tensorflow custom implementation request: SparseReorder, SparseTensorDenseMatMul - Cplusplus

System information - OS Platform and Distribution (e.g., Linux Ubuntu 16.04): windows10, x86-64 - TensorFlow installed from (source or binary): source - TensorFlow version (or github SHA if from source): 2.7.0-dev20210702

Provide the text output from tflite_convert

2021-07-02 16:52:27.481365: W tensorflow/compiler/mlir/lite/flatbuffer_export.cc:1851] The following operation(s) need TFLite custom op implementation(s):
Custom ops: SparseReorder, SparseTensorDenseMatMul

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/Jupyter/any notebook.

Also, please include a link to a GraphDef or the model if possible.

Any other info / logs

Include any logs or source code that would be helpful to diagnose the problem. If including tracebacks, please include the full traceback. Large logs and files should be attached.

Details: tf.SparseReorder(tensor, tensor, tensor<3xi64>) -> (tensor, tensor) : {T = f32, device = ""} tf.SparseTensorDenseMatMul(tensor, tensor, tensor<2xi64>, tensor<6400x256xf32>) -> (tensor) : {T = f32, Tindices = i64, adjointa = false, adjointb = false, device = ""}

Asked Oct 15 '21 09:10
avatar RedJyve

2 Answer:

I adding them as Select TF Ops. Please wait a bit before it get merged.

Answered Jul 05 '21 at 03:20
avatar  of thaink

the fixed is merged to master branch. You can try it in tomorrow nightly. However, you will need to treat them as Select TF Ops. Please follow https://www.tensorflow.org/lite/guide/ops_select

Answered Jul 05 '21 at 04:53
avatar  of thaink