{
  "id": 680420,
  "title": "Notebook Submission not throwing errors but also stuck in running",
  "url": "/competitions/deep-past-initiative-machine-translation/discussion/680420",
  "author_name": "",
  "post_date": "2026-03-08T02:29:32.876768800Z",
  "votes": 1,
  "comment_count": 11,
  "views": 0,
  "content": "<p>I submitted a notebook for this competition, and for 10 hours now it has never submitted and has been stuck in the loading circle. I have not received any kind of submission error message associated with the notebook, but I'm starting to think that I went wrong somewhere in the code because of the runtime.</p>\n<p>If anyone has answers, that would be much appreciated!</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16323783%2F104f38117d841375593dbd9a30e57fc7%2FScreenshot%202026-03-07%20212530.png?generation=1772936970884794&amp;alt=media\" alt=\"\"></p>",
  "messages": [
    {
      "id": "3418426",
      "postDate": "03/08/2026 02:29:32",
      "content": "<p>I submitted a notebook for this competition, and for 10 hours now it has never submitted and has been stuck in the loading circle. I have not received any kind of submission error message associated with the notebook, but I'm starting to think that I went wrong somewhere in the code because of the runtime.</p>\n<p>If anyone has answers, that would be much appreciated!</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16323783%2F104f38117d841375593dbd9a30e57fc7%2FScreenshot%202026-03-07%20212530.png?generation=1772936970884794&amp;alt=media\" alt=\"\"></p>",
      "rawMarkdown": "I submitted a notebook for this competition, and for 10 hours now it has never submitted and has been stuck in the loading circle. I have not received any kind of submission error message associated with the notebook, but I'm starting to think that I went wrong somewhere in the code because of the runtime.\n\nIf anyone has answers, that would be much appreciated!\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16323783%2F104f38117d841375593dbd9a30e57fc7%2FScreenshot%202026-03-07%20212530.png?generation=1772936970884794&alt=media)",
      "votes": null
    },
    {
      "id": "3418468",
      "postDate": "03/08/2026 06:49:59",
      "content": "<p>please check if the gpu is on ?</p>",
      "rawMarkdown": "please check if the gpu is on ?",
      "votes": null
    },
    {
      "id": "3420504",
      "postDate": "03/13/2026 13:02:17",
      "content": "<p>Thanks for the reply AK, I submitted a notebook with the GPU T4, but so far it has taken 30 minutes and I still do not have a score. Is this normal, and is there a difference between the GPU T4 and P100?</p>",
      "rawMarkdown": "Thanks for the reply AK, I submitted a notebook with the GPU T4, but so far it has taken 30 minutes and I still do not have a score. Is this normal, and is there a difference between the GPU T4 and P100?",
      "votes": null
    },
    {
      "id": "3420539",
      "postDate": "03/13/2026 14:16:18",
      "content": "<p>I haven’t tried P100 in this competition but it usually takes double the time than 2*T4 ☺️ </p>",
      "rawMarkdown": "I haven’t tried P100 in this competition but it usually takes double the time than 2*T4 ☺️",
      "votes": null
    },
    {
      "id": "3420578",
      "postDate": "03/13/2026 15:35:53",
      "content": "<p>How long did it take your notebook to submit and get a score? Mine has been running for three hours at this point… I checked and I'm using the T4 x2</p>",
      "rawMarkdown": "How long did it take your notebook to submit and get a score? Mine has been running for three hours at this point... I checked and I'm using the T4 x2",
      "votes": null
    },
    {
      "id": "3420585",
      "postDate": "03/13/2026 15:40:22",
      "content": "<p>Approx 10-12 mins</p>",
      "rawMarkdown": "Approx 10-12 mins",
      "votes": null
    },
    {
      "id": "3423732",
      "postDate": "03/19/2026 00:16:37",
      "content": "<p>Happened to me once with byt5 + transformers==5.x. Switching to transformers==4.57.x fixed it for me. </p>",
      "rawMarkdown": "Happened to me once with byt5 + transformers==5.x. Switching to transformers==4.57.x fixed it for me.",
      "votes": null
    },
    {
      "id": "3424241",
      "postDate": "03/19/2026 13:41:57",
      "content": "<p>that's really interesting. why do you think this was the case? for me I got it fixed by moving the inputs and model to the cuda device.</p>",
      "rawMarkdown": "that's really interesting. why do you think this was the case? for me I got it fixed by moving the inputs and model to the cuda device.",
      "votes": null
    },
    {
      "id": "3425029",
      "postDate": "03/20/2026 08:17:24",
