[GitHub] [incubator-mxnet] Zha0q1 commented on a change in pull request #20048: [v1.x] Onnx Support for Transformer

[GitBox <gi...@apache.org>]

Zha0q1 commented on a change in pull request #20048:
URL: https://github.com/apache/incubator-mxnet/pull/20048#discussion_r598365388



##########
File path: tests/python-pytest/onnx/test_onnxruntime.py
##########
@@ -988,3 +988,170 @@ def test_ernie_inference_onnxruntime(tmp_path, model_name):
 
     finally:
         shutil.rmtree(tmp_path)
+
+
+@with_seed()
+@pytest.mark.parametrize('model_name', ['transformer_en_de_512'])
+def test_transformer_pretrained_inference_onnxruntime(tmp_path, model_name):
+    tmp_path = str(tmp_path)
+    try:
+        import gluonnlp as nlp
+        dataset = 'WMT2014'
+        ctx = mx.cpu(0)
+        model, _, _ = nlp.model.get_model(
+            name=model_name,
+            ctx=ctx,
+            pretrained=True,
+            dataset_name=dataset)
+
+        model.hybridize(static_alloc=False)
+
+        batch = 7
+        seq_length = 16
+        C_in = 512

Review comment:
       You can refer to this file https://github.com/dmlc/gluon-nlp/blob/v0.10.x/src/gluonnlp/model/transformer.py for C_in and C_out. Those are defined in the pretrained model thus we need to set it the same as in the pretrained model 




-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org