Does --resume work the same for yolo11 as yolo5? Do i need to adjust epochs when resuming?

[quitmeyer]

If i run a training session intended for 100 epochs, and then it stops after 20 for some reason. Then if i run the script again with the --resume flag, do i need to change the script to run for only 80 epochs? or does it rememeber it was originally trying to hit 100?

[UltralyticsAssistant]

👋 Hello @quitmeyer, thank you for your interest in Ultralytics 🚀! We recommend visiting the Ultralytics Docs for new users where you can find many helpful resources on usage, including guidance for both Python and CLI examples to address some common questions.

If your query is related to a potential 🐛 Bug Report, please provide a minimum reproducible example to better assist in debugging the issue.

For questions related to custom training or usage scenarios, such as your question about the --resume flag, please provide additional details such as your training script and logs. Meanwhile, please check out our Training Tips for best practices.

An Ultralytics engineer will be along shortly to provide more detailed assistance. In the meantime, consider joining our vibrant community discussions where you can connect with others on Discord 🎧, Discourse, or our Subreddit.

Upgrade

Ensure you are using the latest ultralytics package with all requirements updated. Please confirm this in a Python>=3.8 environment with PyTorch>=1.8:

pip install -U ultralytics

Environments

You can run YOLO in several up-to-date verified environments with all dependencies pre-installed, including CUDA and CUDNN:

• Notebooks with free GPU:
• Google Cloud Deep Learning VM. See GCP Quickstart Guide
• Amazon Deep Learning AMI. See AWS Quickstart Guide
• Docker Image. See Docker Quickstart Guide

Status

Check the status badge above to see if all Ultralytics CI tests are currently passing. This ensures the stability and functionality across all YOLO Modes and Tasks.

[glenn-jocher]

@quitmeyer when using the --resume flag with YOLO11, the training will automatically continue from where it left off, aiming to complete the originally intended 100 epochs without needing to adjust the epoch count.

[tirtab]

If in hyperparameter tuning, something similar might happen, how do you resume it?

[Y-T-G]

You can't resume tuning training.