20230826022805

This model is a fine-tuned version of bert-large-cased on the super_glue dataset. It achieves the following results on the evaluation set:

• Loss: 0.3895
• Accuracy: 0.72

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: 0.01
• train_batch_size: 16
• eval_batch_size: 8
• seed: 11
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 80.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	25	0.5651	0.49
No log	2.0	50	0.5092	0.6
No log	3.0	75	0.5062	0.63
No log	4.0	100	0.4844	0.64
No log	5.0	125	0.4414	0.58
No log	6.0	150	0.4266	0.58
No log	7.0	175	0.4180	0.62
No log	8.0	200	0.4625	0.64
No log	9.0	225	0.4133	0.61
No log	10.0	250	0.4282	0.63
No log	11.0	275	0.4196	0.57
No log	12.0	300	0.4066	0.6
No log	13.0	325	0.4009	0.62
No log	14.0	350	0.3953	0.63
No log	15.0	375	0.3953	0.61
No log	16.0	400	0.4115	0.64
No log	17.0	425	0.3895	0.6
No log	18.0	450	0.4274	0.63
No log	19.0	475	0.3997	0.64
0.6183	20.0	500	0.3965	0.66
0.6183	21.0	525	0.4352	0.68
0.6183	22.0	550	0.4253	0.69
0.6183	23.0	575	0.3891	0.66
0.6183	24.0	600	0.4324	0.69
0.6183	25.0	625	0.4396	0.73
0.6183	26.0	650	0.4316	0.68
0.6183	27.0	675	0.3951	0.67
0.6183	28.0	700	0.4022	0.68
0.6183	29.0	725	0.4209	0.68
0.6183	30.0	750	0.4500	0.7
0.6183	31.0	775	0.4072	0.71
0.6183	32.0	800	0.4018	0.7
0.6183	33.0	825	0.4191	0.7
0.6183	34.0	850	0.3971	0.71
0.6183	35.0	875	0.3999	0.7
0.6183	36.0	900	0.4025	0.71
0.6183	37.0	925	0.4091	0.71
0.6183	38.0	950	0.4060	0.72
0.6183	39.0	975	0.4416	0.71
0.4716	40.0	1000	0.4041	0.71
0.4716	41.0	1025	0.4100	0.72
0.4716	42.0	1050	0.4042	0.73
0.4716	43.0	1075	0.3744	0.71
0.4716	44.0	1100	0.3827	0.71
0.4716	45.0	1125	0.3941	0.71
0.4716	46.0	1150	0.4305	0.73
0.4716	47.0	1175	0.4008	0.73
0.4716	48.0	1200	0.4027	0.73
0.4716	49.0	1225	0.4024	0.72
0.4716	50.0	1250	0.3938	0.72
0.4716	51.0	1275	0.3843	0.73
0.4716	52.0	1300	0.3911	0.73
0.4716	53.0	1325	0.3855	0.73
0.4716	54.0	1350	0.3934	0.72
0.4716	55.0	1375	0.4029	0.73
0.4716	56.0	1400	0.3878	0.73
0.4716	57.0	1425	0.3839	0.72
0.4716	58.0	1450	0.3943	0.75
0.4716	59.0	1475	0.3984	0.74
0.4121	60.0	1500	0.4064	0.71
0.4121	61.0	1525	0.3871	0.72
0.4121	62.0	1550	0.4141	0.73
0.4121	63.0	1575	0.3850	0.72
0.4121	64.0	1600	0.3933	0.73
0.4121	65.0	1625	0.4055	0.72
0.4121	66.0	1650	0.3852	0.72
0.4121	67.0	1675	0.3952	0.73
0.4121	68.0	1700	0.3874	0.72
0.4121	69.0	1725	0.3999	0.72
0.4121	70.0	1750	0.3956	0.72
0.4121	71.0	1775	0.3918	0.72
0.4121	72.0	1800	0.3859	0.72
0.4121	73.0	1825	0.3926	0.72
0.4121	74.0	1850	0.3897	0.72
0.4121	75.0	1875	0.3859	0.72
0.4121	76.0	1900	0.3849	0.72
0.4121	77.0	1925	0.3856	0.72
0.4121	78.0	1950	0.3902	0.72
0.4121	79.0	1975	0.3904	0.72
0.3881	80.0	2000	0.3895	0.72

Framework versions

• Transformers 4.26.1
• Pytorch 2.0.1+cu118
• Datasets 2.12.0
• Tokenizers 0.13.3