distilbert_add_GLUE_Experiment_logit_kd_cola_384

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

• Loss: 0.6805
• Matthews Correlation: 0.1134

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	Matthews Correlation
0.8052	1.0	34	0.6842	0.0
0.7982	2.0	68	0.6842	0.0
0.7961	3.0	102	0.6841	0.0
0.7965	4.0	136	0.6846	0.0
0.799	5.0	170	0.6841	0.0
0.7956	6.0	204	0.6840	0.0
0.798	7.0	238	0.6860	0.0
0.7984	8.0	272	0.6839	0.0
0.7962	9.0	306	0.6875	0.0
0.797	10.0	340	0.6834	0.0
0.7951	11.0	374	0.6813	0.0
0.7771	12.0	408	0.6849	0.0257
0.7055	13.0	442	0.7093	0.0764
0.6664	14.0	476	0.6957	0.1007
0.654	15.0	510	0.6805	0.1134
0.6345	16.0	544	0.6966	0.1176
0.6176	17.0	578	0.7439	0.1155
0.6156	18.0	612	0.7178	0.1406
0.5938	19.0	646	0.7192	0.1212
0.582	20.0	680	0.7765	0.1506

Framework versions

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