22s-dl-sentiment-1
This model is a fine-tuned version of distilbert-base-uncased on the yelp_review_full dataset. It achieves the following results on the evaluation set:
- Loss: 0.2574
- Accuracy: 0.9542
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: 5
Training results
Framework versions
- Transformers 4.19.4
- Pytorch 1.11.0+cu113
- Datasets 2.3.0
- Tokenizers 0.12.1
- Downloads last month
- 3
Dataset used to train DLochmelis33/22s-dl-sentiment-1
Evaluation results
- Accuracy on yelp_review_fullself-reported0.954