add load_op_libraries in docs

Signed-off-by: hwangdeyu <[email protected]>
Co-authored-by: fatcat-z <[email protected]>

Author: hwangdeyu

README.md

@@ -140,6 +140,7 @@ python -m tf2onnx.convert
     [--concrete_function CONCRETE_FUNCTION]
     [--target TARGET]
     [--custom-ops list-of-custom-ops]
+    [--load_op_libraries tf_custom_ops_library]
     [--fold_const]
     [--large_model]
     [--continue_on_error]
@@ -226,6 +227,10 @@ runtime can still open the model. The format is a comma-separated map of tf op n
 OpName:domain. If only an op name is provided (no colon), the default domain of `ai.onnx.converters.tensorflow`
 will be used.
 
+#### --load_op_libraries
+
+Some ops are not covered by the existing TensorFlow library. You can create the custom library following by [Create an op tf wiki](https://www.tensorflow.org/guide/create_op). And use this parameter to load the comma-separated list of tf op library paths, each library file usually ends with `.so`. 
+
 #### --target
 
 Some models require special handling to run on some runtimes. In particular, the model may use unsupported data types. Workarounds are activated with ```--target TARGET```. Currently supported values are listed on this [wiki](https://github.com/onnx/tensorflow-onnx/wiki/target). If your model will be run on Windows ML, you should specify the appropriate target value.

tf2onnx/convert.py

@@ -69,9 +69,9 @@ def get_args():
     parser.add_argument("--use-graph-names", help="(saved model only) skip renaming io using signature names",
                         action="store_true")
     parser.add_argument("--opset", type=int, default=None, help="opset version to use for onnx domain")
-    parser.add_argument("--dequantize", help="Remove quantization from model. Only supported for tflite currently.",
+    parser.add_argument("--dequantize", help="remove quantization from model. Only supported for tflite currently.",
                         action="store_true")
-    parser.add_argument("--custom-ops", help="Comma-separated map of custom ops to domains in format OpName:domain. "
+    parser.add_argument("--custom-ops", help="comma-separated map of custom ops to domains in format OpName:domain. "
                                              "Domain 'ai.onnx.converters.tensorflow' is used by default.")
     parser.add_argument("--extra_opset", default=None,
                         help="extra opset with format like domain:version, e.g. com.microsoft:1")