sguarnaccio/le_signatory

training

This model is a fine-tuned version of nlpaueb/bert-base-uncased-contracts on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0369
• Precision: 0.8095
• Recall: 0.8293
• F1: 0.8193
• Accuracy: 0.9881

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: 5
• eval_batch_size: 5
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	11	0.4084	0.0	0.0	0.0	0.8828
No log	2.0	22	0.2084	0.025	0.0122	0.0164	0.9172
No log	3.0	33	0.1223	0.4615	0.3659	0.4082	0.9488
No log	4.0	44	0.1054	0.5417	0.4756	0.5065	0.9572
No log	5.0	55	0.0662	0.6279	0.6585	0.6429	0.9744
No log	6.0	66	0.0758	0.6104	0.5732	0.5912	0.9723
No log	7.0	77	0.0424	0.7619	0.7805	0.7711	0.9842
No log	8.0	88	0.0377	0.7791	0.8171	0.7976	0.9867
No log	9.0	99	0.0419	0.8235	0.8537	0.8383	0.9881
No log	10.0	110	0.0378	0.8214	0.8415	0.8313	0.9874
No log	11.0	121	0.0367	0.8	0.8293	0.8144	0.9874
No log	12.0	132	0.0390	0.8313	0.8415	0.8364	0.9867
No log	13.0	143	0.0365	0.8095	0.8293	0.8193	0.9884
No log	14.0	154	0.0365	0.8214	0.8415	0.8313	0.9884
No log	15.0	165	0.0369	0.8095	0.8293	0.8193	0.9881

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2