distilbert_sa_GLUE_Experiment_logit_kd_mnli_192

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

• Loss: 0.5326
• Accuracy: 0.5774

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
0.5885	1.0	1534	0.5621	0.5096
0.5572	2.0	3068	0.5481	0.5303
0.5456	3.0	4602	0.5473	0.5347
0.5373	4.0	6136	0.5404	0.5533
0.5299	5.0	7670	0.5371	0.5629
0.5235	6.0	9204	0.5361	0.5671
0.5172	7.0	10738	0.5360	0.5645
0.5114	8.0	12272	0.5391	0.5664
0.5058	9.0	13806	0.5332	0.5839
0.5004	10.0	15340	0.5294	0.5867
0.4951	11.0	16874	0.5284	0.5905
0.4901	12.0	18408	0.5309	0.5892
0.4853	13.0	19942	0.5334	0.5967
0.4804	14.0	21476	0.5344	0.5954
0.4754	15.0	23010	0.5297	0.5987
0.4707	16.0	24544	0.5348	0.5989

Framework versions

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