mobilebert_sa_GLUE_Experiment_logit_kd_data_aug_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: 0.1267
• Accuracy: 0.9926
• F1: 0.9947
• Combined Score: 0.9936

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
0.3017	1.0	1959	0.2241	0.9608	0.9713	0.9661
0.233	2.0	3918	0.2357	0.9828	0.9876	0.9852
0.2241	3.0	5877	0.1908	0.9706	0.9786	0.9746
0.2189	4.0	7836	0.1863	0.9755	0.9824	0.9789
0.2149	5.0	9795	0.1868	0.9804	0.9858	0.9831
0.211	6.0	11754	0.1735	0.9804	0.9859	0.9831
0.2073	7.0	13713	0.1875	0.9828	0.9876	0.9852
0.204	8.0	15672	0.1690	0.9853	0.9894	0.9873
0.2014	9.0	17631	0.1597	0.9853	0.9893	0.9873
0.1992	10.0	19590	0.1604	0.9877	0.9911	0.9894
0.1975	11.0	21549	0.1563	0.9853	0.9894	0.9873
0.1959	12.0	23508	0.1518	0.9853	0.9894	0.9873
0.1948	13.0	25467	0.1429	0.9902	0.9929	0.9915
0.1937	14.0	27426	0.1484	0.9853	0.9894	0.9873
0.1928	15.0	29385	0.1527	0.9804	0.9856	0.9830
0.1919	16.0	31344	0.1433	0.9926	0.9947	0.9936
0.1913	17.0	33303	0.1445	0.9902	0.9929	0.9915
0.1905	18.0	35262	0.1407	0.9926	0.9947	0.9936
0.1899	19.0	37221	0.1402	0.9926	0.9947	0.9936
0.1892	20.0	39180	0.1387	0.9926	0.9947	0.9936
0.1887	21.0	41139	0.1384	0.9926	0.9947	0.9936
0.1882	22.0	43098	0.1430	0.9951	0.9964	0.9958
0.1877	23.0	45057	0.1384	0.9951	0.9964	0.9958
0.1871	24.0	47016	0.1398	0.9951	0.9964	0.9958
0.1867	25.0	48975	0.1336	0.9926	0.9947	0.9936
0.1863	26.0	50934	0.1368	0.9951	0.9964	0.9958
0.1859	27.0	52893	0.1337	0.9951	0.9964	0.9958
0.1855	28.0	54852	0.1352	0.9926	0.9947	0.9936
0.1851	29.0	56811	0.1314	0.9951	0.9964	0.9958
0.1847	30.0	58770	0.1333	0.9951	0.9964	0.9958
0.1844	31.0	60729	0.1368	0.9951	0.9964	0.9958
0.184	32.0	62688	0.1310	0.9951	0.9964	0.9958
0.1837	33.0	64647	0.1321	0.9951	0.9964	0.9958
0.1834	34.0	66606	0.1302	0.9926	0.9947	0.9936
0.183	35.0	68565	0.1320	0.9951	0.9964	0.9958
0.1827	36.0	70524	0.1303	0.9951	0.9964	0.9958
0.1825	37.0	72483	0.1273	0.9951	0.9964	0.9958
0.1822	38.0	74442	0.1293	0.9951	0.9964	0.9958
0.1819	39.0	76401	0.1296	0.9951	0.9964	0.9958
0.1817	40.0	78360	0.1305	0.9926	0.9947	0.9936
0.1814	41.0	80319	0.1267	0.9926	0.9947	0.9936
0.1812	42.0	82278	0.1267	0.9951	0.9964	0.9958
0.1809	43.0	84237	0.1278	0.9902	0.9929	0.9915
0.1807	44.0	86196	0.1293	0.9951	0.9964	0.9958
0.1805	45.0	88155	0.1269	0.9951	0.9964	0.9958
0.1803	46.0	90114	0.1284	0.9951	0.9964	0.9958

Framework versions

• Transformers 4.26.0
• Pytorch 1.14.0a0+410ce96
• Datasets 2.9.0
• Tokenizers 0.13.2