use check_point

[lelnazrezaeel]

Dear cwhgn,

I hope this message finds you well. I've been working on training a model for Market1501, but due to hardware limitations, I can't train for 120 epochs in one go. Instead, I'd like to train gradually, saving checkpoints for every 10 epochs, and then use the last checkpoint as a pretrained model to continue training until I reach 120 epochs.

Currently, I've trained the model for 40 epochs, and my Colab runtime has expired. I'm attempting to use the transformer_40.pth checkpoint as a replacement for the pretrained model (swin_base.pth) using the following command:

!CUDA_VISIBLE_DEVICES=0 python train.py --config_file '/content/drive/MyDrive/SoldierMarket1501/SOLIDER-REID/configs/market/swin_base.yml' MODEL.PRETRAIN_CHOICE 'self' MODEL.PRETRAIN_PATH '/content/drive/MyDrive/SoldierMarket1501/SOLIDER-REID/log/market/swin_base/transformer_40.pth' OUTPUT_DIR './log/market/swin_base' SOLVER.BASE_LR 0.0002 SOLVER.OPTIMIZER_NAME 'SGD' MODEL.SEMANTIC_WEIGHT 0.2
However, I encountered the following error.

I would greatly appreciate your assistance in resolving this issue. If there are any modifications needed in the command or if there's additional information you require, please let me know.

Thank you in advance for your help.

Best regards,
Elnaz

[Eliastoomme]

Hello, I know it's been a long time but did you solve it? I'm currently facing the same issue.