unsupervised_captioning icon indicating copy to clipboard operation
unsupervised_captioning copied to clipboard

Published 20 hours ago • verified publisher icon fengyang0317

question about the training process

Open sakura2233565548 opened this issue 4 years ago • 4 comments

Hi Yang: I use the code that you release in the github, and I run the train code with the parameter: two GPUs, batch_size 256, lr_rate 0.0005, now training 26000 batch_size, but the found_object is so low, about 0.14. I observe your training process and your found_object is increasing so high after 14k batch_size, so can you have some answers for this symptom? I'd appreciate it if you can help solve this problem!

sakura2233565548 avatar Mar 05 '20 05:03 sakura2233565548

Are you using the MSCOCO dataset? Have you tried the data files in https://drive.google.com/drive/folders/1ol8gLj6hYgluldvdj9XFKm16TCqOr7EE?

fengyang0317 avatar Mar 06 '20 13:03 fengyang0317

Are you using the MSCOCO dataset? Have you tried the data files in https://drive.google.com/drive/folders/1ol8gLj6hYgluldvdj9XFKm16TCqOr7EE?

yes,I have checked the dataset ,mscoco 2014, No problem~So I am confused with this problem.

sakura2233565548 avatar Mar 06 '20 13:03 sakura2233565548

Did you generate the training input by yourself? You may try to use the input files from google drive.

fengyang0317 avatar Mar 08 '20 21:03 fengyang0317

Did you generate the training input by yourself? You may try to use the input files from google drive.

I didn't use the input files that you release on google drive~I use the code to preprocess the image dataset by myself.

sakura2233565548 avatar Mar 09 '20 04:03 sakura2233565548