favs_sort_classification_v2

This model is a fine-tuned version of bert-base-cased on the sort_v2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.1553
• F1: 0.9801
• Roc Auc: 0.9805
• Accuracy: 0.8966

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: 25

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.5589	1.0	21	0.5325	0.4815	0.6585	0.0345
0.4621	2.0	42	0.4465	0.5225	0.6780	0.0
0.4144	3.0	63	0.4131	0.5950	0.7172	0.0345
0.3669	4.0	84	0.3793	0.6167	0.7279	0.0345
0.3524	5.0	105	0.3455	0.6880	0.7689	0.0690
0.2987	6.0	126	0.3086	0.8116	0.8533	0.4138
0.2734	7.0	147	0.2767	0.8392	0.8772	0.5172
0.2532	8.0	168	0.2483	0.8472	0.8837	0.5172
0.2166	9.0	189	0.2285	0.8707	0.9032	0.5862
0.19	10.0	210	0.2012	0.9459	0.9525	0.7586
0.1833	11.0	231	0.1856	0.9530	0.9590	0.7931
0.1751	12.0	252	0.1748	0.9595	0.9610	0.7931
0.173	13.0	273	0.1633	0.9467	0.9569	0.7931
0.16	14.0	294	0.1553	0.9801	0.9805	0.8966
0.1396	15.0	315	0.1503	0.9733	0.9740	0.8621
0.1467	16.0	336	0.1417	0.9737	0.9785	0.8621
0.1271	17.0	357	0.1380	0.9669	0.9720	0.8621
0.1228	18.0	378	0.1346	0.9669	0.9720	0.8621
0.1257	19.0	399	0.1308	0.9801	0.9805	0.8966
0.1156	20.0	420	0.1280	0.9801	0.9805	0.8966
0.1242	21.0	441	0.1250	0.9801	0.9805	0.8966
0.1146	22.0	462	0.1236	0.9801	0.9805	0.8966
0.1262	23.0	483	0.1228	0.9801	0.9805	0.8966
0.1268	24.0	504	0.1227	0.9801	0.9805	0.8966
0.1133	25.0	525	0.1224	0.9801	0.9805	0.8966

Framework versions

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