bert_uncased_L-2_H-128_A-2_rte

This model is a fine-tuned version of google/bert_uncased_L-2_H-128_A-2 on the GLUE RTE dataset. It achieves the following results on the evaluation set:

• Loss: 0.6606
• Accuracy: 0.6029

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: 5e-05
• train_batch_size: 256
• eval_batch_size: 256
• seed: 10
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6945	1.0	10	0.6932	0.4874
0.6914	2.0	20	0.6897	0.5271
0.6866	3.0	30	0.6866	0.5451
0.6812	4.0	40	0.6829	0.5415
0.676	5.0	50	0.6797	0.5668
0.6666	6.0	60	0.6769	0.5596
0.6596	7.0	70	0.6736	0.5740
0.6485	8.0	80	0.6693	0.5668
0.6347	9.0	90	0.6639	0.5884
0.6186	10.0	100	0.6625	0.6173
0.6026	11.0	110	0.6606	0.6029
0.5894	12.0	120	0.6638	0.6101
0.5841	13.0	130	0.6629	0.6065
0.566	14.0	140	0.6657	0.6101
0.5529	15.0	150	0.6673	0.5776
0.5312	16.0	160	0.6718	0.5812

Framework versions

• Transformers 4.46.3
• Pytorch 2.2.1+cu118
• Datasets 2.17.0
• Tokenizers 0.20.3