Luchang Li

TensorFlow Lite convert BatchMatMulV2 into Split+multi FullyConnected + Pack

2

### 1. System information - OS Platform and Distribution (e.g., Linux Ubuntu 16.04):ub18.04 - TensorFlow installation (pip package or built from source):tf2.9 When I convert a transformer model with BatchMatMulV2...

ModuleNotFoundError: No module named 'libpymo'

19

When I tried to use aimet 1.14.0 for tensorflow, I got this error. Do you know how to solve it?

GATHER: not supported by armnn: in validate_arguments src/core/CL/kernels/CLGatherKernel.cpp:41: indices->num_dimensions() > 1

3

GATHER: not supported by armnn: in validate_arguments src/core/CL/kernels/CLGatherKernel.cpp:41: indices->num_dimensions() > 1

ERROR: Operator SQUARED_DIFFERENCE is not supported by armnn_delegate.

1

ERROR: Operator SQUARED_DIFFERENCE is not supported by armnn_delegate.

ERROR: Operator FULLY_CONNECTED is not supported by armnn_delegate.

1

ERROR: Operator FULLY_CONNECTED is not supported by armnn_delegate.

RuntimeError: /github/workspace/onnx/version_converter/adapters/softmax_12_13.h:53: adapt_softmax_12_13: Assertion `target_shape.size() != 0` failed: Version conversion for Softmax failed because input shape is unknown.

3

when I convert onnx model by onnx_model = version_convert(onnx_model, tgt_opset=15) this error happens for some model actually there are input and output shape for this node I don't know how...

TfLiteArmnnDelegate: Exception (FullyConnectedLayer: TensorShape set on OutputSlot[0] does not match the infer

3

terminating with uncaught exception of type armnn::Exception: TfLiteArmnnDelegate: Exception (FullyConnectedLayer: TensorShape set on OutputSlot[0] does not match the infer red shape. : [1,512,768] != [512,768]) caught from optimize. Aborted model:...

[Bug]: terminating with uncaught exception of type std::bad_cast: std::bad_cast

3

### 软件环境 ```Markdown fast_tokenizer-android-arm64-v8a-1.0.2.tgz ``` ### 重复问题 - [X] I have searched the existing issues ### 错误描述 ```Markdown 我写了一个简单的android ndk程序测试fast_tokenizer，但是程序运行时提示terminating with uncaught exception of type std::bad_cast: std::bad_cast并挂掉。请问这是哪里出了问题呢？ 代码如下： main.cpp /*...

[BUG] failed when the model size is larger than 2GB

2

**Describe the bug** failed when the model size is larger than 2GB [libprotobuf ERROR /project/third_party/onnx-optimizer/third_party/protobuf/src/google/protobuf/message_lite.cc:449] onnx.ModelProto exceeded maximum protobuf size of 2GB: 3438405053 Traceback (most recent call last): File "convert_init_to_half.py",...

优化命令 onnxsim model.b1.onnx model.b1.sim.onnx 版本0.4.24 Your model contains "Tile" ops or/and "ConstantOfShape" ops. Folding these ops can make the simplified model much larger. If it is not expected, please specify...