Hi everyone, thanks for reading this issue, hope I can receive your precious comments.
I am trying to train this SSD model with VOC2007 dataset with the vgg_16 pre-trained model, and there was something wrong with my checkpoint, it is so much bigger than the original SSD_300_VGG.ckpt...and I can't figure out the reason.
The original checkpoint only has SSD_300_VGG.ckpt 102,678 kb, but the checkpoint trained by me has about 190,000 kb.
And when I fine-tuning the SSD_300_VGG.ckpt with VOC2007, the new checkpoint just became 308,034 kb...
I am a newbie in Deep Learning and Tensorflow, could you please teach me why there is such a different size between the checkpoints?
Thank you for reading this issue, hope you guys have a nice day.
cjnjuwhy
commented
5 years ago
Hi everyone, thanks for reading this issue, hope I can receive your precious comments.
I am trying to train this SSD model with VOC2007 dataset with the vgg_16 pre-trained model, and there was something wrong with my checkpoint, it is so much bigger than the original SSD_300_VGG.ckpt...and I can't figure out the reason. The original checkpoint only has SSD_300_VGG.ckpt 102,678 kb, but the checkpoint trained by me has about 190,000 kb. And when I fine-tuning the SSD_300_VGG.ckpt with VOC2007, the new checkpoint just became 308,034 kb... I am a newbie in Deep Learning and Tensorflow, could you please teach me why there is such a different size between the checkpoints? Thank you for reading this issue, hope you guys have a nice day.