Update README.md
Browse files
README.md
CHANGED
|
@@ -119,6 +119,12 @@ The custom dataset of front page texts of patent specifications was assembled in
|
|
| 119 |
|
| 120 |
Our custom dataset has accurate manual labels generated by a graduate student. The final dataset is split 60-20-20 (train-val-test). In the event that the front page text is too long, we restrict the text to the first 512 tokens.
|
| 121 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 122 |
### Evaluation
|
| 123 |
|
| 124 |
Our evaluation metric is F1 at the full entity-level. That is, we aggregated adjacent-indexed entities into full entities and computed F1 scores requiring an exact match. These scores for the test set are below.
|
|
|
|
| 119 |
|
| 120 |
Our custom dataset has accurate manual labels generated by a graduate student. The final dataset is split 60-20-20 (train-val-test). In the event that the front page text is too long, we restrict the text to the first 512 tokens.
|
| 121 |
|
| 122 |
+
|
| 123 |
+
### Training Procedure
|
| 124 |
+
|
| 125 |
+
We use the standard token classification protocols with the HuggingFace Trainer API. We use cross-entropy loss.
|
| 126 |
+
|
| 127 |
+
|
| 128 |
### Evaluation
|
| 129 |
|
| 130 |
Our evaluation metric is F1 at the full entity-level. That is, we aggregated adjacent-indexed entities into full entities and computed F1 scores requiring an exact match. These scores for the test set are below.
|