End of training

README.md

@@ -1,6 +1,6 @@
 ---
 license: apache-2.0
-base_model: bert-base-uncased
+base_model: distilbert-base-uncased
 tags:
 - generated_from_trainer
 datasets:
@@ -25,16 +25,16 @@ model-index:
     metrics:
     - name: Precision
       type: precision
-      value: 0.9992100120577107
+      value: 0.9624574848236965
     - name: Recall
       type: recall
-      value: 0.999833582958895
+      value: 0.9300632384756199
     - name: F1
       type: f1
-      value: 0.9995217002516272
+      value: 0.9459831157565114
     - name: Accuracy
       type: accuracy
-      value: 0.9997074078999867
+      value: 0.9665913020348451
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -42,13 +42,13 @@ should probably proofread and complete it, then remove this comment. -->
 
 # Bert-NER
 
-This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the ner dataset.
+This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the ner dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0010
-- Precision: 0.9992
-- Recall: 0.9998
-- F1: 0.9995
-- Accuracy: 0.9997
+- Loss: 0.0764
+- Precision: 0.9625
+- Recall: 0.9301
+- F1: 0.9460
+- Accuracy: 0.9666
 
 ## Model description
 
@@ -68,32 +68,20 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 5e-05
-- train_batch_size: 16
-- eval_batch_size: 16
+- train_batch_size: 32
+- eval_batch_size: 32
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- num_epochs: 15
+- num_epochs: 3
 
 ### Training results
 
-| Training Loss | Epoch | Step  | Validation Loss | Precision | Recall | F1     | Accuracy |
-|:-------------:|:-----:|:-----:|:---------------:|:---------:|:------:|:------:|:--------:|
-| 0.0809        | 1.0   | 875   | 0.0838          | 0.9600    | 0.9244 | 0.9419 | 0.9642   |
-| 0.0551        | 2.0   | 1750  | 0.0795          | 0.9622    | 0.9282 | 0.9449 | 0.9660   |
-| 0.0517        | 3.0   | 2625  | 0.0793          | 0.9617    | 0.9275 | 0.9443 | 0.9656   |
-| 0.0508        | 4.0   | 3500  | 0.0739          | 0.9619    | 0.9311 | 0.9463 | 0.9668   |
-| 0.0473        | 5.0   | 4375  | 0.0686          | 0.9604    | 0.9382 | 0.9492 | 0.9685   |
-| 0.0427        | 6.0   | 5250  | 0.0541          | 0.9716    | 0.9610 | 0.9663 | 0.9790   |
-| 0.033         | 7.0   | 6125  | 0.0357          | 0.9934    | 0.9677 | 0.9804 | 0.9880   |
-| 0.0223        | 8.0   | 7000  | 0.0236          | 0.9912    | 0.9815 | 0.9863 | 0.9915   |
-| 0.0151        | 9.0   | 7875  | 0.0167          | 0.9899    | 0.9905 | 0.9902 | 0.9938   |
-| 0.0107        | 10.0  | 8750  | 0.0096          | 0.9955    | 0.9919 | 0.9937 | 0.9960   |
-| 0.0074        | 11.0  | 9625  | 0.0063          | 0.9961    | 0.9970 | 0.9965 | 0.9978   |
-| 0.0051        | 12.0  | 10500 | 0.0042          | 0.9979    | 0.9974 | 0.9977 | 0.9985   |
-| 0.0037        | 13.0  | 11375 | 0.0024          | 0.9988    | 0.9985 | 0.9987 | 0.9992   |
-| 0.0023        | 14.0  | 12250 | 0.0015          | 0.9991    | 0.9994 | 0.9992 | 0.9995   |
-| 0.0014        | 15.0  | 13125 | 0.0010          | 0.9992    | 0.9998 | 0.9995 | 0.9997   |
+| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1     | Accuracy |
+|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
+| No log        | 1.0   | 438  | 0.0869          | 0.9598    | 0.9263 | 0.9427 | 0.9648   |
+| 0.0834        | 2.0   | 876  | 0.0815          | 0.9627    | 0.9280 | 0.9450 | 0.9661   |
+| 0.054         | 3.0   | 1314 | 0.0764          | 0.9625    | 0.9301 | 0.9460 | 0.9666   |
 
 
 ### Framework versions