distilbert-swag

This model is a fine-tuned version of distilbert-base-uncased on the swag dataset. It achieves the following results on the evaluation set:

• Loss: 0.9079
• Accuracy: 0.7105

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.9186	1.0	2298	0.7658	0.6950
0.5843	2.0	4597	0.7453	0.7059
0.3548	3.0	6894	0.9079	0.7105

Framework versions

• Transformers 4.31.0
• Pytorch 2.0.1
• Datasets 2.13.1
• Tokenizers 0.13.3