Train so slow with single 3060TI ,minumun_number_of_images=3 can change more bigger?

[huangpan2507]

Train so slow, I have trained almost 4 days, the iter is just 103999, can I change the parameter minumun_number_of_images bigger?
If i change the parameter , do I need to change a lot code in many position?

[huangpan2507]

minumun_number_of_images from 3 to 6 , Still to slow, need about 6 day to train.

[bismex]

Unfortunately, the number of images for training on the A Large-Scale ReID benchmark exceeds 100,000, which takes a long time to train. We recommend that you do a cross-domain re-id experiment to simply verify your experiment.

In my case, it took me about 2 days to experiment with one 1080ti. For faster results, experiment with larger memory GPUs or use multiple GPUs. However, this code does not yet support multiple GPUs.

[huangpan2507]

Unfortunately, the number of images for training on the A Large-Scale ReID benchmark exceeds 100,000, which takes a long time to train. We recommend that you do a cross-domain re-id experiment to simply verify your experiment.

In my case, it took me about 2 days to experiment with one 1080ti. For faster results, experiment with larger memory GPUs or use multiple GPUs. However, this code does not yet support multiple GPUs.

I used one 3060,with 12G. I use the code python3 ./tools/train_net.py --config-file ./configs/Sample/DG-mobilenet.yml #and it train from scratch, it take me almost 6 days

[bismex]

I'm not sure where the bottleneck is. If you want to get results faster, please stop learning early or experiment on a smaller benchmark.