bert-small-finetuned

This model is a fine-tuned version of prajjwal1/bert-small on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.0048
• Accuracy: 0.6038
• F1 Score: 0.6018

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: 86
• eval_batch_size: 86
• 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	Accuracy	F1 Score
1.3167	1.0	18	1.2414	0.4151	0.3857
1.1845	2.0	36	1.1500	0.5148	0.5148
1.0823	3.0	54	1.0743	0.5499	0.5543
0.995	4.0	72	1.0359	0.5553	0.5529
0.9242	5.0	90	1.0195	0.5849	0.5781
0.8742	6.0	108	1.0028	0.5741	0.5758
0.8237	7.0	126	1.0033	0.5930	0.5901
0.7893	8.0	144	0.9967	0.5930	0.5922
0.7332	9.0	162	1.0088	0.5957	0.5924
0.6997	10.0	180	1.0048	0.6038	0.6018
0.6836	11.0	198	1.0120	0.6011	0.5981
0.6571	12.0	216	1.0084	0.5849	0.5864
0.6253	13.0	234	1.0167	0.5903	0.5938
0.5902	14.0	252	1.0184	0.5930	0.5965
0.5766	15.0	270	1.0340	0.5930	0.5925
0.5591	16.0	288	1.0399	0.5930	0.5931
0.5353	17.0	306	1.0364	0.5930	0.5944
0.5205	18.0	324	1.0412	0.5876	0.5889
0.5197	19.0	342	1.0410	0.5849	0.5867
0.5222	20.0	360	1.0418	0.5984	0.5990

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1