I a a high school student and realised I just improved the PNSR score of Uformer-B on the SIDD de-noising and Gopro de-blurring datasets by about 1db. The same change to training dehazeformer. At least from the first 15 epochs of each training (I am planning on renting an a100 for full training soon).

Is it worth publishing after I have fully trained it or should I further try and further improve it? Also, what does the process of actually publishing look like?

I have reached out to local universities but I was ghosted.