I obtain this error : Please help , Thanks
So you are training on 32x32 images?
No, I am training o 416*416 images
So, how is it possible that the output of first pooling layer is 64x16x16?
For such input images, the last layer of pooling (before
nn.Flatten) outputs tensors of size 256x52x52.
It also happens that 256x52x52 tensor contains 692 224 elements - the number from the error message.
Please, i don’t understang, So, I train on 416 * 416
I begin by the first pooling is 832 * 208 * 208
the second pooling is 1664 * 104 * 104
- 1st: 64x208x208
- 2nd: 128x104x104
- 3rd: 256x52x52 (basically what I said earlier)
No idea where the 832 and 1664 values are from.
You need to fix the linear layer after flattening because it accepts incorrect number of elements.
i amso happy