{
  "id": 409419,
  "title": "Can you build a local evaluation pipeline?",
  "url": "/competitions/image-matching-challenge-2023/discussion/409419",
  "author_name": "Yuxiang Huang",
  "post_date": "2023-05-10T23:06:01.483000",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I wonder if/how we can build a local evaluation pipeline to test our method locally before submission? Submissions to Kaggle take too long to run. </p>",
  "messages": [
    {
      "id": 2254390,
      "postDate": "2023-05-10T23:06:01.483Z",
      "content": "<p>I wonder if/how we can build a local evaluation pipeline to test our method locally before submission? Submissions to Kaggle take too long to run. </p>",
      "rawMarkdown": "I wonder if/how we can build a local evaluation pipeline to test our method locally before submission? Submissions to Kaggle take too long to run. ",
      "votes": 1
    },
    {
      "id": 2254393,
      "postDate": "2023-05-10T23:11:57.927Z",
      "content": "<p>You can download the data locally from <a href=\"https://www.kaggle.com/competitions/image-matching-challenge-2023/data\" target=\"_blank\">here</a>. It's already split into train/test. Then use the <a href=\"https://www.kaggle.com/competitions/image-matching-challenge-2023/overview/evaluation\" target=\"_blank\">evaluation metric</a> to evaluate your predictions on the test set locally. </p>",
      "rawMarkdown": "You can download the data locally from [here](https://www.kaggle.com/competitions/image-matching-challenge-2023/data). It's already split into train/test. Then use the [evaluation metric](https://www.kaggle.com/competitions/image-matching-challenge-2023/overview/evaluation) to evaluate your predictions on the test set locally. ",
      "replies": [
        {
          "id": 2266822,
          "postDate": "2023-05-20T11:50:10.227Z",
          "rawMarkdown": "",
          "isDeleted": true
        }
      ]
    },
    {
      "id": 2254619,
      "postDate": "2023-05-11T05:44:50.407Z",
      "content": "<p>You definitely have to do that for iterating faster.</p>",
      "rawMarkdown": "You definitely have to do that for iterating faster.",
      "replies": [
        {
          "id": 2266824,
          "postDate": "2023-05-20T11:50:33.593Z",
          "content": "<p>So this only label file you can use is the 328 labels in the train_labels.csv, right? And it seems there's no label for the phototourism dataset, or am I missing it?</p>",
          "rawMarkdown": "So this only label file you can use is the 328 labels in the train_labels.csv, right? And it seems there's no label for the phototourism dataset, or am I missing it?"
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 2254393,
      "author_name": "Joakim Arvidsson",
      "author_url": "",
      "post_date": "2023-05-10T23:11:57.927000",
      "content": "<p>You can download the data locally from <a href=\"https://www.kaggle.com/competitions/image-matching-challenge-2023/data\" target=\"_blank\">here</a>. It's already split into train/test. Then use the <a href=\"https://www.kaggle.com/competitions/image-matching-challenge-2023/overview/evaluation\" target=\"_blank\">evaluation metric</a> to evaluate your predictions on the test set locally. </p>",
      "votes": 0,
      "replies": [
        {
          "id": 2266822,
          "author_name": "",
          "author_url": "",
          "post_date": "2023-05-20T11:50:10.227000",
          "content": "",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 2254619,
      "author_name": "Gunes Evitan",
      "author_url": "",
      "post_date": "2023-05-11T05:44:50.407000",
      "content": "<p>You definitely have to do that for iterating faster.</p>",
      "votes": 0,
      "replies": [
        {
          "id": 2266824,
          "author_name": "Yuxiang Huang",
          "author_url": "",
          "post_date": "2023-05-20T11:50:33.593000",
          "content": "<p>So this only label file you can use is the 328 labels in the train_labels.csv, right? And it seems there's no label for the phototourism dataset, or am I missing it?</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2254390": "I wonder if/how we can build a local evaluation pipeline to test our method locally before submission? Submissions to Kaggle take too long to run. ",
    "2254393": "You can download the data locally from [here](https://www.kaggle.com/competitions/image-matching-challenge-2023/data). It's already split into train/test. Then use the [evaluation metric](https://www.kaggle.com/competitions/image-matching-challenge-2023/overview/evaluation) to evaluate your predictions on the test set locally. ",
    "2254619": "You definitely have to do that for iterating faster."
  }
}