metadata

tags:
  - generated_from_trainer
model-index:
  - name: icdar23-entrydetector_labelledtext_breaks_indents_left_diff_right_ref
    results: []

icdar23-entrydetector_labelledtext_breaks_indents_left_diff_right_ref

This model is a fine-tuned version of HueyNemud/das22-10-camembert_pretrained on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.3015
Act: {'precision': 0.806146572104019, 'recall': 0.8938401048492791, 'f1': 0.8477315102548166, 'number': 1526}
Cardinal: {'precision': 0.951349296845306, 'recall': 0.962322183775471, 'f1': 0.9568042813455658, 'number': 2601}
Ebegin: {'precision': 0.9863870493009566, 'recall': 0.9951744617668894, 'f1': 0.9907612712490761, 'number': 2694}
Eend: {'precision': 0.9925678186547752, 'recall': 0.9885270170244264, 'f1': 0.9905432968663082, 'number': 2702}
Ft: {'precision': 0.23076923076923078, 'recall': 0.2857142857142857, 'f1': 0.25531914893617025, 'number': 21}
Loc: {'precision': 0.9102217414818821, 'recall': 0.9339622641509434, 'f1': 0.9219391947411668, 'number': 3604}
Per: {'precision': 0.9238871899422358, 'recall': 0.9366172924560799, 'f1': 0.9302086897023606, 'number': 2903}
Titre: {'precision': 0.5961538461538461, 'recall': 0.8266666666666667, 'f1': 0.6927374301675977, 'number': 150}
Overall Precision: 0.9294
Overall Recall: 0.9527
Overall F1: 0.9409
Overall Accuracy: 0.9452

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0001
train_batch_size: 2
eval_batch_size: 2
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
training_steps: 15000

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	0.07	300	0.2069	0.8798	0.9303	0.9044	0.9571
0.4341	0.14	600	0.1650	0.9456	0.9487	0.9471	0.9658
0.4341	0.21	900	0.1539	0.9370	0.9469	0.9419	0.9644
0.1993	0.29	1200	0.1280	0.9502	0.9558	0.9530	0.9692
0.1532	0.36	1500	0.1575	0.9554	0.9507	0.9530	0.9655
0.1532	0.43	1800	0.1213	0.9403	0.9569	0.9485	0.9670
0.128	0.5	2100	0.1075	0.9538	0.9600	0.9569	0.9745
0.128	0.57	2400	0.1351	0.9485	0.9655	0.9569	0.9696
0.1095	0.64	2700	0.1384	0.9446	0.9600	0.9522	0.9678
0.1308	0.72	3000	0.1082	0.9509	0.9617	0.9563	0.9731
0.1308	0.79	3300	0.1246	0.9546	0.9643	0.9594	0.9712
0.1007	0.86	3600	0.1290	0.9484	0.9612	0.9547	0.9689
0.1007	0.93	3900	0.1185	0.9569	0.9604	0.9586	0.9716
0.0996	1.0	4200	0.1144	0.9561	0.9639	0.9600	0.9753
0.078	1.07	4500	0.1120	0.9483	0.9669	0.9575	0.9746
0.078	1.14	4800	0.1285	0.9522	0.9659	0.9590	0.9719
0.0723	1.22	5100	0.1302	0.9413	0.9720	0.9565	0.9703
0.0723	1.29	5400	0.1171	0.9553	0.9687	0.9619	0.9735
0.0728	1.36	5700	0.1256	0.9475	0.9690	0.9581	0.9733
0.0538	1.43	6000	0.1169	0.9505	0.9694	0.9599	0.9745
0.0538	1.5	6300	0.1125	0.9470	0.9712	0.9590	0.9742
0.062	1.57	6600	0.1096	0.9592	0.9675	0.9633	0.9761
0.062	1.65	6900	0.1258	0.9624	0.9638	0.9631	0.9753
0.0515	1.72	7200	0.1256	0.9586	0.9683	0.9634	0.9733
0.0561	1.79	7500	0.1411	0.9559	0.9685	0.9622	0.9727
0.0561	1.86	7800	0.1152	0.9581	0.9672	0.9626	0.9749
0.0566	1.93	8100	0.1196	0.9618	0.9714	0.9666	0.9768
0.0566	2.0	8400	0.1868	0.8886	0.9154	0.9018	0.9529
0.1759	2.07	8700	0.1458	0.9463	0.9643	0.9552	0.9730
0.0494	2.15	9000	0.1440	0.9543	0.9657	0.9599	0.9750
0.0494	2.22	9300	0.1382	0.9646	0.9680	0.9663	0.9752
0.0532	2.29	9600	0.1284	0.9635	0.9712	0.9673	0.9749
0.0532	2.36	9900	0.1495	0.9624	0.9712	0.9668	0.9745
0.0223	2.43	10200	0.1203	0.9600	0.9726	0.9662	0.9757
0.0275	2.5	10500	0.1318	0.9645	0.9694	0.9670	0.9753
0.0275	2.58	10800	0.1224	0.9623	0.9709	0.9666	0.9756
0.026	2.65	11100	0.1241	0.9633	0.9713	0.9673	0.9756

Framework versions

Transformers 4.26.1
Pytorch 1.13.1+cu116
Datasets 2.9.0
Tokenizers 0.13.2