bert-base-uncased-Winowhy

This model is a fine-tuned version of bert-base-uncased.

It achieves the following results on the evaluation set:

• Loss: 0.8005
• Accuracy: 0.7118

Model description

For more information on how it was created, check out the following link: https://github.com/DunnBC22/NLP_Projects/blob/main/Multiple%20Choice/Winowhy/Winowhy%20-%20Multiple%20Choice%20Using%20BERT.ipynb

Intended uses & limitations

This model is intended to demonstrate my ability to solve a complex problem using technology.

Training and evaluation data

Dataset Source: https://huggingface.co/datasets/tasksource/bigbench/viewer/winowhy/train

Histogram of Input Lengths

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.7028	1.0	115	0.6916	0.5371
0.6119	2.0	230	0.5572	0.7031
0.4959	3.0	345	0.5328	0.7118
0.4537	4.0	460	0.5829	0.7118
0.2275	5.0	575	0.8005	0.7118

Framework versions

• Transformers 4.26.1
• Pytorch 2.0.1
• Datasets 2.13.1
• Tokenizers 0.13.3