Skip to content
This repository has been archived by the owner on Oct 13, 2021. It is now read-only.

Error when converting Keras ConvLSTM2D layer into ONNX #731

Open
YizhouFan opened this issue Jul 8, 2021 · 1 comment
Open

Error when converting Keras ConvLSTM2D layer into ONNX #731

YizhouFan opened this issue Jul 8, 2021 · 1 comment

Comments

@YizhouFan
Copy link

YizhouFan commented Jul 8, 2021

My code snippet for conversion:

MODEL_NAME = "data/pie/intention/context_loc_pretrained/model"
model = load_model(MODEL_NAME + ".h5")
onnx_model = keras2onnx.convert_keras(model, model.name, target_opset=14)
temp_model_file = MODEL_NAME + ".onnx"
onnx.save_model(onnx_model, temp_model_file)

Error message:

ValueError: Unable to find out a correct type for tensor type = 0 of conv_lst_m2d_1/while/TensorArrayWrite/TensorArrayWriteV3/Enter:0

Network definition:

See the open-source repository here,
https://github.com/aras62/PIEPredict/blob/df76b6512dd8afa8e695025e306a9a21f54b1179/pie_intent.py#L451

Versions:

Keras2onnx: 1.9.0
onnxconverter-common: 1.9.0
tensorflow: 1.14.0
Keras: 2.2.4

@nirlviz
Copy link

nirlviz commented Jul 28, 2021

Happens to me too.
Any updates about this?

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants