bert-base-cased-finetuned-squad-bs_16

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

• Loss: 1.3851
• EM: 80.2270
• F1: 88.0794

Overview

Language model: bert-base-cased
Language: English
Downstream-task: Extractive QA
Training data: SQuAD
Eval data: SQuAD

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-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
1.0302	1.0	5546	1.0026
0.7716	2.0	11092	0.9711
0.5512	3.0	16638	1.1097
0.3971	4.0	22184	1.2117
0.2999	5.0	27730	1.3851

Framework versions

• Transformers 4.34.0
• Pytorch 1.12.1
• Datasets 2.14.5
• Tokenizers 0.14.1