HueyNemud
/

icdar23-entrydetector_plaintext

@@ -13,13 +13,13 @@ should probably proofread and complete it, then remove this comment. -->
 This model is a fine-tuned version of [HueyNemud/das22-10-camembert_pretrained](https://huggingface.co/HueyNemud/das22-10-camembert_pretrained) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0191
-- Ebegin: {'precision': 0.993127147766323, 'recall': 0.9626915389740173, 'f1': 0.9776725304465493, 'number': 3002}
-- Eend: {'precision': 0.9910313901345291, 'recall': 0.9576666666666667, 'f1': 0.9740634005763689, 'number': 3000}
-- Overall Precision: 0.9921
-- Overall Recall: 0.9602
-- Overall F1: 0.9759
-- Overall Accuracy: 0.9954
 ## Model description
@@ -44,24 +44,29 @@ The following hyperparameters were used during training:
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- training_steps: 6000
 ### Training results
 | Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1     | Accuracy |
 |:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
-| No log        | 0.07  | 300  | 0.0312          | 0.9804    | 0.9791 | 0.9797 | 0.9960   |
-| 0.1423        | 0.14  | 600  | 0.0232          | 0.9920    | 0.9670 | 0.9793 | 0.9958   |
-| 0.1423        | 0.21  | 900  | 0.0164          | 0.9959    | 0.9696 | 0.9825 | 0.9965   |
-| 0.0242        | 0.29  | 1200 | 0.0174          | 0.9850    | 0.9744 | 0.9797 | 0.9959   |
-| 0.0169        | 0.36  | 1500 | 0.0165          | 0.9913    | 0.9696 | 0.9803 | 0.9960   |
-| 0.0169        | 0.43  | 1800 | 0.0168          | 0.9908    | 0.9715 | 0.9810 | 0.9962   |
-| 0.0153        | 0.5   | 2100 | 0.0164          | 0.9884    | 0.9702 | 0.9793 | 0.9960   |
 ### Framework versions
-- Transformers 4.26.0
 - Pytorch 1.13.1+cu116
 - Datasets 2.9.0
 - Tokenizers 0.13.2

 This model is a fine-tuned version of [HueyNemud/das22-10-camembert_pretrained](https://huggingface.co/HueyNemud/das22-10-camembert_pretrained) on the None dataset.
 It achieves the following results on the evaluation set:
+- Loss: 0.0319
+- Ebegin: {'precision': 0.9928057553956835, 'recall': 0.9341857841293719, 'f1': 0.9626041464832397, 'number': 2659}
+- Eend: {'precision': 0.9772727272727273, 'recall': 0.9641255605381166, 'f1': 0.9706546275395034, 'number': 2676}
+- Overall Precision: 0.9848
+- Overall Recall: 0.9492
+- Overall F1: 0.9667
+- Overall Accuracy: 0.9936
 ## Model description
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
+- training_steps: 7500
 ### Training results
 | Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1     | Accuracy |
 |:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
+| No log        | 0.07  | 300  | 0.0422          | 0.9698    | 0.9675 | 0.9686 | 0.9945   |
+| 0.2178        | 0.14  | 600  | 0.0249          | 0.9900    | 0.9637 | 0.9767 | 0.9952   |
+| 0.2178        | 0.21  | 900  | 0.0236          | 0.9859    | 0.9721 | 0.9790 | 0.9957   |
+| 0.0267        | 0.29  | 1200 | 0.0187          | 0.9908    | 0.9711 | 0.9808 | 0.9961   |
+| 0.0209        | 0.36  | 1500 | 0.0191          | 0.9869    | 0.9727 | 0.9798 | 0.9959   |
+| 0.0209        | 0.43  | 1800 | 0.0199          | 0.9886    | 0.9712 | 0.9798 | 0.9959   |
+| 0.0167        | 0.5   | 2100 | 0.0178          | 0.9912    | 0.9715 | 0.9813 | 0.9962   |
+| 0.0167        | 0.57  | 2400 | 0.0176          | 0.9937    | 0.9595 | 0.9763 | 0.9952   |
+| 0.0147        | 0.64  | 2700 | 0.0213          | 0.9869    | 0.9692 | 0.9779 | 0.9955   |
+| 0.0142        | 0.72  | 3000 | 0.0181          | 0.9854    | 0.9767 | 0.9810 | 0.9962   |
+| 0.0142        | 0.79  | 3300 | 0.0222          | 0.9865    | 0.9744 | 0.9804 | 0.9960   |
+| 0.0121        | 0.86  | 3600 | 0.0190          | 0.9855    | 0.9770 | 0.9813 | 0.9962   |
 ### Framework versions
+- Transformers 4.26.1
 - Pytorch 1.13.1+cu116
 - Datasets 2.9.0
 - Tokenizers 0.13.2