distilbert-base-uncased_fold_4_ternary

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

• Loss: 1.2981
• F1: 0.7565

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: 16
• eval_batch_size: 16
• 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
No log	1.0	289	0.5588	0.6984
0.5547	2.0	578	0.5283	0.7336
0.5547	3.0	867	0.7038	0.7202
0.2479	4.0	1156	0.8949	0.7284
0.2479	5.0	1445	0.9959	0.7286
0.1181	6.0	1734	1.0663	0.7311
0.0508	7.0	2023	1.2377	0.7054
0.0508	8.0	2312	1.2981	0.7565
0.0185	9.0	2601	1.3532	0.7407
0.0185	10.0	2890	1.5365	0.7333
0.0103	11.0	3179	1.5184	0.7423
0.0103	12.0	3468	1.6009	0.7420
0.0123	13.0	3757	1.6395	0.7402
0.008	14.0	4046	1.6838	0.7429
0.008	15.0	4335	1.6176	0.7490
0.0012	16.0	4624	1.7873	0.7345
0.0012	17.0	4913	1.6761	0.7412
0.0044	18.0	5202	1.7356	0.7417
0.0044	19.0	5491	1.7686	0.7502
0.0045	20.0	5780	1.7668	0.7406
0.0017	21.0	6069	1.8411	0.7381
0.0017	22.0	6358	1.8147	0.7469
0.0012	23.0	6647	1.8028	0.7489
0.0012	24.0	6936	1.8147	0.7453
0.0026	25.0	7225	1.8257	0.7475

Framework versions

• Transformers 4.21.0
• Pytorch 1.12.0+cu113
• Datasets 2.4.0
• Tokenizers 0.12.1