Modelpublic

gokuls/distilbert_add_GLUE_Experiment_logit_kd_qqp_384

sourceHugging Faceapache-2.0updated 4y agoView on Hugging Face

0likes19downloads

Model Card

distilbertaddGLUEExperimentlogitkdqqp_384

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

—Loss: 0.6543
—Accuracy: 0.6542
—F1: 0.1220
—Combined Score: 0.3881

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
—trainbatchsize: 256
—evalbatchsize: 256
—seed: 10
—distributed_type: multi-GPU
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—num_epochs: 50
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Combined Score
0.8184	1.0	1422	0.7776	0.6318	0.0	0.3159
0.7189	2.0	2844	0.6932	0.6333	0.0083	0.3208
0.6368	3.0	4266	0.6602	0.6326	0.0044	0.3185
0.5938	4.0	5688	0.6756	0.6516	0.1109	0.3812
0.5663	5.0	7110	0.6744	0.6398	0.0459	0.3428
0.5478	6.0	8532	0.6685	0.6479	0.0888	0.3684
0.5345	7.0	9954	0.6543	0.6542	0.1220	0.3881
0.5241	8.0	11376	0.6601	0.6469	0.0848	0.3659
0.5161	9.0	12798	0.6774	0.6590	0.1482	0.4036
0.5099	10.0	14220	0.6590	0.6610	0.1594	0.4102
0.5048	11.0	15642	0.6704	0.6529	0.1166	0.3847
0.5001	12.0	17064	0.6791	0.6562	0.1347	0.3955

Framework versions

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