distilbert-base-uncased_fold_2_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: 1.5810
• F1: 0.7620

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	294	0.5886	0.7239
0.557	2.0	588	0.5085	0.7524
0.557	3.0	882	0.6332	0.7530
0.2456	4.0	1176	0.8749	0.7161
0.2456	5.0	1470	1.0601	0.7371
0.1112	6.0	1764	1.1885	0.7451
0.0484	7.0	2058	1.3027	0.7240
0.0484	8.0	2352	1.4647	0.7259
0.0259	9.0	2646	1.4476	0.7322
0.0259	10.0	2940	1.4826	0.7388
0.0164	11.0	3234	1.5869	0.7333
0.0109	12.0	3528	1.5954	0.7539
0.0109	13.0	3822	1.5810	0.7620
0.0082	14.0	4116	1.7165	0.7335
0.0082	15.0	4410	1.8152	0.7414
0.004	16.0	4704	1.7411	0.7474
0.004	17.0	4998	1.8692	0.7355
0.0034	18.0	5292	1.8727	0.7303
0.0009	19.0	5586	1.9813	0.7305
0.0009	20.0	5880	1.9764	0.7391
0.0012	21.0	6174	2.0170	0.7291
0.0012	22.0	6468	2.0240	0.7391
0.0004	23.0	6762	2.0311	0.7352
0.0014	24.0	7056	2.0174	0.7334
0.0014	25.0	7350	2.0282	0.7381

Framework versions

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