albert-base-v2-Shakespear

This model is a fine-tuned version of albert-base-v2 on 2 datasets. The first dataset consists of Shakespeare's poems, and the second consists of the CNN news dataset.

It achieves the following results on the evaluation set:

• Loss: 0.0127
• Accuracy: 0.9989
• F1: 0.9989
• Recall: 0.9989

Model description

This model was created for text classification to determine whether a given text is in Shakespeare's style or not. You can use this model to classify texts or as a validation metric for evaluating Shakespeare-style text generation models. The label 'LABEL_1' indicates that the text is in Shakespeare's style, while 'LABEL_0' means that it is not.

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• 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	F1	Recall
No log	1.0	226	0.0208	0.9967	0.9967	0.9967
No log	2.0	452	0.0107	0.9989	0.9989	0.9989
0.027	3.0	678	0.0113	0.9989	0.9989	0.9989
0.027	4.0	904	0.0001	1.0	1.0	1.0
0.0087	5.0	1130	0.0121	0.9989	0.9989	0.9989
0.0087	6.0	1356	0.0123	0.9989	0.9989	0.9989
0.001	7.0	1582	0.0125	0.9989	0.9989	0.9989
0.001	8.0	1808	0.0126	0.9989	0.9989	0.9989
0.0	9.0	2034	0.0127	0.9989	0.9989	0.9989
0.0	10.0	2260	0.0127	0.9989	0.9989	0.9989

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0