Update README.md

README.md

@@ -12,8 +12,27 @@ metrics:
 
 This is a T5 small model finetuned on CoNLL-2003 dataset for named entity recognition (NER).
 
-Example input and output:
+Example Input and Output:
 “Recognize all the named entities in this sequence (replace named entities with one of [PER], [ORG], [LOC], [MISC]): When Alice visited New York” → “When PER visited LOC LOC"
 
+Evaluation Result:
+
+% of match (for comparison with ExT5: https://arxiv.org/pdf/2111.10952.pdf): 
+
+| Model| ExT5_{Base} | This Model | T5_NER_CONLL_OUTPUTLIST 
+| :---: | :---: | :---: | :---: |
+| % of Complete Match| 86.53 | 79.03 | 123 | 
+
+
+
+There are some outputs (212/3453 that does not have the same length as the input)
+
+F1 score on testing set of those with matching length:
+
+| Attempt | #1 | #2 |
+| :---: | :---: | :---: |
+| Seconds | 301 | 283 |
+
+