Default --data imagenette160 training (fastest)

[glenn-jocher]

Signed-off-by: Glenn Jocher glenn.jocher@ultralytics.com

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Adjustment of default dataset and clarification on epoch argument in training scripts. 🎛️

📊 Key Changes

• Changed the default dataset in classify/train.py from cifar100 to imagenette160.
• Added help description for the epochs argument in both classify/train.py and train.py to indicate it specifies the total training epochs.

🎯 Purpose & Impact

• 🎯 The update to a more suitable default dataset is likely for improved out-of-the-box training performance or to showcase the capabilities of YOLOv5 with a different dataset type.
• 📈 The additional help text for the epochs argument improves usability by making it clear to users what this parameter controls, potentially reducing confusion and errors when configuring training sessions.