Training got killed while applying transfer learning using freeze blocks

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

TLT Version → docker_tag: v3.21.08-py3
Network Type → Yolov4
config file :specfile.txt (2.2 KB)

Hi,

Training part got killed after 1st epoch while applying transfer learning using freeze blocks by adding some more samples in dataset.

First I trained my custom dataset till 100 epoch without using freeze blocks
then the outcome model become as pretrained model for more custom dataset So, I am applying transfer learning using that pretrained model file name as yolov4_resnet18_epoch_100.tlt with freeze blocks : 3 but training got killed after 1st epoch. I am using batch size as 8.

I have also attached the spec file for your reference.

3

Please try with a smaller batch-size.

5

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.