{
  "id": 103249,
  "title": "Improving GAN training by progressive augmentation",
  "url": "/competitions/generative-dog-images/discussion/103249",
  "author_name": "Pranav Mahajan",
  "post_date": "2019-08-08T06:01:48.306000",
  "votes": 1,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Hey, has anyone tried Progressive Augmenting in GANs? (<a href=\"https://arxiv.org/pdf/1901.10422.pdf\">https://arxiv.org/pdf/1901.10422.pdf</a>). I've been trying to implement but unable to figure out a few details, moreover if anyone has used, did it stabilize the training process well?</p>",
  "messages": [
    {
      "id": 594517,
      "postDate": "2019-08-08T06:01:48.307Z",
      "content": "<p>Hey, has anyone tried Progressive Augmenting in GANs? (<a href=\"https://arxiv.org/pdf/1901.10422.pdf\">https://arxiv.org/pdf/1901.10422.pdf</a>). I've been trying to implement but unable to figure out a few details, moreover if anyone has used, did it stabilize the training process well?</p>",
      "rawMarkdown": "Hey, has anyone tried Progressive Augmenting in GANs? (https://arxiv.org/pdf/1901.10422.pdf). I've been trying to implement but unable to figure out a few details, moreover if anyone has used, did it stabilize the training process well?",
      "votes": 1
    },
    {
      "id": 595365,
      "postDate": "2019-08-09T06:19:12.697Z",
      "content": "<p>Tried to open the link but got an error:\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F694899%2F94a2dad59ec3f94430fac03f1b9ac0cd%2FScreen%20Shot%202019-08-09%20at%208.16.03%20AM.png?generation=1565331521399849&amp;alt=media\" alt=\"\"></p>\n\n<p>Probably you refer to <a href=\"https://arxiv.org/pdf/1901.10422.pdf\">this paper</a>?</p>",
      "rawMarkdown": "Tried to open the link but got an error:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F694899%2F94a2dad59ec3f94430fac03f1b9ac0cd%2FScreen%20Shot%202019-08-09%20at%208.16.03%20AM.png?generation=1565331521399849&amp;alt=media)\n\n\nProbably you refer to [this paper](https://arxiv.org/pdf/1901.10422.pdf)?",
      "replies": [
        {
          "id": 596053,
          "postDate": "2019-08-10T05:08:50.113Z",
          "content": "<p>Yes, I'm referring to the same paper. Thanks for pointing out, fixed the broken link.</p>",
          "rawMarkdown": "Yes, I'm referring to the same paper. Thanks for pointing out, fixed the broken link."
        }
      ]
    },
    {
      "id": 595493,
      "postDate": "2019-08-09T09:34:32.323Z",
      "rawMarkdown": "",
      "isDeleted": true,
      "replies": [
        {
          "id": 595558,
          "postDate": "2019-08-09T11:22:44.980Z",
          "content": "<p>Is the <em>Progressive Augmenting</em> technique is the same as <em>Progressive GAN</em>? Or similar to? Because I believe that what you've described is the later one. Though it also uses \"blending\" approach to slowly fade-in new layers to keep the previously trained weights in consistent state.</p>",
          "rawMarkdown": "Is the _Progressive Augmenting_ technique is the same as _Progressive GAN_? Or similar to? Because I believe that what you've described is the later one. Though it also uses \"blending\" approach to slowly fade-in new layers to keep the previously trained weights in consistent state."
        },
        {
          "id": 596055,
          "postDate": "2019-08-10T05:11:13.607Z",
          "content": "<p>Yup, I believe Progressive GANs are different from Progressive Augmenting GANs. At the moment I'm trying to make progressive augmenting work in GANs.  </p>",
          "rawMarkdown": "Yup, I believe Progressive GANs are different from Progressive Augmenting GANs. At the moment I'm trying to make progressive augmenting work in GANs.  "
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 595365,
      "author_name": "Ilia Zaitsev",
      "author_url": "",
      "post_date": "2019-08-09T06:19:12.697000",
      "content": "<p>Tried to open the link but got an error:\n<img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F694899%2F94a2dad59ec3f94430fac03f1b9ac0cd%2FScreen%20Shot%202019-08-09%20at%208.16.03%20AM.png?generation=1565331521399849&amp;alt=media\" alt=\"\"></p>\n\n<p>Probably you refer to <a href=\"https://arxiv.org/pdf/1901.10422.pdf\">this paper</a>?</p>",
      "votes": 0,
      "replies": [
        {
          "id": 596053,
          "author_name": "Pranav Mahajan",
          "author_url": "",
          "post_date": "2019-08-10T05:08:50.113000",
          "content": "<p>Yes, I'm referring to the same paper. Thanks for pointing out, fixed the broken link.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 595493,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-08-09T09:34:32.323000",
      "content": "",
      "votes": 0,
      "replies": [
        {
          "id": 595558,
          "author_name": "Ilia Zaitsev",
          "author_url": "",
          "post_date": "2019-08-09T11:22:44.980000",
          "content": "<p>Is the <em>Progressive Augmenting</em> technique is the same as <em>Progressive GAN</em>? Or similar to? Because I believe that what you've described is the later one. Though it also uses \"blending\" approach to slowly fade-in new layers to keep the previously trained weights in consistent state.</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 596055,
          "author_name": "Pranav Mahajan",
          "author_url": "",
          "post_date": "2019-08-10T05:11:13.607000",
          "content": "<p>Yup, I believe Progressive GANs are different from Progressive Augmenting GANs. At the moment I'm trying to make progressive augmenting work in GANs.  </p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "594517": "Hey, has anyone tried Progressive Augmenting in GANs? (https://arxiv.org/pdf/1901.10422.pdf). I've been trying to implement but unable to figure out a few details, moreover if anyone has used, did it stabilize the training process well?",
    "595365": "Tried to open the link but got an error:\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F694899%2F94a2dad59ec3f94430fac03f1b9ac0cd%2FScreen%20Shot%202019-08-09%20at%208.16.03%20AM.png?generation=1565331521399849&amp;alt=media)\n\n\nProbably you refer to [this paper](https://arxiv.org/pdf/1901.10422.pdf)?",
    "595493": ""
  }
}