From 5a9619beb0ed7f32bce2cd90e16eb4bea162507a Mon Sep 17 00:00:00 2001 From: Faith Xu Date: Thu, 26 Aug 2021 17:05:33 -0700 Subject: [PATCH] remove Keras-ONNX references (#8820) --- docs/how-to/tune-performance.md | 2 +- docs/resources/compatibility.md | 1 - docs/tutorials/inferencing/tensorflow.md | 14 ++------------ 3 files changed, 3 insertions(+), 14 deletions(-) diff --git a/docs/how-to/tune-performance.md b/docs/how-to/tune-performance.md index e51afe2775..18ee2f7ff5 100644 --- a/docs/how-to/tune-performance.md +++ b/docs/how-to/tune-performance.md @@ -247,4 +247,4 @@ Depending on which execution provider you're using, it may not have full support NCHW and NHWC are two different memory layout for 4-D tensors. -Most TensorFlow operations used by a CNN support both NHWC and NCHW data format. The Tensorflow team suggests that on GPU NCHW is faster but on CPU NHWC is sometimes faster in Tensorflow. However, ONNX only supports NCHW. As a result, if the original model is in NHWC format, when the model is converted extra transposes may be added. The [tensorflow-onnx](https://github.com/onnx/tensorflow-onnx) and [keras-onnx](https://github.com/onnx/keras-onnx) converters do remove many of these transposes, but if this doesn't help sufficiently, consider retraining the model using NCHW. +Most TensorFlow operations used by a CNN support both NHWC and NCHW data format. The Tensorflow team suggests that on GPU NCHW is faster but on CPU NHWC is sometimes faster in Tensorflow. However, ONNX only supports NCHW. As a result, if the original model is in NHWC format, when the model is converted extra transposes may be added. The [tensorflow-onnx](https://github.com/onnx/tensorflow-onnx) converter does remove many of these transposes, but if this doesn't help sufficiently, consider retraining the model using NCHW. diff --git a/docs/resources/compatibility.md b/docs/resources/compatibility.md index e048b6cafb..2fb81af79f 100644 --- a/docs/resources/compatibility.md +++ b/docs/resources/compatibility.md @@ -90,7 +90,6 @@ Unless otherwise noted, please use the latest released version of the tools to c |[ONNXMLTools](https://pypi.org/project/onnxmltools/)
CoreML, LightGBM, XGBoost, LibSVM|[Latest stable](https://github.com/onnx/onnxmltools/releases)| |[ONNXMLTools](https://pypi.org/project/onnxmltools/)
SparkML|[Latest stable](https://github.com/onnx/onnxmltools/releases)| |[SKLearn-ONNX](https://pypi.org/project/skl2onnx/)|[Latest stable](https://github.com/onnx/sklearn-onnx/releases)| -|[Keras-ONNX](https://pypi.org/project/keras2onnx/)|[Latest stable](https://github.com/onnx/keras-onnx/releases)| |[Tensorflow-ONNX](https://pypi.org/project/tf2onnx/)|[Latest stable](https://github.com/onnx/tensorflow-onnx/releases)| |[WinMLTools](https://docs.microsoft.com/en-us/windows/ai/windows-ml/convert-model-winmltools)|[Latest stable](https://pypi.org/project/winmltools/)| |[AutoML](https://docs.microsoft.com/en-us/azure/machine-learning/service/concept-automated-ml)|[1.0.39+](https://pypi.org/project/azureml-automl-core) (ONNX v1.5)
[1.0.33](https://pypi.org/project/azureml-automl-core/1.0.33/) (ONNX v1.4) | diff --git a/docs/tutorials/inferencing/tensorflow.md b/docs/tutorials/inferencing/tensorflow.md index f1e8100199..151b1c0ce5 100644 --- a/docs/tutorials/inferencing/tensorflow.md +++ b/docs/tutorials/inferencing/tensorflow.md @@ -20,7 +20,7 @@ ONNX Runtime can accelerate inferencing times for TensorFlow, TFLite, and Keras ## Export model to ONNX -### TensorFlow +### TensorFlow/Keras These examples use the [TensorFlow-ONNX converter](https://github.com/onnx/tensorflow-onnx), which supports TensorFlow 1, 2, Keras, and TFLite model formats. @@ -29,21 +29,11 @@ These examples use the [TensorFlow-ONNX converter](https://github.com/onnx/tenso * [TensorFlow: Image classification (efficientnet-edge)](https://github.com/onnx/tensorflow-onnx/blob/master/tutorials/efficientnet-edge.ipynb) * [TensorFlow: Image classification (efficientnet-lite)](https://github.com/onnx/tensorflow-onnx/blob/master/tutorials/efficientnet-lite.ipynb) * [TensorFlow: Natural Language Processing (BERT)](https://github.com/onnx/tensorflow-onnx/blob/master/tutorials/BertTutorial.ipynb) +* [Keras: Image classification (Resnet 50)](https://github.com/onnx/tensorflow-onnx/blob/master/tutorials/keras-resnet50.ipynb) ### TFLite * [TFLite: Image classifciation (mobiledet)](https://github.com/onnx/tensorflow-onnx/blob/master/tutorials/mobiledet-tflite.ipynb) -### Keras -Keras models can be converted using either the [tensorflow-onnx](https://github.com/onnx/tensorflow-onnx) or [Keras-ONNX converter](https://github.com/onnx/keras-onnx). The TensorFlow-ONNX converter supports newer opsets with more active support. - -* [tf2onnx: Image classification (Resnet 50)](https://github.com/onnx/tensorflow-onnx/blob/master/tutorials/keras-resnet50.ipynb) -* [keras2onnx: Image classification (efficientnet)](https://github.com/onnx/keras-onnx/blob/master/tutorial/TensorFlow_Keras_EfficientNet.ipynb) -* [keras2onnx: Image classification (Densenet)](https://www.onnxruntime.ai/python/auto_examples/plot_dl_keras.html#sphx-glr-auto-examples-plot-dl-keras-py) -* [keras2onnx: Natural Language Processing (BERT)](https://github.com/microsoft/onnxruntime/tree/master/onnxruntime/python/tools/transformers/notebooks/Tensorflow_Keras_Bert-Squad_OnnxRuntime_CPU.ipynb) -* [keras2onnx: Handwritten Digit Recognition (MNIST)](https://github.com/onnx/keras-onnx/blob/master/tutorial/TensorFlow_Keras_MNIST.ipynb) - - - ## Accelerate TensorFlow model inferencing * [Accelerate BERT model on CPU](https://github.com/microsoft/onnxruntime/blob/master/onnxruntime/python/tools/transformers/notebooks/PyTorch_Bert-Squad_OnnxRuntime_CPU.ipynb) * [Accelerate BERT model on GPU](https://github.com/microsoft/onnxruntime/blob/master/onnxruntime/python/tools/transformers/notebooks/PyTorch_Bert-Squad_OnnxRuntime_GPU.ipynb) \ No newline at end of file