distilbert-base-uncased-finetuned-sst2

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

• Loss: 0.2842
• Accuracy: 0.9071

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.02	100	0.3316	0.8624
No log	0.05	200	0.3357	0.8612
No log	0.07	300	0.3996	0.8383
No log	0.1	400	0.3012	0.8716
0.3421	0.12	500	0.3227	0.8693
0.3421	0.14	600	0.3643	0.8727
0.3421	0.17	700	0.2734	0.8853
0.3421	0.19	800	0.3077	0.8945
0.3421	0.21	900	0.2709	0.9002
0.2705	0.24	1000	0.2737	0.8899
0.2705	0.26	1100	0.3079	0.8979
0.2705	0.29	1200	0.2713	0.8968
0.2705	0.31	1300	0.2505	0.8933
0.2705	0.33	1400	0.2932	0.8922
0.239	0.36	1500	0.2842	0.9071
0.239	0.38	1600	0.2509	0.9014
0.239	0.4	1700	0.2819	0.8853
0.239	0.43	1800	0.2515	0.8956

Framework versions

• Transformers 4.25.1
• Pytorch 1.12.1+cu113
• Datasets 2.7.1
• Tokenizers 0.13.2