VaccinChatSentenceClassifierDutch_fromBERTjeDIAL

This model is a fine-tuned version of Jeska/BertjeWDialDataQA20k on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.8355
• Accuracy: 0.6322

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

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: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
3.4418	1.0	1457	2.3866	0.5406
1.7742	2.0	2914	1.9365	0.6069
1.1313	3.0	4371	1.8355	0.6322

Framework versions

• Transformers 4.13.0.dev0
• Pytorch 1.10.0
• Datasets 1.16.1
• Tokenizers 0.10.3