Errors in the testing process

[navy63]

The training is normal. The following problems are encountered in the testing process.
wrong with vl_nnbnorm
vl::impl::dispatch_cudnn<C, CU>::operator(): BatchNormForwardWithMomentCudnn::operator(): cuDNN error [cudnn:
"d:\matlab\r2018b\bin\matconvnet-1.0-beta25\matlab\src\bits\nnbnorm_cudnn.cu":155 (CUDNN_STATUS_BAD_PARAM)]

[navy63]

train: epoch 499 : 16/ 20: loss: 92.8000
train: epoch 499 : 17/ 20: loss: 107.7473
train: epoch 499 : 18/ 20: loss: 99.2222
train: epoch 499 : 19/ 20: loss: 92.0033
train: epoch 499 : 20/ 20: loss: 87.1613
train: epoch 500 : 1/ 20: loss: 92.2146
train: epoch 500 : 2/ 20: loss: 94.4258
train: epoch 500 : 3/ 20: loss: 91.5383
train: epoch 500 : 4/ 20: loss: 88.9754
train: epoch 500 : 5/ 20: loss: 120.8435
train: epoch 500 : 6/ 20: loss: 87.1991
train: epoch 500 : 7/ 20: loss: 113.6191
train: epoch 500 : 8/ 20: loss: 65.9736
train: epoch 500 : 9/ 20: loss: 61.3054
train: epoch 500 : 10/ 20: loss: 103.8710
train: epoch 500 : 11/ 20: loss: 84.5174
train: epoch 500 : 12/ 20: loss: 99.2449
train: epoch 500 : 13/ 20: loss: 89.7561
train: epoch 500 : 14/ 20: loss: 107.5378
train: epoch 500 : 15/ 20: loss: 110.6297
train: epoch 500 : 16/ 20: loss: 104.3139
train: epoch 500 : 17/ 20: loss: 82.2448
train: epoch 500 : 18/ 20: loss: 127.6959
train: epoch 500 : 19/ 20: loss: 79.3800
train: epoch 500 : 20/ 20: loss: 94.1195

[tzczsq]

@navy63 I have the same problem. Do you fix it?

[navy63]

出现错误vl_nnconv An input is not a numeric array (or GPU support not compiled).
在命令窗口输入：vl_compilenn('enableGpu', true)

[navy63]

you can try it like this

---

try
nnet.internal.cnngpu.reluForward(1);
catch ME
end

---

Good luck for you