superglue_rte-bert-base-uncased

This model is a fine-tuned version of bert-base-uncased on the super_glue dataset. It achieves the following results on the evaluation set:

• Loss: 1.5070
• Accuracy: 0.6739

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: 4
• eval_batch_size: 32
• 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
0.704	1.0	623	0.6653	0.6159
0.6848	2.0	1246	0.7144	0.4203
0.7083	3.0	1869	0.6922	0.5797
0.7014	4.0	2492	0.7327	0.6232
0.6528	5.0	3115	0.6727	0.6522
0.6471	6.0	3738	0.8413	0.6159
0.5872	7.0	4361	0.8780	0.5507
0.5954	8.0	4984	0.7604	0.6377
0.5566	9.0	5607	0.8578	0.6812
0.5576	10.0	6230	2.0498	0.5362
0.4923	11.0	6853	1.4097	0.6304
0.5688	12.0	7476	1.4146	0.6667
0.433	13.0	8099	1.3354	0.6594
0.4259	14.0	8722	1.3271	0.6957
0.3869	15.0	9345	1.2881	0.6812
0.3641	16.0	9968	1.4485	0.6739
0.3292	17.0	10591	1.3445	0.6739
0.3734	18.0	11214	1.4917	0.6739
0.3227	19.0	11837	1.5281	0.6739
0.3133	20.0	12460	1.5070	0.6739

Framework versions

• Transformers 4.32.1
• Pytorch 1.13.0+cu117
• Datasets 2.15.0
• Tokenizers 0.13.3