Search found 3 matches

by EXHYST
Wed Nov 13, 2019 2:43 pm
Forum: Training Discussion
Topic: 256 High Batch Size Progress
Replies: 2
Views: 5333

256 High Batch Size Progress

Hey I have been trainning at the max batch size of 256 for more than 1 day. Did not have any crashes. But I read that " there are slight quality degradations at very large batch sizes".. So should I reduce it? And reduce it to what? https://i.ibb.co/W5pbkLm/Untitled.png After about 60hours...
by EXHYST
Sun Nov 10, 2019 12:34 am
Forum: Training Support
Topic: CRITICAL unexpected crash has occurred
Replies: 2
Views: 3666

CRITICAL unexpected crash has occurred

Hello, I keep getting this error every like 20m of Training at this moment. I attached the error crash report. Thanks for the help. 11/09/2019 23:57:24 MainProcess training_0 _base save_models INFO [Saved models] - Average since last save: total_loss_A: 0.04119, total_loss_B: 0.03918 11/10/2019 00:0...
by EXHYST
Sat Nov 09, 2019 9:59 am
Forum: Training Support
Topic: 50% of my GPU.
Replies: 1
Views: 2728

50% of my GPU.

Hi.

My GPU is a GTX 1080, 16RAM, CPU i7 4790k .

I put batch size to the maximum, and it seems I am only using 50% of my GPU. Can I do something to use more of it so it trains faster?

Image

Thanks and sorry if this is a stupid question, I'm not that tech savy.