distilbert_sa_GLUE_Experiment_logit_kd_data_aug_sst2

This model is a fine-tuned version of distilbert-base-uncased on the GLUE SST2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4215
• Accuracy: 0.8635

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
0.434	1.0	4374	0.6108	0.8222
0.2568	2.0	8748	0.5465	0.8314
0.207	3.0	13122	0.5067	0.8463
0.1819	4.0	17496	0.4734	0.8544
0.1666	5.0	21870	0.4785	0.8578
0.1563	6.0	26244	0.4539	0.8589
0.1492	7.0	30618	0.4600	0.8589
0.1436	8.0	34992	0.4445	0.8647
0.1394	9.0	39366	0.4270	0.8727
0.1361	10.0	43740	0.4524	0.8601
0.1334	11.0	48114	0.4244	0.8693
0.1313	12.0	52488	0.4469	0.8635
0.1292	13.0	56862	0.4556	0.8498
0.1277	14.0	61236	0.4257	0.8635
0.1263	15.0	65610	0.4392	0.8567
0.1251	16.0	69984	0.4215	0.8635
0.124	17.0	74358	0.4289	0.8578
0.123	18.0	78732	0.4448	0.8601
0.1222	19.0	83106	0.4562	0.8555
0.1214	20.0	87480	0.4377	0.8544
0.1207	21.0	91854	0.4563	0.8555

Framework versions

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