Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

the MAP of verification suddenly becomes 0 during training #8339

Closed
1 task done
yrn1126 opened this issue Jun 25, 2022 · 7 comments
Closed
1 task done

the MAP of verification suddenly becomes 0 during training #8339

yrn1126 opened this issue Jun 25, 2022 · 7 comments
Labels
question Further information is requested Stale

Comments

@yrn1126
Copy link

yrn1126 commented Jun 25, 2022

Search before asking

Question

微信截图_20220625184517
微信截图_20220625184551
During the training of my own dataset, everything was normal. But at the 80th epoch, the mAP of the verification became 0 suddenly . After 10 epochs, it becomes normal again, but the mAP was much lower than before. I feel confused about this question.

Additional

No response

@yrn1126 yrn1126 added the question Further information is requested label Jun 25, 2022
@josebenitezg
Copy link

Hi @yrn1126 I have the same problem

@glenn-jocher
Copy link
Member

@josebenitezg @yrn1126 your losses are increasing.

@josebenitezg
Copy link

josebenitezg commented Jun 26, 2022

hi @glenn-jocher
Actually, first are decreasing but then are incresing again, and recover again.
I try with yolov5s6 model with same dataset and I don't have this problem.
Captura de Pantalla 2022-06-26 a la(s) 12 51 51
Captura de Pantalla 2022-06-26 a la(s) 12 47 39

@glenn-jocher
Copy link
Member

@josebenitezg you may need to use less aggressive hyps or a longer warmup etc. to increase stability.

@josebenitezg
Copy link

josebenitezg commented Jun 27, 2022

@glenn-jocher I put the default hyps but still same error with large model (even with the last commit)
As you can see here, I train yolov5s6 with no problem (orange one). Same dataset with large suddenly becomes 0

hyperparameters: lr0=0.01, lrf=0.01, momentum=0.937, weight_decay=0.0005, warmup_epochs=3.0, warmup_momentum=0.8, warmup_bias_lr=0.1, box=0.05, cls=0.5, cls_pw=1.0, obj=1.0, obj_pw=1.0, iou_t=0.2, anchor_t=4.0, fl_gamma=0.0, hsv_h=0.015, hsv_s=0.7, hsv_v=0.4, degrees=0.0, translate=0.1, scale=0.5, shear=0.0, perspective=0.0, flipud=0.0, fliplr=0.5, mosaic=1.0, mixup=0.0, copy_paste=0.0
Training on EC2 Instance p3.2xlarge Instance
Captura de Pantalla 2022-06-27 a la(s) 13 13 23
Captura de Pantalla 2022-06-27 a la(s) 13 13 15

@glenn-jocher
Copy link
Member

@josebenitezg perhaps related to warmup bug fix #8356 merged today. You should git pull and train again.

@github-actions
Copy link
Contributor

github-actions bot commented Jul 28, 2022

👋 Hello, this issue has been automatically marked as stale because it has not had recent activity. Please note it will be closed if no further activity occurs.

Access additional YOLOv5 🚀 resources:

Access additional Ultralytics ⚡ resources:

Feel free to inform us of any other issues you discover or feature requests that come to mind in the future. Pull Requests (PRs) are also always welcomed!

Thank you for your contributions to YOLOv5 🚀 and Vision AI ⭐!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
question Further information is requested Stale
Projects
None yet
Development

No branches or pull requests

3 participants