add_BERT_24_qqp

This model is a fine-tuned version of gokuls/add_bert_12_layer_model_complete_training_new on the GLUE QQP dataset. It achieves the following results on the evaluation set:

• Loss: 0.4356
• Accuracy: 0.8049
• F1: 0.7302
• Combined Score: 0.7675

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: 128
• eval_batch_size: 128
• 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
0.5487	1.0	2843	0.5164	0.7477	0.6465	0.6971
0.4981	2.0	5686	0.4939	0.7635	0.6487	0.7061
0.4835	3.0	8529	0.4990	0.7568	0.6143	0.6856
0.4719	4.0	11372	0.4912	0.7637	0.6417	0.7027
0.4632	5.0	14215	0.4881	0.7680	0.6619	0.7150
0.4584	6.0	17058	0.4839	0.7679	0.6580	0.7129
0.4425	7.0	19901	0.4774	0.7723	0.6914	0.7319
0.4308	8.0	22744	0.4679	0.7738	0.6650	0.7194
0.4102	9.0	25587	0.4536	0.7873	0.6914	0.7393
0.3909	10.0	28430	0.4512	0.7895	0.7153	0.7524
0.3787	11.0	31273	0.4681	0.7959	0.7134	0.7547
0.3538	12.0	34116	0.4487	0.7981	0.7095	0.7538
0.3313	13.0	36959	0.4356	0.8049	0.7302	0.7675
0.3053	14.0	39802	0.4410	0.8081	0.7448	0.7764
0.2785	15.0	42645	0.4896	0.7942	0.7450	0.7696
0.2516	16.0	45488	0.4969	0.8055	0.7510	0.7782
0.2254	17.0	48331	0.5079	0.8129	0.7535	0.7832
0.2017	18.0	51174	0.5186	0.8113	0.7560	0.7836

Framework versions

• Transformers 4.30.2
• Pytorch 1.14.0a0+410ce96
• Datasets 2.13.0
• Tokenizers 0.13.3