bert-base-uncased-finetuned-semeval2020-task4a

This model is a fine-tuned version of bert-base-uncased on the ComVE dataset which was part of SemEval 2020 Task 4. It achieves the following results on the test set:

• Loss: 0.2782
• Accuracy: 0.9040

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	344	0.2700	0.8940
0.349	2.0	688	0.2782	0.9040

Framework versions

• Transformers 4.12.3
• Pytorch 1.9.1
• Datasets 1.12.1
• Tokenizers 0.10.3