Pytorch has its builtin ONNX exporter check here for details
+ +Pytorch has its builtin ONNX exporter check [here](https://pytorch.org/docs/stable/onnx.html) for details. ## Install + You can install latest release of ONNXMLTools from [PyPi](https://pypi.org/project/onnxmltools/): -``` + +```bash pip install onnxmltools ``` + or install from source: -``` + +```bash pip install git+https://github.com/microsoft/onnxconverter-common pip install git+https://github.com/onnx/onnxmltools ``` + If you choose to install `onnxmltools` from its source code, you must set the environment variable `ONNX_ML=1` before installing the `onnx` package. ## Dependencies + This package relies on ONNX, NumPy, and ProtoBuf. If you are converting a model from scikit-learn, Core ML, Keras, LightGBM, SparkML, XGBoost, H2O, CatBoost or LibSVM, you will need an environment with the respective package installed from the list below: + 1. scikit-learn 2. CoreMLTools (version 3.1 or lower) 3. Keras (version 2.0.8 or higher) with the corresponding Tensorflow version @@ -47,9 +57,11 @@ This package relies on ONNX, NumPy, and ProtoBuf. If you are converting a model ONNXMLTools is tested with Python **3.7+**. # Examples + If you want the converted ONNX model to be compatible with a certain ONNX version, please specify the target_opset parameter upon invoking the convert function. The following Keras model conversion example demonstrates this below. You can identify the mapping from ONNX Operator Sets (referred to as opsets) to ONNX releases in the [versioning documentation](https://github.com/onnx/onnx/blob/master/docs/Versioning.md#released-versions). ## Keras to ONNX Conversion + Next, we show an example of converting a Keras model into an ONNX model with `target_opset=7`, which corresponds to ONNX release version 1.2. ```python @@ -83,6 +95,7 @@ onnx_model = onnxmltools.convert_keras(keras_model, target_opset=7) ``` ## CoreML to ONNX Conversion + Here is a simple code snippet to convert a Core ML model into an ONNX model. ```python @@ -100,7 +113,8 @@ onnxmltools.utils.save_model(onnx_model, 'example.onnx') ``` ## H2O to ONNX Conversion -Below is a code snippet to convert a H2O MOJO model into an ONNX model. The only pre-requisity is to have a MOJO model saved on the local file-system. + +Below is a code snippet to convert a H2O MOJO model into an ONNX model. The only prerequisite is to have a MOJO model saved on the local file-system. ```python import onnxmltools @@ -122,7 +136,7 @@ backend of your choice. You can check the operator set of your converted ONNX model using [Netron](https://github.com/lutzroeder/Netron), a viewer for Neural Network models. Alternatively, you could identify your converted model's opset version through the following line of code. -``` +```python opset_version = onnx_model.opset_import[0].version ``` @@ -138,7 +152,8 @@ All converter unit test can generate the original model and converted model to a [onnxruntime](https://pypi.org/project/onnxruntime/) or [onnxruntime-gpu](https://pypi.org/project/onnxruntime-gpu/). The unit test cases are all the normal python unit test cases, you can run it with pytest command line, for example: -``` + +```bash python -m pytest --ignore .\tests\ ``` @@ -159,4 +174,5 @@ be added in *tests_backend* to compute the prediction with the runtime. # License + [Apache License v2.0](LICENSE) diff --git a/onnxmltools/convert/sparkml/README.md b/onnxmltools/convert/sparkml/README.md index 179dd640..7d3bbd3b 100644 --- a/onnxmltools/convert/sparkml/README.md +++ b/onnxmltools/convert/sparkml/README.md @@ -1,6 +1,6 @@ -# Spark ML to Onnx Model Conversion +# Spark ML to ONNX Model Conversion There is prep work needed above and beyond calling the API. In short these steps are: @@ -9,72 +9,74 @@ There is prep work needed above and beyond calling the API. In short these steps * taking the output Tensor(s) and converting it(them) back to a DataFrame if further processing is required. ## Instructions + For examples, please see the unit tests under `tests/sparkml` -1- Create a list of input types needed to be supplied to the `convert_sparkml()` call. -For simple cases you can use `buildInitialTypesSimple()` function in `convert/sparkml/utils.py`. -To use this function just pass your test DataFrame. +1. Create a list of input types needed to be supplied to the `convert_sparkml()` call. + + For