]

Please let us know which model this issue is about (specify the top-level directory)

slim/inception-resnet-v2

---

The paper's "stem" structure is different with the the code.

And also, the paper indicates that Inception-v4 should share the same "stem" structure with Inception-resnet-v2. But the implementation of v4 is match with the paper.

So I don't know if it's intentional? Thanks for any explanation.

[ghost]

Please refer to https://research.googleblog.com/2016/08/improving-inception-and-image.html

They said "we are also releasing a pre-trained instance of the new Inception-ResNet-v2, as part of the TF-Slim Image Model Library".

May this help :)

[michaelisard]

Closing this, but please reopen if there's an issue.

[ProgramItUp]

Very minor - something is not quite right with @YusongLeng 's link. It should be:

https://research.googleblog.com/2016/08/improving-inception-and-image.html