Thanks for answer. "Don't worry - be happy"
Search found 11 matches
- Sat May 06, 2023 10:23 am
- Forum: Training Support
- Topic: Gape between A and B on training graph
- Replies: 2
- Views: 1298
- Fri May 05, 2023 8:08 am
- Forum: Training Support
- Topic: Gape between A and B on training graph
- Replies: 2
- Views: 1298
Gape between A and B on training graph
Hi to everyone. I use Dfaker model with output 128, and now hit 100k iteration. Each training set contain 1019 faces, batch size set to 32. Everything goes well (by my opinion), but i have strange gap between Face A and Face B on graph. In the beginning they go approx. equally, but then with two jum...
- Tue Apr 04, 2023 4:15 pm
- Forum: Training Support
- Topic: Get error when try to continue training under Linux
- Replies: 1
- Views: 1089
Get error when try to continue training under Linux
I use AMD RX 580 gpu and meet with regular crashes under Windows (i suppose because of DirectML) so i try to install Ubuntu to change DirectML on a ROCM. Change OS solve my problem with training crashes, but now i faced with strange error when try to continue training model. 04/04/2023 18:57:06 INFO...
- Thu Mar 30, 2023 11:27 am
- Forum: General Discussion
- Topic: Adjusted masks during manual adjustment - didn't seem to work with training - error on conversion.
- Replies: 21
- Views: 16554
Re: Adjusted masks during manual adjustment - didn't seem to work with training - error on conversion.
The jawline is entirely unused, so you can probably skip faces where the jawline 'flies away' Don't spend an inordinate amount of time fixing up training data. It should be generally correct, yes, but the model is seeing millions of faces, so a few incorrect alignments won't really hurt anything. I...
- Sat Mar 25, 2023 7:47 am
- Forum: Training Support
- Topic: Mask in training differ from mask in manual tool
- Replies: 6
- Views: 2496
Re: Mask in training differ from mask in manual tool
So, changing Mask type in Settings don't affect on existing model, i must create a new one?
- Fri Mar 24, 2023 7:34 pm
- Forum: Training Support
- Topic: Mask in training differ from mask in manual tool
- Replies: 6
- Views: 2496
- Wed Mar 22, 2023 2:25 pm
- Forum: Training Support
- Topic: Mask in training differ from mask in manual tool
- Replies: 6
- Views: 2496
Re: Mask in training differ from mask in manual tool
Thanks, i missed this moment, and use extended mask by default. Now i changed it to BiseNet-Fp Face, hope this will help.
- Tue Mar 21, 2023 8:24 pm
- Forum: Training Support
- Topic: Training freeze at random iteration - HRESULT failed with 0x887a0005
- Replies: 4
- Views: 3687
Re: Training freeze at random iteration - HRESULT failed with 0x887a0005
Yes, i read that and now my delay set to 60 seconds. I did this before create this thread. As i say early - in my case really helps changing batch size.
- Tue Mar 21, 2023 8:19 pm
- Forum: Training Support
- Topic: Mask in training differ from mask in manual tool
- Replies: 6
- Views: 2496
Mask in training differ from mask in manual tool
While training i see in preview that mask differ from what was on manual tool. I use next gudeline from howto: 1) extract faces from image/video 2) sort extracted faces and remove wrong face detection 3) trune dots and mask using manual tool 4) extract training set from alignment file After that i s...
- Tue Mar 07, 2023 9:24 pm
- Forum: Training Support
- Topic: Training freeze at random iteration - HRESULT failed with 0x887a0005
- Replies: 4
- Views: 3687
Re: Training freeze at random iteration - HRESULT failed with 0x887a0005
Thank you for link. This how to not actually solve the situation, but help to make freeze less friquent by redusing batch size to 4.
- Sat Mar 04, 2023 7:59 pm
- Forum: Training Support
- Topic: Training freeze at random iteration - HRESULT failed with 0x887a0005
- Replies: 4
- Views: 3687
Training freeze at random iteration - HRESULT failed with 0x887a0005
Can't find similar topic by errorcode, so start a new one. Model Original, all settings by default. After some iteration (it may be 36 or 506 but rarely greater than 1000) process freeze. It can be stopped by "Stop" button (with no response on stop signal) or stopped by itself after some t...