The Missing Highly-Relevant Reference : Progressive Network Grafting for Few-Shot Knowledge Distillation

[sccbhxc]

We have read your recent paper, "Exploring Diffusion Transformer Designs via Grafting," with great interest and appreciate your insightful contributions to the field.

We noticed a strong similarity between the "Two-Stage Grafting" methodology you proposed and a method we introduced in our earlier work, "Progressive Network Grafting for Few-Shot Knowledge Distillation," published at AAAI 2021.

Our Progressive Network Grafting was also specifically designed to reduce training costs by progressively grafting and training parts of a network. Given that both methods share the same core idea and motivation, we believe our paper constitutes relevant prior work.

We would be grateful if you could consider acknowledging the contribution of our paper in your work. For your convenience, you can find our paper here: https://ojs.aaai.org/index.php/AAAI/article/view/16356/16163.

Thank you for your time and consideration. We look forward to seeing more of your valuable research.

[keshik6]

Thank you for reaching out and sharing your work.

We will acknowledge your grafting-based approach to few-shot knowledge distillation in the next version of our paper.

Thanks!