-
Notifications
You must be signed in to change notification settings - Fork 1.9k
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
tfjs converter cannot parse complex64 as dtype #8080
Comments
add proper handling for DT.COMPLEX64 in https://github.com/tensorflow/tfjs/blob/master/tfjs-converter/src/operations/operation_mapper.ts?rgh-link-date=2023-11-23T20%3A35%3A02Z#L482 fix the issue. I am interested in submitting a PR but have no idea which test needs to be written to test this case. |
Hi, @Lutra-Fs Thank you for bringing this issue to our attention and I was trying to replicate the same issue from my end and I'm also getting the same error message which you mentioned above in issue template, It seems like we haven't If you want to submit PR to take care of this issue please feel free to do it and please refer this file which may help you to write test cases for this issue, once you submit PR our team may help or guide you to write test cases if possible. Thank you.
|
PR opened |
Hi, @Lutra-Fs We really appreciate your submission of the PR addressing this issue. Our team will thoroughly review your PR and take the necessary actions accordingly. Thank you for your contribution |
System information
Describe the current behavior
currently if a model has
tf.cast(x, tf.complex64)
it could be exported correctly but the browser/nodejs throws an error withSee also kymatio/kymatio#980. That issue's author meets the same issue.
Describe the expected behavior
model runs without error
Standalone code to reproduce the issue
Provide a reproducible test case that is the bare minimum necessary to generate
the problem. If possible, please share a link to Colab/CodePen/any notebook.
I am trying to build one with the minimal info. Our Python code is in https://github.com/techlauncher-mlai-edge-physics/neural_net_physics_tf. I will update this field after I create the link. the tfjs model if helps:
FNO2dLite_from_saved_model.zip
a minimal tfjs
model.json
build with only a cast call. accepts [-1,10] shape.tfjs_test_model.zip
Other info / logs
I am aware of https://github.com/tensorflow/tfjs/blob/master/tfjs-converter/src/operations/operation_mapper.ts#L482 and it seems like the DT_COMPLEX64 is not handled in that function, so a null will always return.
The text was updated successfully, but these errors were encountered: