distilbert_sa_GLUE_Experiment_logit_kd_qqp_96

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: 0.7423
• Accuracy: 0.6329
• F1: 0.0062
• Combined Score: 0.3195

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
0.8963	1.0	1422	0.7832	0.6318	0.0	0.3159
0.7734	2.0	2844	0.7741	0.6318	0.0	0.3159
0.7598	3.0	4266	0.7727	0.6318	0.0	0.3159
0.7474	4.0	5688	0.7675	0.6318	0.0	0.3159
0.7366	5.0	7110	0.7626	0.6318	0.0	0.3159
0.7272	6.0	8532	0.7568	0.6318	0.0	0.3159
0.7177	7.0	9954	0.7539	0.6318	0.0	0.3159
0.7084	8.0	11376	0.7500	0.6318	0.0	0.3159
0.6998	9.0	12798	0.7543	0.6318	0.0	0.3159
0.692	10.0	14220	0.7469	0.6318	0.0	0.3159
0.6846	11.0	15642	0.7481	0.6318	0.0	0.3159
0.6774	12.0	17064	0.7486	0.6318	0.0	0.3159
0.6705	13.0	18486	0.7440	0.6318	0.0	0.3159
0.6648	14.0	19908	0.7464	0.6318	0.0	0.3159
0.659	15.0	21330	0.7430	0.6318	0.0	0.3159
0.6531	16.0	22752	0.7423	0.6329	0.0062	0.3195
0.6479	17.0	24174	0.7452	0.6321	0.0016	0.3169
0.643	18.0	25596	0.7443	0.6354	0.0214	0.3284
0.6387	19.0	27018	0.7431	0.6335	0.0092	0.3213
0.6343	20.0	28440	0.7436	0.6370	0.0318	0.3344
0.6297	21.0	29862	0.7444	0.6362	0.0266	0.3314

Framework versions

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