Tiny_Bert_Cupstone

This model is a fine-tuned version of huawei-noah/TinyBERT_General_4L_312D on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3333
• Accuracy: 0.8550

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: 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: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.524	0.2	500	0.4015	0.8318
0.4268	0.4	1000	0.4274	0.8279
0.39	0.6	1500	0.3743	0.8502
0.3674	0.8	2000	0.3333	0.8550
0.3687	1.0	2500	0.3836	0.8585
0.3489	1.2	3000	0.3927	0.8548
0.3193	1.41	3500	0.3938	0.8669
0.3525	1.61	4000	0.3717	0.8753
0.3327	1.81	4500	0.4589	0.8573
0.3276	2.01	5000	0.3676	0.8791
0.285	2.21	5500	0.4196	0.8811
0.2757	2.41	6000	0.3973	0.8777
0.277	2.61	6500	0.4198	0.8805
0.2834	2.81	7000	0.4955	0.8739
0.338	3.01	7500	0.4383	0.8844
0.2499	3.21	8000	0.4745	0.8785
0.2405	3.41	8500	0.4794	0.8854
0.2648	3.61	9000	0.4576	0.8844
0.2379	3.81	9500	0.4395	0.8886
0.2343	4.01	10000	0.5088	0.8791
0.2011	4.22	10500	0.5272	0.8781
0.2198	4.42	11000	0.5235	0.8765
0.2343	4.62	11500	0.5019	0.8844
0.194	4.82	12000	0.5227	0.8791

Framework versions

• Transformers 4.34.1
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1