Catastrophic Forgetting Fixed

README.md

@@ -1,5 +1,6 @@
 ---
 license: mit
+base_model: dslim/bert-large-NER
 tags:
 - generated_from_trainer
 metrics:
@@ -7,7 +8,6 @@ metrics:
 - recall
 - f1
 - accuracy
-base_model: dslim/bert-large-NER
 model-index:
 - name: bert-finetuned-ner-adam
   results: []
@@ -21,10 +21,10 @@ should probably proofread and complete it, then remove this comment. -->
 This model is a fine-tuned version of [dslim/bert-large-NER](https://huggingface.co/dslim/bert-large-NER) on an unknown dataset.
 It achieves the following results on the evaluation set:
 - Loss: nan
-- Precision: 0.8340
-- Recall: 0.8131
-- F1: 0.8234
-- Accuracy: 0.9216
+- Precision: 0.8845
+- Recall: 0.8749
+- F1: 0.8797
+- Accuracy: 0.9646
 
 ## Model description
 
@@ -55,9 +55,9 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1     | Accuracy |
 |:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
-| 0.1744        | 1.0   | 893  | nan             | 0.8276    | 0.8115 | 0.8195 | 0.9205   |
-| 0.128         | 2.0   | 1786 | nan             | 0.8404    | 0.8256 | 0.8329 | 0.9238   |
-| 0.0768        | 3.0   | 2679 | nan             | 0.8340    | 0.8131 | 0.8234 | 0.9216   |
+| 0.0949        | 1.0   | 3080 | nan             | 0.8914    | 0.8942 | 0.8928 | 0.9663   |
+| 0.0574        | 2.0   | 6160 | nan             | 0.8763    | 0.8784 | 0.8773 | 0.9635   |
+| 0.0376        | 3.0   | 9240 | nan             | 0.8845    | 0.8749 | 0.8797 | 0.9646   |
 
 
 ### Framework versions