search

xuyanwu / SIDDMs-UFOGen

This is the official git report for SIDDMs in NeurIPS2023 and officially unofficial implementation for UFOGen CVPR2024
11 stars 0 forks source link

Questions Regarding Loss Scale and numtimes Parameter #2

Closed Guanys-dar closed 4 days ago

Guanys-dar commented 4 days ago

Hello, thank you for sharing your excellent work!

I have a few questions regarding the use of your proposed framework:

  1. When obtaining reasonable results, to what magnitude does the loss_c (the mse loss) typically decrease?
  2. Can reasonable results be produced with num_times set to 1? Based on the supplementary materials (Table 9), it seems that num_times = 1 can also yield acceptable results, but I would like further clarification on this.

Thank you very much for your time and assistance!