deberta-v3-large__sst2__train-32-1

This model is a fine-tuned version of microsoft/deberta-v3-large on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4201
• Accuracy: 0.8759

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: 4
• eval_batch_size: 4
• seed: 42
• 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
0.7162	1.0	13	0.6832	0.5385
0.6561	2.0	26	0.7270	0.4615
0.4685	3.0	39	1.0674	0.5385
0.2837	4.0	52	1.0841	0.5385
0.1129	5.0	65	0.3502	0.9231
0.0118	6.0	78	0.4829	0.9231
0.0022	7.0	91	0.7430	0.8462
0.0007	8.0	104	0.8219	0.8462
0.0005	9.0	117	0.8787	0.8462
0.0003	10.0	130	0.8713	0.8462
0.0003	11.0	143	0.8473	0.8462
0.0002	12.0	156	0.8482	0.8462
0.0002	13.0	169	0.8494	0.8462
0.0002	14.0	182	0.8638	0.8462
0.0002	15.0	195	0.8492	0.8462

Framework versions

• Transformers 4.15.0
• Pytorch 1.10.2+cu102
• Datasets 1.18.2
• Tokenizers 0.10.3