favs-filtersort-multilabel-classification-bert-base-cased

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

• Loss: 0.3066
• F1: 0.7429
• Roc Auc: 0.8142
• Accuracy: 0.2

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: 1.5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.7601	1.0	12	0.6966	0.2564	0.4518	0.0
0.6757	2.0	24	0.5629	0.6667	0.7785	0.0
0.5796	3.0	36	0.4652	0.6286	0.7477	0.0
0.5026	4.0	48	0.4161	0.6479	0.7605	0.0
0.4282	5.0	60	0.3830	0.6849	0.7862	0.0
0.4085	6.0	72	0.3658	0.7273	0.7962	0.0
0.3847	7.0	84	0.3538	0.7353	0.8052	0.0
0.3829	8.0	96	0.3457	0.6761	0.7772	0.0
0.3758	9.0	108	0.3409	0.6857	0.7810	0.0
0.3487	10.0	120	0.3327	0.7143	0.7976	0.0
0.3421	11.0	132	0.3268	0.6866	0.7758	0.0
0.3351	12.0	144	0.3183	0.7059	0.7886	0.0
0.3245	13.0	156	0.3149	0.7246	0.8014	0.0
0.3191	14.0	168	0.3087	0.7246	0.8014	0.1
0.3083	15.0	180	0.3066	0.7429	0.8142	0.2
0.3061	16.0	192	0.3062	0.7429	0.8142	0.2
0.2935	17.0	204	0.3017	0.7429	0.8142	0.2
0.2888	18.0	216	0.3009	0.7429	0.8142	0.2
0.297	19.0	228	0.3022	0.7429	0.8142	0.2
0.2868	20.0	240	0.3014	0.7429	0.8142	0.2

Framework versions

• Transformers 4.21.1
• Pytorch 1.12.1
• Datasets 2.4.0
• Tokenizers 0.12.1