Week 13: Final Week

Wrapping Up

Fixing operational bugs.
All the codes are stored in https://github.com/Zhongheng-Cheng/gsoc2024-frame-blending and in Gallina home of CWRU HPC, with README.md as the documentation.
Preparing the demonstration video (with explanation).
Will submit the final evaluation form in GSoC dashboard by deadline.

Enhance Model Performance: By the conclusion of this project, newer and more advanced models, such as LLaMA 3, have been released. Additionally, more powerful variants of LLaMA 2, including models with 13 billion and 70 billion parameters, are now available. Future work should consider integrating these advancements to improve the overall performance of the model.
Incorporate Training: Utilizing the Frame Blender workflow and the evaluation data we anticipate collecting, the generation results, along with their corresponding evaluation data, can be fed back into the model. This process will enable the model to learn how to generate more accurate and refined frame blending examples through iterative training.