结业必修 | 文本检测实战题第二题遇到的bug，怎么破？

[2021/12/23 10:18:11] root INFO: Initialize indexs of datasets:['/home/aistudio/det_data_lesson_demo/eval.txt']
W1223 10:18:11.152596  1019 device_context.cc:447] Please NOTE: device: 0, GPU Compute Capability: 7.0, Driver API Version: 10.1, Runtime API Version: 10.1
W1223 10:18:11.156044  1019 device_context.cc:465] device: 0, cuDNN Version: 7.6.
[2021/12/23 10:18:15] root WARNING: The shape of model params neck.in2_conv.weight [256, 16, 1, 1] not matched with loaded params last_conv.weight [1280, 480, 1, 1] !
[2021/12/23 10:18:15] root WARNING: The shape of model params neck.in3_conv.weight [256, 24, 1, 1] not matched with loaded params out.weight [1280, 1000] !
[2021/12/23 10:18:15] root WARNING: The shape of model params neck.in4_conv.weight [256, 56, 1, 1] not matched with loaded params out.bias [1000] !
[2021/12/23 10:18:15] root INFO: load pretrain successful from ./pretrain_models/MobileNetV3_large_x0_5_pretrained
[2021/12/23 10:18:15] root INFO: train dataloader has 24 iters
[2021/12/23 10:18:15] root INFO: valid dataloader has 250 iters
[2021/12/23 10:18:15] root INFO: During the training process, after the 200th iteration, an evaluation is run every 100 iterations
[2021/12/23 10:18:15] root INFO: Initialize indexs of datasets:['/home/aistudio/det_data_lesson_demo/train.txt']
Corrupt JPEG data: 18 extraneous bytes before marker 0xc4
Corrupt JPEG data: bad Huffman code