distilbert-base-uncased-lora-text-classification

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: 0.8437
• Accuracy: {'accuracy': 0.881}

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: 0.001
• 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: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	250	0.3423	{'accuracy': 0.886}
0.4235	2.0	500	0.3493	{'accuracy': 0.892}
0.4235	3.0	750	0.5340	{'accuracy': 0.881}
0.207	4.0	1000	0.6471	{'accuracy': 0.868}
0.207	5.0	1250	0.7612	{'accuracy': 0.874}
0.0831	6.0	1500	0.8176	{'accuracy': 0.875}
0.0831	7.0	1750	0.8788	{'accuracy': 0.872}
0.0284	8.0	2000	0.8236	{'accuracy': 0.886}
0.0284	9.0	2250	0.8466	{'accuracy': 0.881}
0.0128	10.0	2500	0.8437	{'accuracy': 0.881}

Framework versions

• PEFT 0.10.1.dev0
• Transformers 4.41.0.dev0
• Pytorch 2.1.0+cpu
• Datasets 2.19.0
• Tokenizers 0.19.1