superglue_rte-t5-base

This model is a fine-tuned version of t5-base on the super_glue dataset. It achieves the following results on the evaluation set:

• Loss: 1.8826
• Accuracy: 0.8406

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
0.7037	1.0	623	0.6646	0.5797
0.6448	2.0	1246	0.5461	0.7899
0.4943	3.0	1869	0.8069	0.7536
0.3854	4.0	2492	1.2553	0.8188
0.1244	5.0	3115	1.4887	0.7826
0.0836	6.0	3738	1.7422	0.7681
0.0672	7.0	4361	1.7002	0.8116
0.0449	8.0	4984	1.9237	0.7971
0.0246	9.0	5607	1.7064	0.7899
0.0239	10.0	6230	1.4433	0.8551
0.0233	11.0	6853	2.1623	0.7754
0.0348	12.0	7476	2.2059	0.7754
0.0268	13.0	8099	1.9322	0.8261
0.0076	14.0	8722	2.5687	0.7464
0.0117	15.0	9345	2.3024	0.7899
0.0129	16.0	9968	2.0848	0.7971
0.0206	17.0	10591	1.9453	0.8333
0.0162	18.0	11214	2.1232	0.7971
0.0132	19.0	11837	1.9754	0.8406
0.0098	20.0	12460	1.8826	0.8406

Framework versions

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