{
  "id": 200276,
  "title": "Submission Scoring Error",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/200276",
  "author_name": "",
  "post_date": "2020-11-29T19:17:01.674561900Z",
  "votes": 1,
  "comment_count": 8,
  "views": 0,
  "content": "<p>Submission errors with \"Submission Scoring Error\" with no additional details.  <br>\nAny clues how can we get over this issue. Please advise.</p>",
  "messages": [
    {
      "id": "1095618",
      "postDate": "11/29/2020 19:17:01",
      "content": "<p>Submission errors with \"Submission Scoring Error\" with no additional details.  <br>\nAny clues how can we get over this issue. Please advise.</p>",
      "rawMarkdown": "Submission errors with \"Submission Scoring Error\" with no additional details.  \nAny clues how can we get over this issue. Please advise.",
      "votes": null
    },
    {
      "id": "1095736",
      "postDate": "11/29/2020 22:24:02",
      "content": "<p>Double check you <code>submission.csv</code>.</p>\n<ol>\n<li>Save it without index, like <code>submission_df.to_csv('submission.csv', index=False)</code>;</li>\n<li>Your <code>submission.csv</code> should contains only two columns with correct names: <code>image_id</code> and <code>label</code>;</li>\n<li><code>image_id</code> should be a correct file name with extension, without quotes, like <code>1000015157.jpg</code>;</li>\n<li><code>label</code> should be in range [0, 4] without quotes too;</li>\n<li>Your submission should contains labels for all images in <code>test_images</code> directory (I think it is your case);</li>\n</ol>\n<p>Check your <code>submission.csv</code> file format locally, it should looks like <code>train.csv</code>, for example:</p>\n<pre><code>image_id,label\n1000015157.jpg,0\n1000201771.jpg,3\n100042118.jpg,1\n1000723321.jpg,1\n</code></pre>\n<p>Try to read <code>sample_submission.csv</code>, add random labels and save it back as <code>submission.csv</code>, it should works and will be a good starter to debug your problem.</p>\n<p>Bonus: when your score on LB will be lower than your CV score check order of labels in <code>submission.csv</code>.</p>",
      "rawMarkdown": "Double check you `submission.csv`.\n1. Save it without index, like `submission_df.to_csv('submission.csv', index=False)`;\n2. Your `submission.csv` should contains only two columns with correct names: `image_id` and `label`;\n3. `image_id` should be a correct file name with extension, without quotes, like `1000015157.jpg`;\n4. `label` should be in range [0, 4] without quotes too;\n5. Your submission should contains labels for all images in `test_images` directory (I think it is your case);\n\nCheck your `submission.csv` file format locally, it should looks like `train.csv`, for example:\n```\nimage_id,label\n1000015157.jpg,0\n1000201771.jpg,3\n100042118.jpg,1\n1000723321.jpg,1\n```\n\nTry to read `sample_submission.csv`, add random labels and save it back as `submission.csv`, it should works and will be a good starter to debug your problem.\n\nBonus: when your score on LB will be lower than your CV score check order of labels in `submission.csv`.",
      "votes": null
    },
    {
      "id": "1095765",
      "postDate": "11/29/2020 23:34:16",
      "content": "<p>image_id,label<br>\n2216849948.jpg,4</p>\n<p>Contents of submission.csv looks like above.  I do not see any issues.  There is only one test image.</p>",
      "rawMarkdown": "image_id,label\n2216849948.jpg,4\n\nContents of submission.csv looks like above.  I do not see any issues.  There is only one test image.",
      "votes": null
    },
    {
      "id": "1095772",
      "postDate": "11/29/2020 23:52:39",
      "content": "<p>Please take a look at five point in comment below. For successful submit <code>submission.csv</code> should contains all images from <code>sample_submission.csv</code>. Now you can see only one image in test folder, but when you click submit, you solution will start in another environment with another <code>test_images</code> directory, contains 15 000 images. In testing mode (after submit) <code>sample_submission.csv</code> will contains all of these images too, and all 15k images must be in your <code>submission.csv</code>.</p>",
      "rawMarkdown": "Please take a look at five point in comment below. For successful submit `submission.csv` should contains all images from `sample_submission.csv`. Now you can see only one image in test folder, but when you click submit, you solution will start in another environment with another `test_images` directory, contains 15 000 images. In testing mode (after submit) `sample_submission.csv` will contains all of these images too, and all 15k images must be in your `submission.csv`.",
