Merge branch 'master' into r1.8

Author: guschmue

.github/ISSUE_TEMPLATE/bug-performance-issue.md

@@ -4,6 +4,13 @@ about: Use this template for reporting a bug or a performance issue.
 
 ---
 
+**Dubugging advice**
+[delete this section if it doesn't solve your issue]
+- Add a `--opset` flag with the highest possible opset you can use. Some ops only convert in higher opsets.
+- Try installing the latest tf2onnx from master. Some bug fixes might not have been released to PyPI. Run `pip uninstall tf2onnx` and `pip install git+https://github.com/onnx/tensorflow-onnx`
+- If using a saved model, use the Tensorflow `saved_model_cli` to determine the correct `--tag` and `--signature_def` flags to use. If the signature you need is not listed, use the `--concrete_function` flag to index into the model's defined functions.
+- If your model was made in tf1.x, try running tf2onnx in a venv with tensorflow 1.x installed. tf2.x should be able to read tf1 models, but sometimes there are bugs.
+
 **Describe the bug**
 A clear and concise description of what the bug is.
 
@@ -16,10 +23,7 @@ If there are particular important use cases blocked by this or strict project-re
 - Python version:
 
 **To Reproduce**
-Describe steps/code to reproduce the behavior:
-
-**Expected behavior**
-A clear and concise description of what you expected to happen.
+Describe steps/code to reproduce the behavior. Please upload/link the model you are trying to convert if possible.
 
 **Screenshots**
 If applicable, add screenshots to help explain your problem.

README.md

@@ -197,6 +197,10 @@ def convert_to_tflite(self, graph_def, feed_dict, outputs):
             sess_outputs = [sess.graph.get_tensor_by_name(n) for n in outputs]
             converter = tf_lite.TFLiteConverter.from_session(sess, sess_inputs, sess_outputs)
             #converter.optimizations = [tf.lite.Optimize.DEFAULT]
+            converter.target_spec.supported_ops = [
+                tf.lite.OpsSet.TFLITE_BUILTINS,    # enable TensorFlow Lite ops.
+                tf.lite.OpsSet.SELECT_TF_OPS,      # enable TensorFlow flex ops.
+            ]
 
             from tensorflow.lite.python.convert import ConverterError
             try: