{
  "id": 78528,
  "title": "kernel submit error",
  "url": "/competitions/quora-insincere-questions-classification/discussion/78528",
  "author_name": "",
  "post_date": "2019-01-25T03:03:14.241429600Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi,</p>\n\n<p>I'm new to kernels and I have been struggling with an error when calling keras model.fit(). I searched the website and the link <em><a href=\"https://devtalk.nvidia.com/default/topic/1030610/cuda-setup-and-installation/fail-to-find-the-dnn-implementation-/\">https://devtalk.nvidia.com/default/topic/1030610/cuda-setup-and-installation/fail-to-find-the-dnn-implementation-/</a></em> said it's cudnn issue? Anyone know how to deal with it? The code runs smoothly on my local machine but I cannot submit successfully with this error. </p>\n\n<hr>\n\n<p>UnknownError: Fail to find the dnn implementation.\n     [[{{node bidirectional_1/CudnnRNN}} = CudnnRNN[T=DT_FLOAT, _class=[\"loc:<a href=\"/training\">@training</a>/Adam/gradients/bidirectional_1/CudnnRNN_grad/CudnnRNNBackprop\"], direction=\"unidirectional\", dropout=0, input_mode=\"linear_input\", is_training=true, rnn_mode=\"lstm\", seed=87654321, seed2=0, _device=\"/job:localhost/replica:0/task:0/device:GPU:0\"](bidirectional_1/transpose, bidirectional_1/ExpandDims_1, bidirectional_1/ExpandDims_1, bidirectional_1/concat)]]\n     [[{{node metrics/acc/Mean_1/_193}} = _Recvclient_terminated=false, recv_device=\"/job:localhost/replica:0/task:0/device:CPU:0\", send_device=\"/job:localhost/replica:0/task:0/device:GPU:0\", send_device_incarnation=1, tensor_name=\"edge_2370_metrics/acc/Mean_1\", tensor_type=DT_FLOAT, _device=\"/job:localhost/replica:0/task:0/device:CPU:0\"]]</p>",
  "messages": [
    {
      "id": "461028",
      "postDate": "01/25/2019 03:03:14",
      "content": "<p>Hi,</p>\n\n<p>I'm new to kernels and I have been struggling with an error when calling keras model.fit(). I searched the website and the link <em><a href=\"https://devtalk.nvidia.com/default/topic/1030610/cuda-setup-and-installation/fail-to-find-the-dnn-implementation-/\">https://devtalk.nvidia.com/default/topic/1030610/cuda-setup-and-installation/fail-to-find-the-dnn-implementation-/</a></em> said it's cudnn issue? Anyone know how to deal with it? The code runs smoothly on my local machine but I cannot submit successfully with this error. </p>\n\n<hr>\n\n<p>UnknownError: Fail to find the dnn implementation.\n     [[{{node bidirectional_1/CudnnRNN}} = CudnnRNN[T=DT_FLOAT, _class=[\"loc:<a href=\"/training\">@training</a>/Adam/gradients/bidirectional_1/CudnnRNN_grad/CudnnRNNBackprop\"], direction=\"unidirectional\", dropout=0, input_mode=\"linear_input\", is_training=true, rnn_mode=\"lstm\", seed=87654321, seed2=0, _device=\"/job:localhost/replica:0/task:0/device:GPU:0\"](bidirectional_1/transpose, bidirectional_1/ExpandDims_1, bidirectional_1/ExpandDims_1, bidirectional_1/concat)]]\n     [[{{node metrics/acc/Mean_1/_193}} = _Recvclient_terminated=false, recv_device=\"/job:localhost/replica:0/task:0/device:CPU:0\", send_device=\"/job:localhost/replica:0/task:0/device:GPU:0\", send_device_incarnation=1, tensor_name=\"edge_2370_metrics/acc/Mean_1\", tensor_type=DT_FLOAT, _device=\"/job:localhost/replica:0/task:0/device:CPU:0\"]]</p>",
      "rawMarkdown": "Hi,\n\nI'm new to kernels and I have been struggling with an error when calling keras model.fit(). I searched the website and the link *https://devtalk.nvidia.com/default/topic/1030610/cuda-setup-and-installation/fail-to-find-the-dnn-implementation-/* said it's cudnn issue? Anyone know how to deal with it? The code runs smoothly on my local machine but I cannot submit successfully with this error. \n\n\n\n\n---------------------------------------------------------------------------\nUnknownError: Fail to find the dnn implementation.\n\t [[{{node bidirectional_1/CudnnRNN}} = CudnnRNN[T=DT_FLOAT, _class=[\"loc:@training/Adam/gradients/bidirectional_1/CudnnRNN_grad/CudnnRNNBackprop\"], direction=\"unidirectional\", dropout=0, input_mode=\"linear_input\", is_training=true, rnn_mode=\"lstm\", seed=87654321, seed2=0, _device=\"/job:localhost/replica:0/task:0/device:GPU:0\"](bidirectional_1/transpose, bidirectional_1/ExpandDims_1, bidirectional_1/ExpandDims_1, bidirectional_1/concat)]]\n\t [[{{node metrics/acc/Mean_1/_193}} = _Recv[client_terminated=false, recv_device=\"/job:localhost/replica:0/task:0/device:CPU:0\", send_device=\"/job:localhost/replica:0/task:0/device:GPU:0\", send_device_incarnation=1, tensor_name=\"edge_2370_metrics/acc/Mean_1\", tensor_type=DT_FLOAT, _device=\"/job:localhost/replica:0/task:0/device:CPU:0\"]()]]",
