{
  "id": 137033,
  "title": "Bengali 2D UMAP Embedding",
  "url": "/competitions/bengaliai-cv19/discussion/137033",
  "author_name": "",
  "post_date": "2020-03-18T23:11:12.422023Z",
  "votes": 4,
  "comment_count": 5,
  "views": 0,
  "content": "<p>At one point I decided to make UMAP embeddings of the original data. It turns out this was a huge undertaking, and the final UMAP model was 250 GB. It took 2.5 days on a 16 core CPU / 128 GB machine to compute. Due to its size, we did not use it, but the embedding can be found <a href=\"https://www.kaggle.com/tunguz/bengali-umap-2d-embedding\">here</a>.</p>\n\n<p>A simple notebook that generates the UMAP model and the files can be found <a href=\"https://www.kaggle.com/tunguz/bengali-2d-umap\">here</a>.</p>\n\n<p><img src=\"https://pbs.twimg.com/media/ETbVYlJWoAENXK6?format=png&amp;name=small\" alt=\"bengali_umap\"></p>",
  "messages": [
    {
      "id": "778989",
      "postDate": "03/18/2020 23:11:12",
      "content": "<p>At one point I decided to make UMAP embeddings of the original data. It turns out this was a huge undertaking, and the final UMAP model was 250 GB. It took 2.5 days on a 16 core CPU / 128 GB machine to compute. Due to its size, we did not use it, but the embedding can be found <a href=\"https://www.kaggle.com/tunguz/bengali-umap-2d-embedding\">here</a>.</p>\n\n<p>A simple notebook that generates the UMAP model and the files can be found <a href=\"https://www.kaggle.com/tunguz/bengali-2d-umap\">here</a>.</p>\n\n<p><img src=\"https://pbs.twimg.com/media/ETbVYlJWoAENXK6?format=png&amp;name=small\" alt=\"bengali_umap\"></p>",
      "rawMarkdown": "At one point I decided to make UMAP embeddings of the original data. It turns out this was a huge undertaking, and the final UMAP model was 250 GB. It took 2.5 days on a 16 core CPU / 128 GB machine to compute. Due to its size, we did not use it, but the embedding can be found [here](https://www.kaggle.com/tunguz/bengali-umap-2d-embedding).\n\nA simple notebook that generates the UMAP model and the files can be found [here](https://www.kaggle.com/tunguz/bengali-2d-umap).\n\n![bengali_umap](https://pbs.twimg.com/media/ETbVYlJWoAENXK6?format=png&amp;name=small)",
      "votes": null
    },
    {
      "id": "778991",
      "postDate": "03/18/2020 23:17:08",
      "content": "<p>Thank you for sharing!</p>\n\n<p>&gt; the final UMAP model was 250 GB. It took 2.5 hours on a 16 core CPU / 128 GB</p>\n\n<p>😂 </p>",
      "rawMarkdown": "Thank you for sharing!\n\n&gt; the final UMAP model was 250 GB. It took 2.5 hours on a 16 core CPU / 128 GB\n\n😂",
      "votes": null
    },
    {
      "id": "778996",
      "postDate": "03/18/2020 23:21:14",
      "content": "<p>Actually, that was a mistake. It took 2.5 days. 😄 </p>",
      "rawMarkdown": "Actually, that was a mistake. It took 2.5 days. 😄",
      "votes": null
    },
    {
      "id": "779004",
      "postDate": "03/18/2020 23:31:48",
      "content": "<p>Hahaha ... 2.5 days?! OMG😂 </p>",
      "rawMarkdown": "Hahaha ... 2.5 days?! OMG😂",
      "votes": null
    },
    {
      "id": "779323",
      "postDate": "03/19/2020 07:50:37",
      "content": "<p><a href=\"/tunguz\">@tunguz</a> I'm sure you guys at Rapids will find a way to make this on the GPU soon? ;)</p>",
      "rawMarkdown": "tunguz I'm sure you guys at Rapids will find a way to make this on the GPU soon? ;)",
      "votes": null
    },
    {
      "id": "2978368",
      "postDate": "09/03/2024 19:44:53",
      "content": "<p>It's a…blob!</p>",
      "rawMarkdown": "It's a...blob!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2978368,
      "author_name": "scharlesworth",
      "author_url": "",
      "post_date": "09/03/2024 19:44:53",
      "content": "<p>It's a…blob!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 778991,
      "author_name": "ttahara",
      "author_url": "",
      "post_date": "03/18/2020 23:17:08",
      "content": "<p>Thank you for sharing!</p>\n\n<p>&gt; the final UMAP model was 250 GB. It took 2.5 hours on a 16 core CPU / 128 GB</p>\n\n<p>😂 </p>",
      "votes": null,
      "replies": [
        {
          "id": 778996,
          "author_name": "tunguz",
          "author_url": "",
          "post_date": "03/18/2020 23:21:14",
          "content": "<p>Actually, that was a mistake. It took 2.5 days. 😄 </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 779004,
          "author_name": "ttahara",
          "author_url": "",
          "post_date": "03/18/2020 23:31:48",
          "content": "<p>Hahaha ... 2.5 days?! OMG😂 </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 779323,
          "author_name": "maxlenormand",
          "author_url": "",
          "post_date": "03/19/2020 07:50:37",
          "content": "<p><a href=\"/tunguz\">@tunguz</a> I'm sure you guys at Rapids will find a way to make this on the GPU soon? ;)</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "778989": "At one point I decided to make UMAP embeddings of the original data. It turns out this was a huge undertaking, and the final UMAP model was 250 GB. It took 2.5 days on a 16 core CPU / 128 GB machine to compute. Due to its size, we did not use it, but the embedding can be found [here](https://www.kaggle.com/tunguz/bengali-umap-2d-embedding).\n\nA simple notebook that generates the UMAP model and the files can be found [here](https://www.kaggle.com/tunguz/bengali-2d-umap).\n\n![bengali_umap](https://pbs.twimg.com/media/ETbVYlJWoAENXK6?format=png&amp;name=small)",
    "778991": "Thank you for sharing!\n\n&gt; the final UMAP model was 250 GB. It took 2.5 hours on a 16 core CPU / 128 GB\n\n😂",
    "778996": "Actually, that was a mistake. It took 2.5 days. 😄",
    "779004": "Hahaha ... 2.5 days?! OMG😂",
    "779323": "tunguz I'm sure you guys at Rapids will find a way to make this on the GPU soon? ;)",
    "2978368": "It's a...blob!"
  },
  "source": "meta"
}