distilbert_add_GLUE_Experiment_logit_kd_qqp_192

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: 0.6704
• Accuracy: 0.6417
• F1: 0.0575
• Combined Score: 0.3496

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
0.8489	1.0	1422	0.8086	0.6318	0.0	0.3159
0.776	2.0	2844	0.7797	0.6318	0.0	0.3159
0.7501	3.0	4266	0.7462	0.6318	0.0	0.3159
0.7008	4.0	5688	0.7089	0.6318	0.0	0.3159
0.6547	5.0	7110	0.7026	0.6318	0.0	0.3159
0.6221	6.0	8532	0.6962	0.6338	0.0115	0.3226
0.5981	7.0	9954	0.6812	0.6437	0.0693	0.3565
0.5797	8.0	11376	0.6846	0.6361	0.0261	0.3311
0.565	9.0	12798	0.6835	0.6423	0.0609	0.3516
0.5537	10.0	14220	0.6816	0.6516	0.1130	0.3823
0.5446	11.0	15642	0.6907	0.6392	0.0427	0.3409
0.5368	12.0	17064	0.6788	0.6476	0.0921	0.3699
0.5305	13.0	18486	0.6729	0.6531	0.1216	0.3874
0.525	14.0	19908	0.6704	0.6417	0.0575	0.3496
0.5206	15.0	21330	0.6757	0.6467	0.0873	0.3670
0.5165	16.0	22752	0.6805	0.6481	0.0940	0.3711
0.513	17.0	24174	0.6760	0.6474	0.0901	0.3688
0.5103	18.0	25596	0.6767	0.6505	0.1071	0.3788
0.5074	19.0	27018	0.6798	0.6486	0.0964	0.3725

Framework versions

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