mobilebert_sa_GLUE_Experiment_qqp_256

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

• Loss: 0.4349
• Accuracy: 0.7976
• F1: 0.7297
• Combined Score: 0.7637

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: 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
0.526	1.0	2843	0.5088	0.7492	0.6674	0.7083
0.4762	2.0	5686	0.4782	0.7695	0.6583	0.7139
0.4438	3.0	8529	0.4532	0.7847	0.6829	0.7338
0.4161	4.0	11372	0.4602	0.7869	0.7135	0.7502
0.3968	5.0	14215	0.4395	0.7955	0.7212	0.7583
0.3815	6.0	17058	0.4392	0.7985	0.7190	0.7587
0.3659	7.0	19901	0.4349	0.7976	0.7297	0.7637
0.352	8.0	22744	0.4419	0.8005	0.7300	0.7652
0.3399	9.0	25587	0.4454	0.7998	0.7317	0.7658
0.327	10.0	28430	0.4614	0.7995	0.7359	0.7677
0.3157	11.0	31273	0.4733	0.8000	0.7246	0.7623
0.3041	12.0	34116	0.4738	0.8041	0.7283	0.7662

Framework versions

• Transformers 4.26.0
• Pytorch 1.14.0a0+410ce96
• Datasets 2.8.0
• Tokenizers 0.13.2