distilhubert-finetuned-mixed-data

This model is a fine-tuned version of ntu-spml/distilhubert on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8806
• Accuracy: 0.7912
• F1: 0.7772
• Precision: 0.8022
• Recall: 0.7912
• Confusion Matrix: [[59, 1, 1, 2], [20, 35, 22, 0], [2, 7, 68, 0], [2, 0, 0, 54]]

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: 0.0005
• train_batch_size: 128
• eval_batch_size: 128
• seed: 123
• gradient_accumulation_steps: 2
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine_with_restarts
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 40
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall	Confusion Matrix
0.4221	22.2222	100	0.8806	0.7912	0.7772	0.8022	0.7912	[[59, 1, 1, 2], [20, 35, 22, 0], [2, 7, 68, 0], [2, 0, 0, 54]]

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Tokenizers 0.19.1