bert-base-uncased-Masked_Language_Modeling-Reddit_Comments

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

• Loss: 2.5415

Model description

This is a masked language modeling project.

For more information on how it was created, check out the following link: https://github.com/DunnBC22/NLP_Projects/blob/main/Masked%20Language%20Model/Datasets%20for%20NLP%20-%20Reddit%20Comments/Datasets_for_NLP_MLM.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://www.kaggle.com/datasets/toygarr/datasets-for-natural-language-processing

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 3

Training results

Perplexity: 12.70

Framework versions

• Transformers 4.27.0
• Pytorch 1.13.1+cu116
• Datasets 2.10.1
• Tokenizers 0.13.2