{
  "id": 459977,
  "title": "Detect and Init TPU",
  "url": "/competitions/stanford-ribonanza-rna-folding/discussion/459977",
  "author_name": "",
  "post_date": "2023-12-07T09:22:04.985521600Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Reference to  \"Stanford Ribonanza RNA Folding competition\". <br>\n<a href=\"https://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/\" target=\"_blank\">https://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/</a></p>\n<p>Recently, I ran out of GPU limit of 30hrs and was planning to utilize 20 hrs quota of TPU for<br>\ncomputation.  However, when I tried the standard code for detecting and initiating TPU:<br>\ntpu = tf.distribute.cluster_resolver.TPUClusterResolver.connect()</p>\n<p>I am getting the following error:<br>\nTypeError: initialize_tpu_system_impl() missing 1 required positional argument: 'tpu_cluster_resolver_cls'</p>\n<p>Appreciate getting tips on solving this issue. Also, appreciate advise on utilizing  TPU instead of GPU to speedup computation. </p>",
  "messages": [
    {
      "id": "2552188",
      "postDate": "12/07/2023 09:22:04",
      "content": "<p>Reference to  \"Stanford Ribonanza RNA Folding competition\". <br>\n<a href=\"https://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/\" target=\"_blank\">https://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/</a></p>\n<p>Recently, I ran out of GPU limit of 30hrs and was planning to utilize 20 hrs quota of TPU for<br>\ncomputation.  However, when I tried the standard code for detecting and initiating TPU:<br>\ntpu = tf.distribute.cluster_resolver.TPUClusterResolver.connect()</p>\n<p>I am getting the following error:<br>\nTypeError: initialize_tpu_system_impl() missing 1 required positional argument: 'tpu_cluster_resolver_cls'</p>\n<p>Appreciate getting tips on solving this issue. Also, appreciate advise on utilizing  TPU instead of GPU to speedup computation. </p>",
      "rawMarkdown": "Reference to  \"Stanford Ribonanza RNA Folding competition\". \nhttps://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/\n\nRecently, I ran out of GPU limit of 30hrs and was planning to utilize 20 hrs quota of TPU for\ncomputation.  However, when I tried the standard code for detecting and initiating TPU:\ntpu = tf.distribute.cluster_resolver.TPUClusterResolver.connect()\n\nI am getting the following error:\nTypeError: initialize_tpu_system_impl() missing 1 required positional argument: 'tpu_cluster_resolver_cls'\n\nAppreciate getting tips on solving this issue. Also, appreciate advise on utilizing  TPU instead of GPU to speedup computation.",
      "votes": null
    },
    {
      "id": "2552222",
      "postDate": "12/07/2023 09:50:42",
      "content": "<p>This is no longer works in the new kaggle image (tensorflow 14.0). Copy a notebook that use the old image and it will work.</p>",
      "rawMarkdown": "This is no longer works in the new kaggle image (tensorflow 14.0). Copy a notebook that use the old image and it will work.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2552222,
      "author_name": "shlomoron",
      "author_url": "",
      "post_date": "12/07/2023 09:50:42",
      "content": "<p>This is no longer works in the new kaggle image (tensorflow 14.0). Copy a notebook that use the old image and it will work.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2552188": "Reference to  \"Stanford Ribonanza RNA Folding competition\". \nhttps://www.kaggle.com/competitions/stanford-ribonanza-rna-folding/\n\nRecently, I ran out of GPU limit of 30hrs and was planning to utilize 20 hrs quota of TPU for\ncomputation.  However, when I tried the standard code for detecting and initiating TPU:\ntpu = tf.distribute.cluster_resolver.TPUClusterResolver.connect()\n\nI am getting the following error:\nTypeError: initialize_tpu_system_impl() missing 1 required positional argument: 'tpu_cluster_resolver_cls'\n\nAppreciate getting tips on solving this issue. Also, appreciate advise on utilizing  TPU instead of GPU to speedup computation.",
    "2552222": "This is no longer works in the new kaggle image (tensorflow 14.0). Copy a notebook that use the old image and it will work."
  },
  "source": "meta"
}