watsonx.ai

 View Only

 fine-tuning the Granite 3.2 model and encountering overfitting issues

  • Watson studio
Muhammad Tahir Ashraf's profile image
Muhammad Tahir Ashraf posted Sat April 26, 2025 05:02 AM

Working on fine-tuning the Granite 3.2 model for a domain-specific application. While the base model performs well, the fine-tuned version seems to overfit on our dataset. Has anyone experienced similar challenges? What strategies did you employ to mitigate overfitting during fine-tuning?