simple cases you can use `buildInitialTypesSimple()` function in `convert/sparkml/utils.py`. + To use this function just pass your test DataFrame. + + Otherwise, the conversion code requires a list of tuples with input names and their corresponding Tensor types, as shown below: -Otherwise, the conversion code requires a list of tuples with input names and their corresponding Tensor types, as shown below: -```python -initial_types = [ - ("label", StringTensorType([1, 1])), - # (repeat for the required inputs) -] -``` -Note that the input names are the same as columns names from your DataFrame and they must match the "inputCol(s)" values + ```python + initial_types = [ + ("label", StringTensorType([1, 1])), + # (repeat for the required inputs) + ] + ``` -you provided when you created your Pipeline. + Note that the input names are the same as columns names from your DataFrame and they must match the "inputCol(s)" values -2- Now you can create the ONNX model from your pipeline model like so: -```python -pipeline_model = pipeline.fit(training_data) -onnx_model = convert_sparkml(pipeline_model, 'My Sparkml Pipeline', initial_types) -``` + you provided when you created your Pipeline. -3- (optional) You could save the ONNX model for future use or further examination by using the `SerializeToString()` +2. Now you can create the ONNX model from your pipeline model like so: + + ```python + pipeline_model = pipeline.fit(training_data) + onnx_model = convert_sparkml(pipeline_model, 'My Sparkml Pipeline', initial_types) + ``` + +3. (optional) You could save the ONNX model for future use or further examination by using the `SerializeToString()` method of ONNX model -```python -with open("model.onnx", "wb") as f: - f.write(onnx_model.SerializeToString()) -``` + ```python + with open("model.onnx", "wb") as f: + f.write(onnx_model.SerializeToString()) + ``` -4- Before running this model (e.g. using `onnxruntime`) you need to create a `dict` from the input data. This dictionay +4. Before running this model (e.g. using `onnxruntime`) you need to create a `dict` from the input data. This dictionary will have entries for each input name and its corresponding TensorData. For simple cases you could use the function `buildInputDictSimple()` and pass your testing DataFrame to it. Otherwise, you need to create something like the following: -```python -input_data = {} -input_data['label'] = test_df.select('label').toPandas().values -# ... (repeat for all desired inputs) -``` + ```python + input_data = {} + input_data['label'] = test_df.select('label').toPandas().values + # ... (repeat for all desired inputs) + ``` +5. (optional) You could save the converted input data for possible debugging or future reuse. See below: -5- (optional) You could save the converted input data for possible debugging or future reuse. See below: -```python -with open("input_data", "wb") as f: - pickle.dump(input, f) -``` + ```python + with open("input_data", "wb") as f: + pickle.dump(input, f) + ``` -6- And finally run the newly converted ONNX model in the runtime: -```python -sess = onnxruntime.InferenceSession(onnx_model) -output = sess.run(None, input_data) +6. And finally run the newly converted ONNX model in the runtime: -``` - This output may need further conversion back to a DataFrame. + ```python + sess = onnxruntime.InferenceSession(onnx_model) + output = sess.run(None, input_data) + ``` +This output may need further conversion back to a DataFrame. ## Known Issues -1. Overall invalid data handling is problematic and not implemented in most cases. -Make sure your data is clean. +1. Overall invalid data handling is problematic and not implemented in most cases. Make sure your data is clean. -2. OneHotEncoderEstimator must not drop the last bit: OneHotEncoderEstimator has an option -which you can use to make sure the last bit is included in the vector: `dropLast=False` +2. When converting `OneHotEncoderModel` to ONNX, if `handleInvalid` is set to `"keep"`, then `dropLast` must be set to `True`. If `handleInvalid` is set to `"error"`, then `dropLast` must be set to `False`. -3. Use FloatTensorType for all numbers (instead of Int6t4Tensor or other variations) +3. Use `FloatTensorType` for all numbers (instead of `Int64Tensor` or other variations) 4. Some conversions, such as the one for Word2Vec, can only handle batch size of 1 (one input row) -