{
  "id": 111305,
  "title": "EfficientNets are now available in pytorch segmentation model repo.",
  "url": "/competitions/understanding_cloud_organization/discussion/111305",
  "author_name": "",
  "post_date": "2019-10-04T17:09:38.134668400Z",
  "votes": 8,
  "comment_count": 4,
  "views": 0,
  "content": "<p>looks like it was added yesterday. \nEnjoy!\nrepo :- <a href=\"https://github.com/qubvel/segmentation_models.pytorch\">link</a></p>",
  "messages": [
    {
      "id": "641415",
      "postDate": "10/04/2019 17:09:38",
      "content": "<p>looks like it was added yesterday. \nEnjoy!\nrepo :- <a href=\"https://github.com/qubvel/segmentation_models.pytorch\">link</a></p>",
      "rawMarkdown": "looks like it was added yesterday. \nEnjoy!\nrepo :- [link](https://github.com/qubvel/segmentation_models.pytorch)",
      "votes": null
    },
    {
      "id": "641533",
      "postDate": "10/04/2019 19:19:19",
      "content": "<p>Not so much interested after seeing this issue comments : <a href=\"https://github.com/qubvel/segmentation_models.pytorch/issues/19\">https://github.com/qubvel/segmentation_models.pytorch/issues/19</a> almost 1 month ago.don't know if his newly released efficientnet worked better than he said 4 months ago</p>",
      "rawMarkdown": "Not so much interested after seeing this issue comments : https://github.com/qubvel/segmentation_models.pytorch/issues/19 almost 1 month ago.don't know if his newly released efficientnet worked better than he said 4 months ago",
      "votes": null
    },
    {
      "id": "641536",
      "postDate": "10/04/2019 19:22:36",
      "content": "<p>Just tried them and must say results are disappointing. </p>",
      "rawMarkdown": "Just tried them and must say results are disappointing.",
      "votes": null
    },
    {
      "id": "641538",
      "postDate": "10/04/2019 19:25:28",
      "content": "<p>Sad to know :(</p>",
      "rawMarkdown": "Sad to know :(",
      "votes": null
    },
    {
      "id": "641966",
      "postDate": "10/05/2019 11:33:48",
      "content": "<p>Quoting <a href=\"/hengck23\">@hengck23</a> from <a href=\"https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/111110#latest-641411\">here</a>:</p>\n\n<blockquote>\n  <p>My efficientnet b3 and b5 with fpn gives good results. Input size is crop of 256x400. You may want to use this as reference.</p>\n</blockquote>\n\n<p>In terms of the validation loss, the EfficientNet Models I tried (B1, B2) are comparable to Resnet Architectures.</p>",
      "rawMarkdown": "Quoting @hengck23 from [here](https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/111110#latest-641411):\n&gt; My efficientnet b3 and b5 with fpn gives good results. Input size is crop of 256x400. You may want to use this as reference.\n\nIn terms of the validation loss, the EfficientNet Models I tried (B1, B2) are comparable to Resnet Architectures.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 641533,
      "author_name": "mobassir",
      "author_url": "",
      "post_date": "10/04/2019 19:19:19",
      "content": "<p>Not so much interested after seeing this issue comments : <a href=\"https://github.com/qubvel/segmentation_models.pytorch/issues/19\">https://github.com/qubvel/segmentation_models.pytorch/issues/19</a> almost 1 month ago.don't know if his newly released efficientnet worked better than he said 4 months ago</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 641536,
      "author_name": "dhananjay3",
      "author_url": "",
      "post_date": "10/04/2019 19:22:36",
      "content": "<p>Just tried them and must say results are disappointing. </p>",
      "votes": null,
      "replies": [
        {
          "id": 641538,
          "author_name": "mobassir",
          "author_url": "",
          "post_date": "10/04/2019 19:25:28",
          "content": "<p>Sad to know :(</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 641966,
          "author_name": "maxjeblick",
          "author_url": "",
          "post_date": "10/05/2019 11:33:48",
          "content": "<p>Quoting <a href=\"/hengck23\">@hengck23</a> from <a href=\"https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/111110#latest-641411\">here</a>:</p>\n\n<blockquote>\n  <p>My efficientnet b3 and b5 with fpn gives good results. Input size is crop of 256x400. You may want to use this as reference.</p>\n</blockquote>\n\n<p>In terms of the validation loss, the EfficientNet Models I tried (B1, B2) are comparable to Resnet Architectures.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "641415": "looks like it was added yesterday. \nEnjoy!\nrepo :- [link](https://github.com/qubvel/segmentation_models.pytorch)",
    "641533": "Not so much interested after seeing this issue comments : https://github.com/qubvel/segmentation_models.pytorch/issues/19 almost 1 month ago.don't know if his newly released efficientnet worked better than he said 4 months ago",
    "641536": "Just tried them and must say results are disappointing.",
    "641538": "Sad to know :(",
    "641966": "Quoting @hengck23 from [here](https://www.kaggle.com/c/severstal-steel-defect-detection/discussion/111110#latest-641411):\n&gt; My efficientnet b3 and b5 with fpn gives good results. Input size is crop of 256x400. You may want to use this as reference.\n\nIn terms of the validation loss, the EfficientNet Models I tried (B1, B2) are comparable to Resnet Architectures."
  },
  "source": "meta"
}