The model-converter failed with a simple Kaldi model that I trained myself

[rzy6461]

Hello,when I used a simple Kaldi model with only full connection layer and softmax layer, I failed to convert from. mdl to IR(.xml and .bin).The error message is as follows：

<class 'extensions.load.kaldi.loader.kaldiloader'="">Model Optimizer is not able to parse Kaldi model

Does this mean that even if the training model only contains the listed Kaldi * supported layers, it may not be converted either? Or what details did I miss？

Looking forward to your reply.

Best regards,

rzy

[rzy6461]

When I used the nnet-am-copy command to convert. mdl to a TXT file, I found that there were two layers, affinecomponentprecondition and sumgroupcomponent, which did not appear in Kaldi * supported layers.

Now I have two questions：

1.What are the differences between the affinecomponentprecondition and the affinecomponent in the model transformation? Will the affinecomponentprecondition layer be supported later？

2. Sumgroupcomponent seems to be a common layer. Why is it not supported now?

Best regards,

rzy

[Adli]

Hi zy,

Thank you for reaching out to us. We are currently investigating this issue and will get back to you soon.

Regards,

Adli

[Adli]

Hi Zy,

A quick check, did you use the 'mo_kaldi.py' command to convert the Kaldi model?

Regards,

Adli

[Adli]

Hi rzy6461,

Thank you for your question. If you need any additional information from Intel, please submit a new question as this thread is no longer being monitored.

Regards,

Adli