      "votes": null
    },
    {
      "id": "1095775",
      "postDate": "11/30/2020 00:04:33",
      "content": "<p>To test for memory or processing time images, substitute the train data for the test data during inference. Make sure it runs correctly.</p>",
      "rawMarkdown": "To test for memory or processing time images, substitute the train data for the test data during inference. Make sure it runs correctly.",
      "votes": null
    },
    {
      "id": "1096247",
      "postDate": "11/30/2020 10:40:52",
      "content": "<p>Can we use the test images path as  '/kaggle/input/cassava-leaf-disease-classification/test_images/' .  Please advise.</p>",
      "rawMarkdown": "Can we use the test images path as  '/kaggle/input/cassava-leaf-disease-classification/test_images/' .  Please advise.",
      "votes": null
    },
    {
      "id": "1096250",
      "postDate": "11/30/2020 10:43:48",
      "content": "<p>Yes, we can</p>",
      "rawMarkdown": "Yes, we can",
      "votes": null
    },
    {
      "id": "1096539",
      "postDate": "11/30/2020 15:30:13",
      "content": "<p>Now, the submission did not error, some progress.  But the public score is at 0.00.  Not sure where is it going wrong!</p>",
      "rawMarkdown": "Now, the submission did not error, some progress.  But the public score is at 0.00.  Not sure where is it going wrong!",
      "votes": null
    },
    {
      "id": "1098317",
      "postDate": "12/01/2020 15:12:08",
      "content": "<p>I just resolved my submission error.  Initially, I tried to make a data set containing all of the test images. It worked during testing since there is only one image there.  When I replaced the training directory for testing directory, the notebook fails with out of memory error.  So I have to rewrite my prediction code to perform prediction in batches, which resolved the memory issue.  Now,  I am able to submit without issue.</p>",
      "rawMarkdown": "I just resolved my submission error.  Initially, I tried to make a data set containing all of the test images. It worked during testing since there is only one image there.  When I replaced the training directory for testing directory, the notebook fails with out of memory error.  So I have to rewrite my prediction code to perform prediction in batches, which resolved the memory issue.  Now,  I am able to submit without issue.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1095736,
      "author_name": "khlevnov",
      "author_url": "",
      "post_date": "11/29/2020 22:24:02",
      "content": "<p>Double check you <code>submission.csv</code>.</p>\n<ol>\n<li>Save it without index, like <code>submission_df.to_csv('submission.csv', index=False)</code>;</li>\n<li>Your <code>submission.csv</code> should contains only two columns with correct names: <code>image_id</code> and <code>label</code>;</li>\n<li><code>image_id</code> should be a correct file name with extension, without quotes, like <code>1000015157.jpg</code>;</li>\n<li><code>label</code> should be in range [0, 4] without quotes too;</li>\n<li>Your submission should contains labels for all images in <code>test_images</code> directory (I think it is your case);</li>\n</ol>\n<p>Check your <code>submission.csv</code> file format locally, it should looks like <code>train.csv</code>, for example:</p>\n<pre><code>image_id,label\n1000015157.jpg,0\n1000201771.jpg,3\n100042118.jpg,1\n1000723321.jpg,1\n</code></pre>\n<p>Try to read <code>sample_submission.csv</code>, add random labels and save it back as <code>submission.csv</code>, it should works and will be a good starter to debug your problem.</p>\n<p>Bonus: when your score on LB will be lower than your CV score check order of labels in <code>submission.csv</code>.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1095765,
      "author_name": "godina",
      "author_url": "",
      "post_date": "11/29/2020 23:34:16",
      "content": "<p>image_id,label<br>\n2216849948.jpg,4</p>\n<p>Contents of submission.csv looks like above.  I do not see any issues.  There is only one test image.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1095772,
          "author_name": "khlevnov",
          "author_url": "",
          "post_date": "11/29/2020 23:52:39",
          "content": "<p>Please take a look at five point in comment below. For successful submit <code>submission.csv</code> should contains all images from <code>sample_submission.csv</code>. Now you can see only one image in test folder, but when you click submit, you solution will start in another environment with another <code>test_images</code> directory, contains 15 000 images. In testing mode (after submit) <code>sample_submission.csv</code> will contains all of these images too, and all 15k images must be in your <code>submission.csv</code>.</p>",
