mobilebert_sa_GLUE_Experiment_mrpc_256

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

• Loss: 0.6111
• Accuracy: 0.6912
• F1: 0.7948
• Combined Score: 0.7430

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
0.6431	1.0	29	0.6261	0.6838	0.8122	0.7480
0.6296	2.0	58	0.6235	0.6838	0.8122	0.7480
0.6306	3.0	87	0.6237	0.6838	0.8122	0.7480
0.6297	4.0	116	0.6238	0.6838	0.8122	0.7480
0.6276	5.0	145	0.6207	0.6838	0.8122	0.7480
0.6197	6.0	174	0.6213	0.6838	0.8122	0.7480
0.6065	7.0	203	0.6284	0.6912	0.8043	0.7478
0.5258	8.0	232	0.6111	0.6912	0.7948	0.7430
0.4596	9.0	261	0.6506	0.7034	0.8052	0.7543
0.3953	10.0	290	0.7271	0.7034	0.7932	0.7483
0.3426	11.0	319	0.9509	0.6740	0.7542	0.7141
0.2821	12.0	348	1.0021	0.6863	0.7808	0.7335
0.2177	13.0	377	1.0359	0.6691	0.7676	0.7184

Framework versions

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