tensorflow-onnx

tensorflow-onnx copied to clipboard

ValueError: Graph has cycles

1

**Describe the bug** ` 2023-09-22 13:51:09.484965: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used. 2023-09-22 13:51:09.533379: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers...

duckheada

Recent ml_dtypes release (0.3.0) removed ml_dtypes.float8_e4m3b11 breaking tf2onnx

3

**Describe the bug** Title says it all, but here's a stack trace: ``` AttributeError: module 'ml_dtypes' has no attribute 'float8_e4m3b11'. Did you mean: 'float8_e4m3fn'? Traceback (most recent call last): File...

blthayer

Cannot assign a device for operation global_step: {{node global_step}} was explicitly assigned to /job:ps/task:0 but available devices are [ /job:localhost/replica:0/task:0/device:CPU:0, /job:localhost/replica:0/task:0/device:GPU:0 ]. Make sure the device specification refers to a valid device.

1

**Describe the bug** **Urgency** **System information** - OS Platform and Distribution (e.g., Linux Ubuntu 18.04*): - TensorFlow Version: - Python version: - ONNX version (if applicable, e.g. 1.11*): - ONNXRuntime...

jingwei0109

tensorflow.python.framework.errors_impl.NotFoundError: Op type not registered 'ParseFeatureOp' in binary running on NQ9PG7CVXG. Make sure the Op and Kernel are registered in the binary running in this process. Note that if you are loading a saved graph which used ops from tf.contrib, accessing (e.g.) `tf.contrib.resampler` should be done before importing the graph, as contrib ops are lazily registered when the module is first accessed.

1

**Describe the bug** **Urgency** **System information** - OS Platform and Distribution (e.g., Linux Ubuntu 18.04*):Apple M2 Pro - TensorFlow Version:2.13.0 - Python version:3.10.12 - ONNX version (if applicable, e.g. 1.11*):1.15.1...

jingwei0109

AttributeError: 'FuncGraph'

1

Hi, I am using the below command to convert a tf-nlp model to onnx. It has a custom-ops built in. !/anaconda/envs/onnxet/bin/python3.10 -m tf2onnx.convert --saved-model {checkpoint_path} --output {onnx_model_output_path} --extra_opset ai.onnx.contrib:1 --custom-ops...

ashtarimo

What does it actually do?

4

# What does it actually do? ### Question The description is unclear. 1. What does this repository actually do? 2. Can I use it to convert ckpt to onnx? 3....

f1am3d

Support for `LayerNormalization` from ONNX opset 17

4

When converting a `tensorflow.keras.layers.LayerNormalization` layer to ONNX, `tf2onnx` currently decomposes layer normalizations into rather complex subgraphs with batch norms and more basic building blocks. Inference engines (like TensorRT in the...

pwuertz

**Describe the bug** I am getting a segfault when attempting to convert a model. **Urgency** None. **System information** - OS Platform and Distribution: macOS Ventura 13.5.2 - TensorFlow Version: 2.13.0...

pjjajal

Tensorflow Lambda layer gives Squeeze error

1

**Describe the bug** I have a tf2 model that uses a lambda layer. When converted to ONNX, I get a strange error involving a Squeeze op. **Urgency** This is blocking...

mchonofsky

Batch normalization layer from Keras only converts to BatchNormalization op if dims >= 4

1

Hi Onnx team, I'm running into a model conversion issue for Kera's BatchNorm layer. The BatchNorm layer is only converted into a BatchNorm Op in Onnx when input Dim>=4. I...

ewenw