      "votes": null
    },
    {
      "id": "461045",
      "postDate": "01/25/2019 04:32:38",
      "content": "<p>Good day. Please look this work: <a href=\"https://www.kaggle.com/gmhost/gru-capsule\">https://www.kaggle.com/gmhost/gru-capsule</a> in Kernel section. Author creates model with Keras and CuDNNGRU support. Everything works very well. Maybe you miss something.</p>",
      "rawMarkdown": "Good day. Please look this work: https://www.kaggle.com/gmhost/gru-capsule in Kernel section. Author creates model with Keras and CuDNNGRU support. Everything works very well. Maybe you miss something.",
      "votes": null
    },
    {
      "id": "461588",
      "postDate": "01/26/2019 13:01:14",
      "content": "<p>I changed the order of my \"imports\" and my code works now. I still don't know the reason of the error. Thank you for your recommendation I will study this kernel.</p>",
      "rawMarkdown": "I changed the order of my \"imports\" and my code works now. I still don't know the reason of the error. Thank you for your recommendation I will study this kernel.",
      "votes": null
    },
    {
      "id": "461591",
      "postDate": "01/26/2019 13:07:31",
      "content": "<p>Good luck in competition )</p>",
      "rawMarkdown": "Good luck in competition )",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 461045,
      "author_name": "alexanderkireev",
      "author_url": "",
      "post_date": "01/25/2019 04:32:38",
      "content": "<p>Good day. Please look this work: <a href=\"https://www.kaggle.com/gmhost/gru-capsule\">https://www.kaggle.com/gmhost/gru-capsule</a> in Kernel section. Author creates model with Keras and CuDNNGRU support. Everything works very well. Maybe you miss something.</p>",
      "votes": null,
      "replies": [
        {
          "id": 461588,
          "author_name": "wowfattie",
          "author_url": "",
          "post_date": "01/26/2019 13:01:14",
          "content": "<p>I changed the order of my \"imports\" and my code works now. I still don't know the reason of the error. Thank you for your recommendation I will study this kernel.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 461591,
          "author_name": "alexanderkireev",
          "author_url": "",
          "post_date": "01/26/2019 13:07:31",
          "content": "<p>Good luck in competition )</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "461028": "Hi,\n\nI'm new to kernels and I have been struggling with an error when calling keras model.fit(). I searched the website and the link *https://devtalk.nvidia.com/default/topic/1030610/cuda-setup-and-installation/fail-to-find-the-dnn-implementation-/* said it's cudnn issue? Anyone know how to deal with it? The code runs smoothly on my local machine but I cannot submit successfully with this error. \n\n\n\n\n---------------------------------------------------------------------------\nUnknownError: Fail to find the dnn implementation.\n\t [[{{node bidirectional_1/CudnnRNN}} = CudnnRNN[T=DT_FLOAT, _class=[\"loc:@training/Adam/gradients/bidirectional_1/CudnnRNN_grad/CudnnRNNBackprop\"], direction=\"unidirectional\", dropout=0, input_mode=\"linear_input\", is_training=true, rnn_mode=\"lstm\", seed=87654321, seed2=0, _device=\"/job:localhost/replica:0/task:0/device:GPU:0\"](bidirectional_1/transpose, bidirectional_1/ExpandDims_1, bidirectional_1/ExpandDims_1, bidirectional_1/concat)]]\n\t [[{{node metrics/acc/Mean_1/_193}} = _Recv[client_terminated=false, recv_device=\"/job:localhost/replica:0/task:0/device:CPU:0\", send_device=\"/job:localhost/replica:0/task:0/device:GPU:0\", send_device_incarnation=1, tensor_name=\"edge_2370_metrics/acc/Mean_1\", tensor_type=DT_FLOAT, _device=\"/job:localhost/replica:0/task:0/device:CPU:0\"]()]]",
    "461045": "Good day. Please look this work: https://www.kaggle.com/gmhost/gru-capsule in Kernel section. Author creates model with Keras and CuDNNGRU support. Everything works very well. Maybe you miss something.",
    "461588": "I changed the order of my \"imports\" and my code works now. I still don't know the reason of the error. Thank you for your recommendation I will study this kernel.",
    "461591": "Good luck in competition )"
  },
  "source": "meta"
}