{
  "id": 529325,
  "title": "Notebook Threw Exception",
  "url": "/competitions/ariel-data-challenge-2024/discussion/529325",
  "author_name": "Umar IGAN",
  "post_date": "2024-08-20T06:02:20.017000",
  "votes": 0,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Hello,<br>\nI am trying to submit my notebook but getting Notebook Threw Exception,<br>\nAlthough when i run my notebook,  I face no error and and I can see my submission file. The notebook simple read data getssome image features and model using a simple NN model but still submission ends up with error. <br>\nDo you have any advice to debug this? Internet access is closed too.</p>",
  "messages": [
    {
      "id": 2964804,
      "postDate": "2024-08-20T07:40:00.990Z",
      "content": "<p>Injecting random data to test the submission code for robustness often helps:</p>\n<pre><code> = pd.read_csv('/kaggle/input/ariel-data-challenge-/test_adc_info.csv',\n                           ='planet_id')\n\n\n = pd.concat([test_adc_info] * , axis=)\n.iloc[:, -] =  # unseen star\n\n = f_read_and_preprocess('test', test_adc_info, test_adc_info.index)\n = a_read_and_preprocess('test', test_adc_info, test_adc_info.index)\n\n\n = np.random.default_rng()\n[:] = rng.normal(size=(, f_raw_test.shape[]))\n[:] = rng.normal(size=(, a_raw_test.shape[]))\n</code></pre>\n<p>You can as well test the inference code for memory overflow by feeding it a dataset with 800+ planets. And I assume you've made sure that the inference notebook does not access the training dataset.</p>",
      "rawMarkdown": "Injecting random data to test the submission code for robustness often helps:\n\n```\ntest_adc_info = pd.read_csv('/kaggle/input/ariel-data-challenge-2024/test_adc_info.csv',\n                           index_col='planet_id')\n\n# Check robustness (add more rows and a new star)\ntest_adc_info = pd.concat([test_adc_info] * 10, axis=0)\ntest_adc_info.iloc[5:, -1] = 2 # unseen star\n\nf_raw_test = f_read_and_preprocess('test', test_adc_info, test_adc_info.index)\na_raw_test = a_read_and_preprocess('test', test_adc_info, test_adc_info.index)\n\n# Check robustness (inject unexpected data)\nrng = np.random.default_rng()\nf_raw_test[2:] = rng.normal(size=(8, f_raw_test.shape[1]))\na_raw_test[2:] = rng.normal(size=(8, a_raw_test.shape[1]))\n\n```\n\nYou can as well test the inference code for memory overflow by feeding it a dataset with 800+ planets. And I assume you've made sure that the inference notebook does not access the training dataset.",
      "votes": 2,
      "replies": [
        {
          "id": 2966287,
          "postDate": "2024-08-21T17:44:42.057Z",
          "content": "<p>Hey thanks fot the response and sorry for the late response. Since i saved data which i generated as features from images and used those directly, the code runs very fast and no memory issue. For inference notebook to not access to the training data, are purposing to generate anew notebook and use trained model there? I checked other codes and they mostly first saved the model and then use it for test prediction in the training notebook, I did the same yet I faced same error.</p>",
          "rawMarkdown": "Hey thanks fot the response and sorry for the late response. Since i saved data which i generated as features from images and used those directly, the code runs very fast and no memory issue. For inference notebook to not access to the training data, are purposing to generate anew notebook and use trained model there? I checked other codes and they mostly first saved the model and then use it for test prediction in the training notebook, I did the same yet I faced same error.",
          "votes": 1
        }
      ]
    },
    {
      "id": 2964735,
      "postDate": "2024-08-20T06:02:20.017Z",
      "content": "<p>Hello,<br>\nI am trying to submit my notebook but getting Notebook Threw Exception,<br>\nAlthough when i run my notebook,  I face no error and and I can see my submission file. The notebook simple read data getssome image features and model using a simple NN model but still submission ends up with error. <br>\nDo you have any advice to debug this? Internet access is closed too.</p>",
      "rawMarkdown": "Hello,\nI am trying to submit my notebook but getting Notebook Threw Exception,\nAlthough when i run my notebook,  I face no error and and I can see my submission file. The notebook simple read data getssome image features and model using a simple NN model but still submission ends up with error. \nDo you have any advice to debug this? Internet access is closed too."
    }
  ],
  "comments": [
    {
      "id": 2964804,
      "author_name": "AmbrosM",
      "author_url": "",
      "post_date": "2024-08-20T07:40:00.990000",
      "content": "<p>Injecting random data to test the submission code for robustness often helps:</p>\n<pre><code> = pd.read_csv('/kaggle/input/ariel-data-challenge-/test_adc_info.csv',\n                           ='planet_id')\n\n\n = pd.concat([test_adc_info] * , axis=)\n.iloc[:, -] =  # unseen star\n\n = f_read_and_preprocess('test', test_adc_info, test_adc_info.index)\n = a_read_and_preprocess('test', test_adc_info, test_adc_info.index)\n\n\n = np.random.default_rng()\n[:] = rng.normal(size=(, f_raw_test.shape[]))\n[:] = rng.normal(size=(, a_raw_test.shape[]))\n</code></pre>\n<p>You can as well test the inference code for memory overflow by feeding it a dataset with 800+ planets. And I assume you've made sure that the inference notebook does not access the training dataset.</p>",
      "votes": 2,
      "replies": [
        {
          "id": 2966287,
          "author_name": "Umar IGAN",
          "author_url": "",
          "post_date": "2024-08-21T17:44:42.057000",
          "content": "<p>Hey thanks fot the response and sorry for the late response. Since i saved data which i generated as features from images and used those directly, the code runs very fast and no memory issue. For inference notebook to not access to the training data, are purposing to generate anew notebook and use trained model there? I checked other codes and they mostly first saved the model and then use it for test prediction in the training notebook, I did the same yet I faced same error.</p>",
          "votes": 1,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2964804": "Injecting random data to test the submission code for robustness often helps:\n\n```\ntest_adc_info = pd.read_csv('/kaggle/input/ariel-data-challenge-2024/test_adc_info.csv',\n                           index_col='planet_id')\n\n# Check robustness (add more rows and a new star)\ntest_adc_info = pd.concat([test_adc_info] * 10, axis=0)\ntest_adc_info.iloc[5:, -1] = 2 # unseen star\n\nf_raw_test = f_read_and_preprocess('test', test_adc_info, test_adc_info.index)\na_raw_test = a_read_and_preprocess('test', test_adc_info, test_adc_info.index)\n\n# Check robustness (inject unexpected data)\nrng = np.random.default_rng()\nf_raw_test[2:] = rng.normal(size=(8, f_raw_test.shape[1]))\na_raw_test[2:] = rng.normal(size=(8, a_raw_test.shape[1]))\n\n```\n\nYou can as well test the inference code for memory overflow by feeding it a dataset with 800+ planets. And I assume you've made sure that the inference notebook does not access the training dataset.",
    "2964735": "Hello,\nI am trying to submit my notebook but getting Notebook Threw Exception,\nAlthough when i run my notebook,  I face no error and and I can see my submission file. The notebook simple read data getssome image features and model using a simple NN model but still submission ends up with error. \nDo you have any advice to debug this? Internet access is closed too."
  }
}