          "votes": null,
          "replies": [
            {
              "id": 1095775,
              "author_name": "richardepstein",
              "author_url": "",
              "post_date": "11/30/2020 00:04:33",
              "content": "<p>To test for memory or processing time images, substitute the train data for the test data during inference. Make sure it runs correctly.</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 1096247,
      "author_name": "godina",
      "author_url": "",
      "post_date": "11/30/2020 10:40:52",
      "content": "<p>Can we use the test images path as  '/kaggle/input/cassava-leaf-disease-classification/test_images/' .  Please advise.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1096250,
          "author_name": "khlevnov",
          "author_url": "",
          "post_date": "11/30/2020 10:43:48",
          "content": "<p>Yes, we can</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1096539,
      "author_name": "godina",
      "author_url": "",
      "post_date": "11/30/2020 15:30:13",
      "content": "<p>Now, the submission did not error, some progress.  But the public score is at 0.00.  Not sure where is it going wrong!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1098317,
      "author_name": "luqing2",
      "author_url": "",
      "post_date": "12/01/2020 15:12:08",
      "content": "<p>I just resolved my submission error.  Initially, I tried to make a data set containing all of the test images. It worked during testing since there is only one image there.  When I replaced the training directory for testing directory, the notebook fails with out of memory error.  So I have to rewrite my prediction code to perform prediction in batches, which resolved the memory issue.  Now,  I am able to submit without issue.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1095618": "Submission errors with \"Submission Scoring Error\" with no additional details.  \nAny clues how can we get over this issue. Please advise.",
    "1095736": "Double check you `submission.csv`.\n1. Save it without index, like `submission_df.to_csv('submission.csv', index=False)`;\n2. Your `submission.csv` should contains only two columns with correct names: `image_id` and `label`;\n3. `image_id` should be a correct file name with extension, without quotes, like `1000015157.jpg`;\n4. `label` should be in range [0, 4] without quotes too;\n5. Your submission should contains labels for all images in `test_images` directory (I think it is your case);\n\nCheck your `submission.csv` file format locally, it should looks like `train.csv`, for example:\n```\nimage_id,label\n1000015157.jpg,0\n1000201771.jpg,3\n100042118.jpg,1\n1000723321.jpg,1\n```\n\nTry to read `sample_submission.csv`, add random labels and save it back as `submission.csv`, it should works and will be a good starter to debug your problem.\n\nBonus: when your score on LB will be lower than your CV score check order of labels in `submission.csv`.",
    "1095765": "image_id,label\n2216849948.jpg,4\n\nContents of submission.csv looks like above.  I do not see any issues.  There is only one test image.",
    "1095772": "Please take a look at five point in comment below. For successful submit `submission.csv` should contains all images from `sample_submission.csv`. Now you can see only one image in test folder, but when you click submit, you solution will start in another environment with another `test_images` directory, contains 15 000 images. In testing mode (after submit) `sample_submission.csv` will contains all of these images too, and all 15k images must be in your `submission.csv`.",
    "1095775": "To test for memory or processing time images, substitute the train data for the test data during inference. Make sure it runs correctly.",
    "1096247": "Can we use the test images path as  '/kaggle/input/cassava-leaf-disease-classification/test_images/' .  Please advise.",
    "1096250": "Yes, we can",
    "1096539": "Now, the submission did not error, some progress.  But the public score is at 0.00.  Not sure where is it going wrong!",
    "1098317": "I just resolved my submission error.  Initially, I tried to make a data set containing all of the test images. It worked during testing since there is only one image there.  When I replaced the training directory for testing directory, the notebook fails with out of memory error.  So I have to rewrite my prediction code to perform prediction in batches, which resolved the memory issue.  Now,  I am able to submit without issue."
  },
  "source": "meta"
}