search

kingo233 / FCT-Pytorch

pytorch implementation for The Fully Convolutional Transformer(FCT)
38 stars 6 forks source link

Model weights #7

Open ales-git opened 10 months ago

ales-git commented 10 months ago

Hi kingo233! When running your code with the same parameters and settings on the ACDC dataset using the official train/test split, I achieved a validation dice score of 0.7178 after 120 epochs. However, this result is significantly lower than the performance you reported. Do you have any insights into why this difference might occur? Additionally, could you please share the weights of your trained model? Thank you.

kingo233 commented 10 months ago

Hi ales! Maybe you can change the lr and continue learning? I'm afraid that I have forgotten how I achieve this result. By the way, this work is not worth spending too much time on it because I think it overfit on ACDC.

amirrezarajabi commented 7 months ago

Hi kingo233! When running your code with the same parameters and settings on the ACDC dataset using the official train/test split, I achieved a validation dice score of 0.7178 after 120 epochs. However, this result is significantly lower than the performance you reported. Do you have any insights into why this difference might occur? Additionally, could you please share the weights of your trained model? Thank you.

Hi i hope you are doing well can you please share me your checkpoints and logs? thnaks