san_2409_1325

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.0992
• F1: 0.7727

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• 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	F1
No log	0.91	5	1.9727	0.1939
No log	1.91	10	1.5642	0.3535
No log	2.91	15	1.2698	0.6818
No log	3.91	20	1.3642	0.6429
No log	4.91	25	1.3411	0.6818
No log	5.91	30	1.2627	0.6818
No log	6.91	35	1.1269	0.7727
No log	7.91	40	1.0719	0.7727
No log	8.91	45	1.0567	0.7727
No log	9.91	50	1.1256	0.7727
No log	10.91	55	0.7085	0.7727
No log	11.91	60	0.9290	0.7727
No log	12.91	65	1.0355	0.7727
No log	13.91	70	1.0866	0.7727
No log	14.91	75	1.0992	0.7727

Framework versions

• Transformers 4.22.1
• Pytorch 1.12.1+cu113
• Datasets 2.5.1
• Tokenizers 0.12.1