{
  "id": 413212,
  "title": "What are numbers in t5embeds/train_embeds.npy ?",
  "url": "/competitions/cafa-5-protein-function-prediction/discussion/413212",
  "author_name": "",
  "post_date": "2023-05-27T14:12:49.406971700Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Are them the same as those proteins in train_sequences.fasta? If yes, where are them from? I am new to this field, thanks!</p>",
  "messages": [
    {
      "id": "2277188",
      "postDate": "05/27/2023 14:12:49",
      "content": "<p>Are them the same as those proteins in train_sequences.fasta? If yes, where are them from? I am new to this field, thanks!</p>",
      "rawMarkdown": "Are them the same as those proteins in train_sequences.fasta? If yes, where are them from? I am new to this field, thanks!",
      "votes": null
    },
    {
      "id": "2281342",
      "postDate": "05/30/2023 19:31:48",
      "content": "<p>Those numbers are embedding vectors derived from protein language models. Just like word embeddings for <code>king</code> and <code>queen</code> are close to each other, similar proteins will have embeddings with a relatively small Euclidean distance in high-dimensional space. The same is true for low-dimensional space - see <a href=\"https://www.kaggle.com/competitions/cafa-5-protein-function-prediction/discussion/406573\" target=\"_blank\"><strong>here</strong></a>.</p>",
      "rawMarkdown": "Those numbers are embedding vectors derived from protein language models. Just like word embeddings for `king` and `queen` are close to each other, similar proteins will have embeddings with a relatively small Euclidean distance in high-dimensional space. The same is true for low-dimensional space - see [**here**](https://www.kaggle.com/competitions/cafa-5-protein-function-prediction/discussion/406573).",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2281342,
      "author_name": "tilii7",
      "author_url": "",
      "post_date": "05/30/2023 19:31:48",
      "content": "<p>Those numbers are embedding vectors derived from protein language models. Just like word embeddings for <code>king</code> and <code>queen</code> are close to each other, similar proteins will have embeddings with a relatively small Euclidean distance in high-dimensional space. The same is true for low-dimensional space - see <a href=\"https://www.kaggle.com/competitions/cafa-5-protein-function-prediction/discussion/406573\" target=\"_blank\"><strong>here</strong></a>.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2277188": "Are them the same as those proteins in train_sequences.fasta? If yes, where are them from? I am new to this field, thanks!",
    "2281342": "Those numbers are embedding vectors derived from protein language models. Just like word embeddings for `king` and `queen` are close to each other, similar proteins will have embeddings with a relatively small Euclidean distance in high-dimensional space. The same is true for low-dimensional space - see [**here**](https://www.kaggle.com/competitions/cafa-5-protein-function-prediction/discussion/406573)."
  },
  "source": "meta"
}