sa_BERT_no_pretrain_mrpc

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

• Loss: 0.6003
• Accuracy: 0.6814
• F1: 0.7782
• Combined Score: 0.7298

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: 4e-05
• train_batch_size: 96
• eval_batch_size: 96
• seed: 10
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Combined Score
0.6845	1.0	39	0.6307	0.6838	0.8122	0.7480
0.6398	2.0	78	0.6313	0.6838	0.8122	0.7480
0.6384	3.0	117	0.6247	0.6838	0.8122	0.7480
0.6428	4.0	156	0.6467	0.6667	0.7806	0.7237
0.6021	5.0	195	0.6003	0.6814	0.7782	0.7298
0.5125	6.0	234	0.6875	0.6863	0.7874	0.7368
0.3735	7.0	273	0.8672	0.6422	0.7355	0.6888
0.2662	8.0	312	0.9928	0.6765	0.7857	0.7311
0.2247	9.0	351	0.9605	0.6789	0.7798	0.7294
0.1655	10.0	390	1.0684	0.6275	0.7206	0.6740

Framework versions

• Transformers 4.30.2
• Pytorch 1.14.0a0+410ce96
• Datasets 2.12.0
• Tokenizers 0.13.3