metadata
license: apache-2.0
datasets:
- wikimedia/wikipedia
language:
- en
metrics:
- glue
- accuracy
- f1
Model Card for Model ID
This model detects if you are writing in a format that is more similar to Simple English or English.
Model Details
It was finetuned with the wikiedia datasets for English (20220301.en) and Simple English (20220301.simple). Evaluation data after training:
- 'eval_loss': 0.100
- 'eval_accuracy': 0.982
- 'eval_f1': 0.982
- 'eval_runtime': 10.236
- 'eval_samples_per_second': 58.616
- 'eval_steps_per_second': 7.327
- 'epoch': 10.0
Model Description
- Model type: Text Classification
- Language(s) (NLP): English, Simple English
- License: Apache 2.0
- Finetuned from model [optional]: distilbert-base-uncased