Skip to content

Can't convert some OPs in my TF2 model #1870

Open
@sniper0110

Description

@sniper0110

Describe the bug
I'm trying to convert a tensorflow 2 model to onnx and although I get a model at the end, I just can't use it because some of the OPs were not converted to onnx.

Urgency
Very important to get it to work soon.

System information

  • OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Windows
  • Tensorflow Version: 2.8
  • Python version: 3.8

To Reproduce
I ran the command :
python -m tf2onnx.convert --saved-model model_v0 --opset 15 --output model_v0.onnx
The model that I used is model_v0.zip

Additional context
The exact errors that I got are these:
ERROR - Tensorflow op [StatefulPartitionedCall/attend_copy_parse_model/mul: SparseDenseCwiseMul] is not supported
ERROR - Tensorflow op [StatefulPartitionedCall/attend_copy_parse_model/SparseReduceSum: SparseReduceSum] is not supported
ERROR - Unsupported ops: Counter({'SparseDenseCwiseMul': 1, 'SparseReduceSum': 1})

Thanks in advance!

Metadata

Metadata

Assignees

No one assigned

    Labels

    unsupported opsIssues related to unsupported operators

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions