复现错误

[luodi789567890]

(LR_conv): Conv2d(50, 50, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) (upsampler): Sequential( (0): Conv2d(50, 48, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1)) (1): PixelShuffle(upscale_factor=4) ) ) ) 372,848 total parameters. 372,848 training parameters. [Epoch 0] Learning rate: 5.00e-4 [1600/16000] [L1: 0.0862] 27.3+4.9s [3200/16000] [L1: 0.0660] 27.4+1.1s [4800/16000] [L1: 0.0567] 27.6+2.0s [6400/16000] [L1: 0.0513] 27.7+0.9s [8000/16000] [L1: 0.0479] 27.7+2.2s [9600/16000] [L1: 0.0455] 27.9+0.8s [11200/16000] [L1: 0.0437] 27.9+0.3s [12800/16000] [L1: 0.0423] 28.0+0.2s [14400/16000] [L1: 0.0412] 28.0+1.2s [16000/16000] [L1: 0.0402] 28.1+0.7s

Evaluation: 0it [00:00, ?it/s] Traceback (most recent call last): File "main.py", line 39, in main() File "main.py", line 34, in main t.test() File "/root/autodl-tmp/src/trainer.py", line 132, in test PSNR_values.append(eval_acc / len(d)) ZeroDivisionError: division by zero

[kekiii]

你好，我遇到了同样的问题，请问解决了吗？

[Bobodesu]

我把测试集放在生成的benchmark文件夹里就对了，我其实也没搞明白原理