distilbert-base-uncased-rvl-cdip

This model is a fine-tuned version of distilbert-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.2055
• Accuracy: 0.7079

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	48	2.4392	0.45
No log	2.0	96	1.5529	0.5632
No log	3.0	144	1.3164	0.6132
No log	4.0	192	1.1269	0.6395
No log	5.0	240	1.0145	0.7
No log	6.0	288	1.0839	0.6816
No log	7.0	336	1.1414	0.6868
No log	8.0	384	1.1220	0.7053
No log	9.0	432	1.1402	0.7105
No log	10.0	480	1.1805	0.7132
0.8154	11.0	528	1.1923	0.7132
0.8154	12.0	576	1.2007	0.7079
0.8154	13.0	624	1.1973	0.7079
0.8154	14.0	672	1.2049	0.7105
0.8154	15.0	720	1.2055	0.7079

Framework versions

• Transformers 4.36.0
• Pytorch 2.1.0+cu118
• Datasets 2.15.0
• Tokenizers 0.15.0