distilbert_sa_GLUE_Experiment_logit_kd_data_aug_mrpc_192

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

• Loss: 0.3961
• Accuracy: 0.8848
• F1: 0.9080
• Combined Score: 0.8964

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: 256
• eval_batch_size: 256
• 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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Combined Score
0.4524	1.0	980	0.4024	0.9779	0.9837	0.9808
0.4195	2.0	1960	0.3999	0.9779	0.9837	0.9808
0.416	3.0	2940	0.3984	0.9706	0.9781	0.9743
0.4145	4.0	3920	0.3981	0.9853	0.9892	0.9872
0.4133	5.0	4900	0.3983	0.9926	0.9946	0.9936
0.4128	6.0	5880	0.3982	0.9951	0.9964	0.9958
0.4124	7.0	6860	0.3968	0.9951	0.9964	0.9958
0.4121	8.0	7840	0.3968	0.9951	0.9964	0.9958
0.4118	9.0	8820	0.3969	0.9926	0.9946	0.9936
0.4115	10.0	9800	0.3967	0.9926	0.9946	0.9936
0.4114	11.0	10780	0.3967	0.9902	0.9928	0.9915
0.4113	12.0	11760	0.3970	0.9608	0.9705	0.9656
0.4112	13.0	12740	0.3967	0.9902	0.9928	0.9915
0.4112	14.0	13720	0.3967	0.9926	0.9946	0.9936
0.4111	15.0	14700	0.3966	0.9877	0.9910	0.9894
0.411	16.0	15680	0.3966	0.9779	0.9836	0.9808
0.4109	17.0	16660	0.3965	0.9681	0.9761	0.9721
0.4109	18.0	17640	0.3969	0.9608	0.9705	0.9656
0.4108	19.0	18620	0.3964	0.9804	0.9855	0.9829
0.4107	20.0	19600	0.3966	0.9681	0.9761	0.9721
0.4106	21.0	20580	0.3962	0.9926	0.9946	0.9936
0.4107	22.0	21560	0.3965	0.8627	0.8884	0.8756
0.4105	23.0	22540	0.3962	0.9755	0.9818	0.9786
0.4105	24.0	23520	0.3964	0.9118	0.9310	0.9214
0.4105	25.0	24500	0.3963	0.9167	0.9351	0.9259
0.4104	26.0	25480	0.3962	0.9142	0.9331	0.9236
0.4104	27.0	26460	0.3962	0.9069	0.9269	0.9169
0.4104	28.0	27440	0.3962	0.8701	0.8950	0.8826
0.4104	29.0	28420	0.3962	0.875	0.8994	0.8872
0.4104	30.0	29400	0.3961	0.8848	0.9080	0.8964
0.4103	31.0	30380	0.3961	0.8922	0.9144	0.9033
0.4102	32.0	31360	0.3961	0.8897	0.9123	0.9010
0.4102	33.0	32340	0.3961	0.8971	0.9186	0.9078
0.4102	34.0	33320	0.3961	0.8505	0.8773	0.8639
0.4103	35.0	34300	0.3962	0.8333	0.8612	0.8473

Framework versions

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