Added unit test with pretrained model

examples/text_classification_with_scriptable_tokenizer/README.md

@@ -33,6 +33,7 @@ Subsequently, we need to add the command ```torch.save(model.state_dict(), "mode
 python sst2_classification_non_distributed.py
 ```
 
+A pretrained ```model.pt``` is also available for download [here](https://bert-mar-file.s3.us-west-2.amazonaws.com/text_classification_with_scriptable_tokenizer/model.pt).
 The trained model can then be combined and compiled with TorchScript using the script_tokenizer_and_model.py script. Here ```model.pt``` are the model weights saved after training and ```model_jit.pt``` is the combination of tokenizer and model compiled with TorchScript.
 
 ```bash

test/pytest/test_example_scriptable_tokenzier.py

@@ -3,6 +3,7 @@
 import os
 import shutil
 import sys
+import time
 from argparse import Namespace
 
 import pytest
@@ -145,10 +146,12 @@ def mar_file_path(work_dir, session_mocker, jit_file_path, archiver):
     """
     Create mar file and return file path.
     """
-    mar_file_path = os.path.join(work_dir, "scriptable_tokenizer.mar")
+    model_name = "scriptable_tokenizer_untrained"
+
+    mar_file_path = os.path.join(work_dir, model_name + ".mar")
 
     args = Namespace(
-        model_name="scriptable_tokenizer",
+        model_name=model_name,
         version="1.0",
         serialized_file=jit_file_path,
         model_file=None,
@@ -229,9 +232,12 @@ def model_name(mar_file_path, model_store, torchserve):
     test_utils.unregister_model(model_name)
 
 
-def test_inference_with_sample_text(model_name):
+@pytest.fixture
+def test_file():
+    return os.path.join(EXAMPLE_ROOT_DIR, "sample_text.txt")
+
 
-    test_file = os.path.join(EXAMPLE_ROOT_DIR, "sample_text.txt")
+def test_inference_with_untrained_model_and_sample_text(model_name, test_file):
 
     with open(test_file, "rb") as f:
         response = requests.post(
@@ -250,7 +256,7 @@ def test_inference_with_sample_text(model_name):
     assert float(result_entries["Positive"]) == pytest.approx(0.36607906222343445, 1e-3)
 
 
-def test_inference_with_empty_string(model_name):
+def test_inference_with_untrained_model_and_empty_string(model_name):
 
     data = "".encode("utf8")
 
@@ -268,3 +274,42 @@ def test_inference_with_empty_string(model_name):
     # We're using an untrained model for the unit test, so results do not make sense but should be consistent
     assert float(result_entries["Negative"]) == pytest.approx(0.6082412600517273, 1e-3)
     assert float(result_entries["Positive"]) == pytest.approx(0.3917587101459503, 1e-3)
+
+
+def test_inference_with_pretrained_model(model_store, test_file, torchserve):
+    model_name = "scriptable_tokenizer"
+
+    params = (
+        ("model_name", model_name),
+        (
+            "url",
+            "https://bert-mar-file.s3.us-west-2.amazonaws.com/text_classification_with_scriptable_tokenizer/scriptable_tokenizer.mar",
+        ),
+        ("initial_workers", "1"),
+    )
+
+    test_utils.reg_resp = test_utils.register_model_with_params(params)
+
+    # Give test some time for model to be downloaded from S3 bucket
+    for sleep_time in [2, 4, 8, 16, 32, 64]:
+        with open(test_file, "rb") as f:
+            response = requests.post(
+                url=f"http://localhost:8080/predictions/{model_name}", data=f
+            )
+        if response.status_code == 200:
+            break
+        time.sleep(sleep_time)
+
+    assert response.status_code == 200
+
+    result_entries = json.loads(response.text)
+
+    assert "Negative" in result_entries
+    assert "Positive" in result_entries
+
+    assert float(result_entries["Negative"]) == pytest.approx(
+        0.0001851904089562595, 1e-3
+    )
+    assert float(result_entries["Positive"]) == pytest.approx(0.9998148083686829, 1e-3)
+
+    test_utils.unregister_model(model_name)