Training complete on TPU - full dataset - lr=1e-4

README.md

@@ -20,11 +20,11 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [dslim/distilbert-NER](https://huggingface.co/dslim/distilbert-NER) on an unknown dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0367
-- Precision: 0.9569
-- Recall: 0.9677
-- F1: 0.9623
-- Accuracy: 0.9915
+- Loss: 0.6583
+- Precision: 0.2957
+- Recall: 0.0831
+- F1: 0.1297
+- Accuracy: 0.8101
 
 ## Model description
 
@@ -43,7 +43,7 @@ More information needed
 ### Training hyperparameters
 
 The following hyperparameters were used during training:
-- learning_rate: 1e-05
+- learning_rate: 0.0001
 - train_batch_size: 8
 - eval_batch_size: 8
 - seed: 42
@@ -55,7 +55,7 @@ The following hyperparameters were used during training:
 
 | Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1     | Accuracy |
 |:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
-| 0.0315        | 1.0   | 1500 | 0.0367          | 0.9569    | 0.9677 | 0.9623 | 0.9915   |
+| 0.7138        | 1.0   | 1500 | 0.6583          | 0.2957    | 0.0831 | 0.1297 | 0.8101   |
 
 
 ### Framework versions