distilbert_sa_GLUE_Experiment_logit_kd_data_aug_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: 0.7029
• Accuracy: 0.6540
• F1: 0.1254
• Combined Score: 0.3897

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
0.8495	1.0	29671	0.7150	0.6333	0.0086	0.3210
0.7654	2.0	59342	0.7273	0.6339	0.0121	0.3230
0.7305	3.0	89013	0.7241	0.6400	0.0479	0.3440
0.7108	4.0	118684	0.7147	0.6381	0.0380	0.3381
0.698	5.0	148355	0.7192	0.6414	0.0564	0.3489
0.6891	6.0	178026	0.7239	0.6357	0.0232	0.3295
0.6823	7.0	207697	0.7141	0.6442	0.0723	0.3583
0.6771	8.0	237368	0.7112	0.6491	0.1004	0.3748
0.6729	9.0	267039	0.7156	0.6494	0.1022	0.3758
0.6694	10.0	296710	0.7185	0.6502	0.1053	0.3777
0.6664	11.0	326381	0.7129	0.6508	0.1085	0.3796
0.6639	12.0	356052	0.7112	0.6508	0.1080	0.3794
0.6617	13.0	385723	0.7105	0.6542	0.1260	0.3901
0.6597	14.0	415394	0.7029	0.6540	0.1254	0.3897
0.658	15.0	445065	0.7094	0.6486	0.0964	0.3725
0.6564	16.0	474736	0.7072	0.6510	0.1084	0.3797
0.655	17.0	504407	0.7049	0.6557	0.1333	0.3945
0.6537	18.0	534078	0.7051	0.6542	0.1269	0.3905
0.6526	19.0	563749	0.7096	0.6601	0.1573	0.4087

Framework versions

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