bert-nlp-project-ft-google
This model is a fine-tuned version of jestemleon/bert-nlp-project-google on the steciuk/google dataset. It achieves the following results on the evaluation set:
- Loss: 0.3255
- Accuracy: 0.9105
- F1: 0.9174
and flowing results on the testing set:
- Accuracy: 0.9115
- F1: 0.9180
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: 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: 3
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
---|---|---|---|---|---|
0.3506 | 0.37 | 196 | 0.2922 | 0.8876 | 0.9017 |
0.2724 | 0.75 | 392 | 0.2456 | 0.9038 | 0.9099 |
0.2299 | 1.12 | 588 | 0.2781 | 0.9124 | 0.9192 |
0.2009 | 1.49 | 784 | 0.2934 | 0.8981 | 0.9016 |
0.182 | 1.86 | 980 | 0.2854 | 0.9095 | 0.9164 |
0.1569 | 2.24 | 1176 | 0.2932 | 0.9086 | 0.9150 |
0.118 | 2.61 | 1372 | 0.3258 | 0.9067 | 0.9139 |
0.1188 | 2.98 | 1568 | 0.3255 | 0.9105 | 0.9174 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2
- Downloads last month
- 7
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.