https://www.philschmid.de/getting-started-setfit
 |
Outperform OpenAI GPT-3 with SetFit for text-classifiation 2. Create Dataset. We are going to use the ag_news dataset, which a news article classification dataset with 4 classes: World (0), Sports (1), Business (2), Sci/Tech (3).. The test split of the dataset contains 7600 examples, which is will be used to evaluate our model. The train split contains 120000 examples, which is a nice amount of data for fine-tuning a regular model. www.philschmid.de |