Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

XceptionNet bug #6501

Closed
fradino opened this issue Sep 30, 2018 · 7 comments

Comments

@fradino
Copy link

commented Sep 30, 2018

Exception in thread "main" java.lang.IllegalStateException: Invalid configuration: disconnected vertices found - [block1_conv2_bn, block12_sepconv3_bn, block14_sepconv1_act]. Disconnected vertices are those that do not connect to either another vertex, and are also not a network output. To disable this error (i.e., allow network configurations with disconnected vertices) use GraphBuilder.allowDisconnected(true)
at org.deeplearning4j.nn.conf.ComputationGraphConfiguration.validate(ComputationGraphConfiguration.java:346)
at org.deeplearning4j.nn.conf.ComputationGraphConfiguration$GraphBuilder.build(ComputationGraphConfiguration.java:1033)
at org.deeplearning4j.zoo.model.Xception.init(Xception.java:78)
at org.deeplearning4j.zoo.model.Xception.init(Xception.java:36)

@maxpumperla

This comment has been minimized.

Copy link

commented Sep 30, 2018

disconnected vertices have been allowed for keras import a while ago (which seems to be the underlying issue here). can you please share which DL4J version you're using?

@fradino

This comment has been minimized.

Copy link
Author

commented Sep 30, 2018

<dl4j.version>1.0.0-beta</dl4j.version>
Btw, I'm not using keras import, just XceptionNet in org.deeplearning4j.zoo.model

@AlexDBlack

This comment has been minimized.

Copy link
Contributor

commented Sep 30, 2018

We do have tests for instantiating xception here:
https://github.com/deeplearning4j/deeplearning4j/blob/451dd76b50355358dc176f2b704e98c43423c5b8/deeplearning4j/deeplearning4j-zoo/src/test/java/org/deeplearning4j/zoo/TestInstantiation.java#L191-L196

Maybe try on the latest version - 1.0.0-beta2. If it still occurs there, can you share the exact code you are using to produce this issue?

@fradino

This comment has been minimized.

Copy link
Author

commented Oct 1, 2018

I use 1.0.0-beta2
and the code i write is
"
ZooModel zooModel = Xception.builder().build();
ComputationGraph sqnet = zooModel.init();
log.info(sqnet.summary());
"
still has this problem

@AlexDBlack AlexDBlack self-assigned this Oct 1, 2018

@AlexDBlack

This comment has been minimized.

Copy link
Contributor

commented Oct 1, 2018

OK, thanks for the code. It looks like this issue is limited to the non-pretrained models.
Pretrained models (via initPretrained()) as per what I linked in my previous post, are OK.

If you want the pretrained model, use initPretrained(). Otherwise (for non-pretrained) it looks like the config needs to be fixed.

@AlexDBlack

This comment has been minimized.

Copy link
Contributor

commented Oct 1, 2018

Fixed here: #6504

Once it's merged, you can get the fix using snapshots: https://deeplearning4j.org/docs/latest/deeplearning4j-config-snapshots

Or, in this case, you can probably just copy the updated model code to a local project (as the changes are isolated to the zoo model class) - https://deeplearning4j.org/docs/latest/deeplearning4j-config-snapshots

AlexDBlack added a commit that referenced this issue Oct 3, 2018
Misc fixes (#6504)
* #6501 Fix multiple issues with Xception configuration

* #6500 Fix SqueezeNet (non-pretrained) config + add tests

* #6497 Fix multi-output net evaluation for ComputationGraph

* #6502 fix formatting for ComputationGraph summary

* #6502 fix formatting for MultiLayerNetwork summary

* #6489 fix JSON mapping for keras import preprocessors

* Switch default output layer activation to softmax

* Fix transposei issues with autoencoder

* Minor test fixes

* ComputationGraph: Fix possible double application of preprocessors and setting input for output layers when doing backprop

* Workspace fix for output layer dropout in compgraph
@lock

This comment has been minimized.

Copy link

commented Nov 2, 2018

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.

@lock lock bot locked and limited conversation to collaborators Nov 2, 2018

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
None yet
3 participants
You can’t perform that action at this time.