Comment améliorer les résultats
Bonjour,
D'abord merci pour ton modèle.
Les résultats ne sont pas toujours pertinents… Comment peut on les améliorer ?
Merci
Thank you for your message and for trying out my model!
I'm responding in English since Hugging Face is a global AI community, which allows us to reach a broader audience. I hope you understand.
Regarding the relevance of results, you're right that there's room for improvement. The model's performance depends heavily on the dataset used for fine-tuning the base model. Currently, I'm working on enhancing its performance in several ways:
- Expanding the Dataset: I need more high-quality training examples
- Improving Data Quality: Particularly through real-world evaluations
- Fine-tuning Iterations: Regular model updates based on feedback
As a French teacher, you could make a valuable contribution by adding authentic evaluations to the dataset. Based on my analysis, reaching approximately 2,000 evaluations would significantly improve the model's effectiveness and relevance.
Feel free to reach out if you have any other questions or would like to contribute!
Best regards,
Ronan