Week07

Blog Post Reflection

Our group chose this project because we all share a deep interest in Artificial Intelligence. Having previously worked with Hugging Face, we recognized it as one of the hottest platforms of our time. Beyond just curiosity, we saw it as an opportunity to gain hands-on experience that would be invaluable for our future careers in the AI field. We believe that by engaging with a platform at the cutting edge of technology, we can not only refine our technical skills but also better understand industry trends and challenges.

Issue Selection Challenges

  • Limited Understanding:
    Initially, our experience with Hugging Face was mostly confined to downloading and using models. This limited our insight into the platform’s internal workings. With an overwhelming number of repositories available, deciding on a specific direction was quite challenging.

  • Nature of Issues:
    Many of the issues we encountered were simply questions seeking answers, which did not align with our goal of making a meaningful contribution through our project.

  • Complexity and Skill Requirements:
    Some issues required advanced expertise that went beyond our undergraduate knowledge. Consequently, we had to carefully filter and select tasks that matched our current skill level.

Our Current Approach

  • Local Reproduction:
    We plan to start by reproducing GitHub projects on our local machines. This involves setting up the required environments and running test cases to ensure that everything functions correctly.

  • Resource Search:
    Alongside local development, we are actively looking for suitable GPU and computational resources. Given our limited resources, we might focus our efforts on contributing to the inference of smaller models.

  • Step-by-Step Enhancement:
    Our approach is to first ensure that the basic workflows are stable. Once we have a solid foundation, we intend to gradually tackle issues related to smaller models that fall within our capabilities. This methodical progression will help us steadily build our expertise and confidence for more complex challenges in the future.

Written before or on March 7, 2025