deberta-v3-large__sst2__train-16-1

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: 0.6804
• Accuracy: 0.5497

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
0.7086	1.0	7	0.7176	0.2857
0.6897	2.0	14	0.7057	0.2857
0.6491	3.0	21	0.6582	0.8571
0.567	4.0	28	0.4480	0.8571
0.4304	5.0	35	0.5465	0.7143
0.0684	6.0	42	0.5408	0.8571
0.0339	7.0	49	0.6501	0.8571
0.0082	8.0	56	0.9152	0.8571
0.0067	9.0	63	2.5162	0.5714
0.0045	10.0	70	1.1136	0.8571
0.0012	11.0	77	1.1668	0.8571
0.0007	12.0	84	1.2071	0.8571
0.0005	13.0	91	1.2310	0.8571
0.0006	14.0	98	1.2476	0.8571

Framework versions

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