This week was critical in advancing my project, focusing on the fine-tuning of the Whisper model to produce outputs that are more aligned with the needs of the Elderly Care System. I started by experimenting with the Whisper-small model, which provided valuable insights into the fine-tuning process. Following this, I encountered and overcame hardware challenges while fine-tuning the Whisper-large-v3 model, thanks to support from my supervisor.

From Day 18 to Day 20, I dedicated my efforts to sourcing and preparing resources for the fine-tuning process. I explored platforms like YouTube, Bilibili, and other online sources to gather relevant materials. These resources were then carefully processed and formatted to ensure they were suitable for use in fine-tuning the Whisper model.

The week culminated on Day 21 with the first complete fine-tune of the Whisper-large-v3 model using the newly prepared resources. This fine-tune was specifically tailored to the project’s requirements and marked a major achievement, as the model now produces outputs that are well-suited for the Elderly Care System. This successful fine-tuning sets a solid foundation for the continued development and refinement of the project in the coming weeks.