distilbert_sa_GLUE_Experiment_logit_kd_mrpc_384

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

• Loss: 0.5217
• Accuracy: 0.3260
• F1: 0.0351
• Combined Score: 0.1805

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
0.5343	1.0	15	0.5288	0.3162	0.0	0.1581
0.5306	2.0	30	0.5289	0.3162	0.0	0.1581
0.5294	3.0	45	0.5281	0.3162	0.0	0.1581
0.5277	4.0	60	0.5269	0.3162	0.0	0.1581
0.518	5.0	75	0.5217	0.3260	0.0351	0.1805
0.5035	6.0	90	0.5230	0.3971	0.2635	0.3303
0.4866	7.0	105	0.5301	0.3652	0.1618	0.2635
0.4624	8.0	120	0.5491	0.5147	0.5123	0.5135
0.4424	9.0	135	0.5479	0.5245	0.5530	0.5388
0.4295	10.0	150	0.5660	0.5392	0.5766	0.5579

Framework versions

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