gerskill-bert

This model is a fine-tuned version of bert-base-german-cased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1187
• Hard: {'precision': 0.7063339731285988, 'recall': 0.8070175438596491, 'f1': 0.7533265097236438, 'number': 456}
• Soft: {'precision': 0.7111111111111111, 'recall': 0.7804878048780488, 'f1': 0.7441860465116279, 'number': 82}
• Overall Precision: 0.7070
• Overall Recall: 0.8030
• Overall F1: 0.7520
• Overall Accuracy: 0.9644

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: 2e-05
• 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: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Hard	Soft	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
No log	1.0	178	0.1433	{'precision': 0.5503355704697986, 'recall': 0.7192982456140351, 'f1': 0.6235741444866921, 'number': 456}	{'precision': 0.5137614678899083, 'recall': 0.6829268292682927, 'f1': 0.5863874345549738, 'number': 82}	0.5447	0.7138	0.6179	0.9448
No log	2.0	356	0.1181	{'precision': 0.7031578947368421, 'recall': 0.7324561403508771, 'f1': 0.7175080558539205, 'number': 456}	{'precision': 0.6585365853658537, 'recall': 0.6585365853658537, 'f1': 0.6585365853658537, 'number': 82}	0.6966	0.7212	0.7087	0.9544
0.1645	3.0	534	0.1079	{'precision': 0.6605839416058394, 'recall': 0.793859649122807, 'f1': 0.7211155378486056, 'number': 456}	{'precision': 0.6966292134831461, 'recall': 0.7560975609756098, 'f1': 0.7251461988304092, 'number': 82}	0.6656	0.7881	0.7217	0.9603
0.1645	4.0	712	0.1146	{'precision': 0.7030651340996169, 'recall': 0.8048245614035088, 'f1': 0.7505112474437627, 'number': 456}	{'precision': 0.6739130434782609, 'recall': 0.7560975609756098, 'f1': 0.7126436781609194, 'number': 82}	0.6987	0.7974	0.7448	0.9621
0.1645	5.0	890	0.1187	{'precision': 0.7063339731285988, 'recall': 0.8070175438596491, 'f1': 0.7533265097236438, 'number': 456}	{'precision': 0.7111111111111111, 'recall': 0.7804878048780488, 'f1': 0.7441860465116279, 'number': 82}	0.7070	0.8030	0.7520	0.9644

Framework versions

• Transformers 4.38.1
• Pytorch 2.1.2+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2