fix the post-processing bug in NER (#534)

* fix conll bug

* update DataCollatorForAuto

* adding label_list comments

test/automl/test_classification.py

@@ -347,3 +347,4 @@ class TestClassification(unittest.TestCase):
 
 if __name__ == "__main__":
     unittest.main()
+    test = TestClassification()

test/nlp/test_autohf_tokenclassification.py

@@ -13,7 +13,18 @@ def test_tokenclassification():
 
     automl_settings = get_automl_settings()
     automl_settings["task"] = "token-classification"
-    automl_settings["metric"] = "seqeval"
+    automl_settings["metric"] = "seqeval:overall_f1"  # evaluating based on the overall_f1 of seqeval
+    automl_settings["fit_kwargs_by_estimator"]["transformer"]["label_list"] = [
+        "O",
+        "B-PER",
+        "I-PER",
+        "B-ORG",
+        "I-ORG",
+        "B-LOC",
+        "I-LOC",
+        "B-MISC",
+        "I-MISC",
+    ]
 
     try:
         automl.fit(