      "content": "<p>I checked my <code>generation_config.json</code> and it saved some more fields like <code>'use_cache': False</code>. I suspect that caused my program to run longer.</p>\n<p>I added <code>use_cache=True</code> in <code>model.generate</code> explicitly, with <code>model.generation_config.use_cache = True</code> alongside <code>model.config.use_cache = True</code>, and it helped solve the issue.</p>",
      "rawMarkdown": "I checked my `generation_config.json` and it saved some more fields like `'use_cache': False`. I suspect that caused my program to run longer.\n\nI added `use_cache=True` in `model.generate` explicitly, with `model.generation_config.use_cache = True` alongside `model.config.use_cache = True`, and it helped solve the issue.",
      "votes": null
    },
    {
      "id": "3425962",
      "postDate": "03/21/2026 15:50:25",
      "content": "<p>Interesting. Why do you think doing this helped? I'm just curious in learning as much as I can about this stuff</p>",
      "rawMarkdown": "Interesting. Why do you think doing this helped? I'm just curious in learning as much as I can about this stuff",
      "votes": null
    },
    {
      "id": "3426157",
      "postDate": "03/22/2026 01:15:38",
      "content": "<p>I don't know for sure but I asked ChatGPT:\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17202047%2F7edb20d7cc560936832e36d815b38a61%2FSnipaste_2026-03-22_09-10-22.png?generation=1774141890331585&amp;alt=media\" alt=\"\"></p>\n<p>It seems that the priority of the <code>use_cache</code> field in <code>generation_config.json</code> is higher than that in <code>model.config</code>. Previously I only set <code>model.config.use_cache = True</code> so it got overridden.</p>",
      "rawMarkdown": "I don't know for sure but I asked ChatGPT:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17202047%2F7edb20d7cc560936832e36d815b38a61%2FSnipaste_2026-03-22_09-10-22.png?generation=1774141890331585&alt=media)\n\nIt seems that the priority of the `use_cache` field in `generation_config.json` is higher than that in `model.config`. Previously I only set `model.config.use_cache = True` so it got overridden.",
      "votes": null
    },
    {
      "id": "3426197",
      "postDate": "03/22/2026 04:12:50",
      "content": "<p>I do not really know. I suspect it is something to do with tokenizer.batch_decode when model generates some weird characters. </p>",
      "rawMarkdown": "I do not really know. I suspect it is something to do with tokenizer.batch_decode when model generates some weird characters.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3418468,
      "author_name": "aman1391",
      "author_url": "",
      "post_date": "03/08/2026 06:49:59",
      "content": "<p>please check if the gpu is on ?</p>",
      "votes": null,
      "replies": [
        {
          "id": 3420504,
          "author_name": "theswimmer",
          "author_url": "",
          "post_date": "03/13/2026 13:02:17",
          "content": "<p>Thanks for the reply AK, I submitted a notebook with the GPU T4, but so far it has taken 30 minutes and I still do not have a score. Is this normal, and is there a difference between the GPU T4 and P100?</p>",
          "votes": null,
          "replies": [
            {
              "id": 3420539,
              "author_name": "aman1391",
              "author_url": "",
              "post_date": "03/13/2026 14:16:18",
              "content": "<p>I haven’t tried P100 in this competition but it usually takes double the time than 2*T4 ☺️ </p>",
              "votes": null,
              "replies": [
                {
                  "id": 3420578,
                  "author_name": "theswimmer",
                  "author_url": "",
                  "post_date": "03/13/2026 15:35:53",
                  "content": "<p>How long did it take your notebook to submit and get a score? Mine has been running for three hours at this point… I checked and I'm using the T4 x2</p>",
                  "votes": null,
                  "replies": [
                    {
                      "id": 3420585,
                      "author_name": "aman1391",
                      "author_url": "",
                      "post_date": "03/13/2026 15:40:22",
                      "content": "<p>Approx 10-12 mins</p>",
                      "votes": null,
                      "replies": []
                    }
                  ]
                }
              ]
            }
          ]
        }
      ]
    },
    {
      "id": 3423732,
      "author_name": "ky8ori",
      "author_url": "",
      "post_date": "03/19/2026 00:16:37",
      "content": "<p>Happened to me once with byt5 + transformers==5.x. Switching to transformers==4.57.x fixed it for me. </p>",
      "votes": null,
      "replies": [
        {
          "id": 3424241,
