./train_erfnet.sh: 5: ./train_erfnet.sh: --resume: not found

[foxkw]

Hi,cardwing! I met such errors when I run train_erfnet.py in ERFNet-CULane-PyTorch .
1) miou /loss/acc didn't change 2)--resume: not found. => loading checkpoint './pretrained/ERFNet_pretrained.tar'

reused param: 302

=> loaded checkpoint 'False' (epoch 16)

but I try to change path to absolute path in train_erfnet.sh, but it still doesn't work do you know why . thanks you so much!

Test: [0/1210] Time 9.200 (9.200) Loss 1.9601 (1.9601) Pixels Acc 0.103 mIoU 0.029 Test: [1/1210] Time 0.161 (4.681) Loss 1.9305 (1.9453) Pixels Acc 0.103 mIoU 0.029 Test: [2/1210] Time 0.161 (3.174) Loss 1.9759 (1.9555) Pixels Acc 0.103 mIoU 0.029 Test: [3/1210] Time 0.161 (2.421) Loss 1.9295 (1.9490) Pixels Acc 0.103 mIoU 0.029 Test: [4/1210] Time 0.182 (1.973) Loss 1.8976 (1.9387) Pixels Acc 0.100 mIoU 0.028 ..... ..... Test: [1208/1210] Time 0.193 (0.228) Loss 1.9116 (1.9595) Pixels Acc 0.106 mIoU 0.031 Test: [1209/1210] Time 0.451 (0.228) Loss 1.9140 (1.9595) Pixels Acc 0.106 mIoU 0.031 Testing Results: Pixels Acc 0.106 mIoU 0.031 (0.0312) Loss 1.95952 ./train_erfnet.sh: 5: ./train_erfnet.sh: --resume: not found

[cardwing]

Can you print the resume path? I think the problem is not difficult to fix.