TFLite Model Maker, how to persist the trained model

I use Colab Pro to train my Object Detection using EfficientDet-Lite with TFLite Model Maker, I let the colab run 24h and export to TFLite model after it done all epoch, but if I want to train for more epoch I need to start again from the beginning. Is there anyway to save the trained model to Google Drive and reload again for further training?
More over, when it is training, if I click “stop” on Colab, the result gone, so what does it mean about this sentence on the Tutorial: “You can look at the validation accuracy during training and stop early to avoid overfitting.” (TensorFlow Lite Model Maker를 사용한 객체 감지)

TFLite Model Maker is still lacking of document so I need your help.

Thank you,

Hi @Yen-Thanh_Le

Check this answer from the TensorFlow team:
https://tensorflow-prod.ospodiscourse.com/t/how-to-stop-and-resume-object-detector-training-object-detection-model-maker/3487

1 Like