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: 0.898124
• Accuracy: {'accuracy': 0.893}

Model description

Using LoRA to fine-tune distilbert/distilbert-base-uncased to classify movie reviews

Training and evaluation data

https://huggingface.co/datasets/stanfordnlp/imdb

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: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	250	0.7278	{'accuracy': 0.833}
0.358	2.0	500	0.6268	{'accuracy': 0.852}
0.358	3.0	750	0.6568	{'accuracy': 0.872}
0.1873	4.0	1000	0.7663	{'accuracy': 0.883}
0.1873	5.0	1250	0.7704	{'accuracy': 0.877}
0.0437	6.0	1500	0.8981	{'accuracy': 0.893}
0.0437	7.0	1750	0.9872	{'accuracy': 0.886}
0.0148	8.0	2000	1.0022	{'accuracy': 0.888}
0.0148	9.0	2250	1.0471	{'accuracy': 0.892}
0.0006	10.0	2500	1.0335	{'accuracy': 0.889}

Framework versions

• PEFT 0.14.0
• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0