distilbert_add_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: 0.7229
• Accuracy: 0.6349
• F1: 0.0187
• Combined Score: 0.3268

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
0.9266	1.0	1422	0.8016	0.6318	0.0	0.3159
0.79	2.0	2844	0.7941	0.6318	0.0	0.3159
0.7769	3.0	4266	0.7865	0.6318	0.0	0.3159
0.7686	4.0	5688	0.8044	0.6318	0.0	0.3159
0.7604	5.0	7110	0.7942	0.6318	0.0	0.3159
0.7508	6.0	8532	0.8087	0.6318	0.0	0.3159
0.7395	7.0	9954	0.7740	0.6318	0.0	0.3159
0.7283	8.0	11376	0.7776	0.6318	0.0	0.3159
0.7149	9.0	12798	0.7925	0.6318	0.0	0.3159
0.7017	10.0	14220	0.7980	0.6318	0.0	0.3159
0.6888	11.0	15642	0.7555	0.6318	0.0	0.3159
0.6762	12.0	17064	0.7617	0.6318	0.0	0.3159
0.6651	13.0	18486	0.7643	0.6318	0.0	0.3159
0.6547	14.0	19908	0.7432	0.6318	0.0	0.3159
0.6457	15.0	21330	0.7386	0.6318	0.0001	0.3160
0.6364	16.0	22752	0.7638	0.6318	0.0005	0.3162
0.6288	17.0	24174	0.7437	0.6323	0.0034	0.3178
0.6211	18.0	25596	0.7229	0.6349	0.0187	0.3268
0.6151	19.0	27018	0.7449	0.6329	0.0072	0.3201
0.6091	20.0	28440	0.7420	0.6337	0.0121	0.3229
0.6034	21.0	29862	0.7284	0.6339	0.0129	0.3234
0.5986	22.0	31284	0.7301	0.6339	0.0131	0.3235
0.5935	23.0	32706	0.7277	0.6361	0.0254	0.3308

Framework versions

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