CLIP_prefix_caption icon indicating copy to clipboard operation
CLIP_prefix_caption copied to clipboard
verified publisher icon rmokady

model overfitting issue

Open lilyzhng opened this issue 3 years ago • 5 comments

image Dear authors, We have tested the COCO model with different frames, but outputting the same caption results. It seems to be a overfitting issue. Could you suggest if any parameters need to be specified at inference time, or it could be a data issue that the input frames are similar to some extent?

lilyzhng avatar Nov 22 '22 20:11 lilyzhng

Hi, did you find a solution to this problem?

ArminBaz avatar Feb 02 '23 21:02 ArminBaz

@alchemz Can you show your hyperparameters?

ret7020 avatar Feb 24 '23 14:02 ret7020

Me too. I have the same problem

zzk2021 avatar Mar 03 '23 18:03 zzk2021

It seems like the model isn‘t train completely. The loss at a high level. the outputs are high-frequency words image

zzk2021 avatar Mar 04 '23 09:03 zzk2021

Hi all, unfortunately I haven't find a solution to this.

@ret7020 For your question, we did not perform hyparameter tuning on the model with new datasets, but simply run inference with the existing model.

lilyzhng avatar Mar 24 '23 05:03 lilyzhng