Fine-tuned GPT2 model on the Quotes-500K dataset. For a given user prompt, it can generate motivational quotes starting with it. The model weights are deployed on Hugging Face Models Hub. Check it out at https://huggingface.co/nandinib1999/quote-generator.
This is the distribution of the total dataset into training, validation and test dataset for the fine-tuning task.
| train | 349796 |
|---|---|
| validation | 99942 |
| test | 49971 |
The model was fine-tuned using the Google Colab GPU for one epoch. The weights of the pre-trained GPT2 model were used as a base.
| Epoch | Perplexity |
|---|---|
| 1 | 15.180 |