          "author_name": "theswimmer",
          "author_url": "",
          "post_date": "03/19/2026 13:41:57",
          "content": "<p>that's really interesting. why do you think this was the case? for me I got it fixed by moving the inputs and model to the cuda device.</p>",
          "votes": null,
          "replies": [
            {
              "id": 3426197,
              "author_name": "ky8ori",
              "author_url": "",
              "post_date": "03/22/2026 04:12:50",
              "content": "<p>I do not really know. I suspect it is something to do with tokenizer.batch_decode when model generates some weird characters. </p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 3425029,
      "author_name": "joey111mygo",
      "author_url": "",
      "post_date": "03/20/2026 08:17:24",
      "content": "<p>I checked my <code>generation_config.json</code> and it saved some more fields like <code>'use_cache': False</code>. I suspect that caused my program to run longer.</p>\n<p>I added <code>use_cache=True</code> in <code>model.generate</code> explicitly, with <code>model.generation_config.use_cache = True</code> alongside <code>model.config.use_cache = True</code>, and it helped solve the issue.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3425962,
          "author_name": "theswimmer",
          "author_url": "",
          "post_date": "03/21/2026 15:50:25",
          "content": "<p>Interesting. Why do you think doing this helped? I'm just curious in learning as much as I can about this stuff</p>",
          "votes": null,
          "replies": [
            {
              "id": 3426157,
              "author_name": "joey111mygo",
              "author_url": "",
              "post_date": "03/22/2026 01:15:38",
              "content": "<p>I don't know for sure but I asked ChatGPT:\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17202047%2F7edb20d7cc560936832e36d815b38a61%2FSnipaste_2026-03-22_09-10-22.png?generation=1774141890331585&amp;alt=media\" alt=\"\"></p>\n<p>It seems that the priority of the <code>use_cache</code> field in <code>generation_config.json</code> is higher than that in <code>model.config</code>. Previously I only set <code>model.config.use_cache = True</code> so it got overridden.</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3418426": "I submitted a notebook for this competition, and for 10 hours now it has never submitted and has been stuck in the loading circle. I have not received any kind of submission error message associated with the notebook, but I'm starting to think that I went wrong somewhere in the code because of the runtime.\n\nIf anyone has answers, that would be much appreciated!\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16323783%2F104f38117d841375593dbd9a30e57fc7%2FScreenshot%202026-03-07%20212530.png?generation=1772936970884794&alt=media)",
    "3418468": "please check if the gpu is on ?",
    "3420504": "Thanks for the reply AK, I submitted a notebook with the GPU T4, but so far it has taken 30 minutes and I still do not have a score. Is this normal, and is there a difference between the GPU T4 and P100?",
    "3420539": "I haven’t tried P100 in this competition but it usually takes double the time than 2*T4 ☺️",
    "3420578": "How long did it take your notebook to submit and get a score? Mine has been running for three hours at this point... I checked and I'm using the T4 x2",
    "3420585": "Approx 10-12 mins",
    "3423732": "Happened to me once with byt5 + transformers==5.x. Switching to transformers==4.57.x fixed it for me.",
    "3424241": "that's really interesting. why do you think this was the case? for me I got it fixed by moving the inputs and model to the cuda device.",
    "3425029": "I checked my `generation_config.json` and it saved some more fields like `'use_cache': False`. I suspect that caused my program to run longer.\n\nI added `use_cache=True` in `model.generate` explicitly, with `model.generation_config.use_cache = True` alongside `model.config.use_cache = True`, and it helped solve the issue.",
    "3425962": "Interesting. Why do you think doing this helped? I'm just curious in learning as much as I can about this stuff",
    "3426157": "I don't know for sure but I asked ChatGPT:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F17202047%2F7edb20d7cc560936832e36d815b38a61%2FSnipaste_2026-03-22_09-10-22.png?generation=1774141890331585&alt=media)\n\nIt seems that the priority of the `use_cache` field in `generation_config.json` is higher than that in `model.config`. Previously I only set `model.config.use_cache = True` so it got overridden.",
    "3426197": "I do not really know. I suspect it is something to do with tokenizer.batch_decode when model generates some weird characters."
  },
  "source": "meta"
}