{
  "id": 125492,
  "title": "How Deep will you GO??",
  "url": "/competitions/bengaliai-cv19/discussion/125492",
  "author_name": "",
  "post_date": "2020-01-11T04:56:19.667467700Z",
  "votes": 3,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi Kagglers!!</p>\n\n<p>I was trying out ResNet with various depth (than using default from Keras)!!, I tired with 20, 40, 50 layers deep!! till now 50 got me the highest LB of 0.96 trained for 25 epochs, How deep did you go to get your LB??\nthanks in advance</p>",
  "messages": [
    {
      "id": "715983",
      "postDate": "01/11/2020 04:56:19",
      "content": "<p>Hi Kagglers!!</p>\n\n<p>I was trying out ResNet with various depth (than using default from Keras)!!, I tired with 20, 40, 50 layers deep!! till now 50 got me the highest LB of 0.96 trained for 25 epochs, How deep did you go to get your LB??\nthanks in advance</p>",
      "rawMarkdown": "Hi Kagglers!!\n\nI was trying out ResNet with various depth (than using default from Keras)!!, I tired with 20, 40, 50 layers deep!! till now 50 got me the highest LB of 0.96 trained for 25 epochs, How deep did you go to get your LB??\nthanks in advance",
      "votes": null
    },
    {
      "id": "716071",
      "postDate": "01/11/2020 07:52:09",
      "content": "<p>This is an interesting thread!! I'm also curious about how deep we need to go. So far, deeper models have performed better in my experiments; the best one being <code>serex50</code></p>",
      "rawMarkdown": "This is an interesting thread!! I'm also curious about how deep we need to go. So far, deeper models have performed better in my experiments; the best one being `serex50`",
      "votes": null
    },
    {
      "id": "717455",
      "postDate": "01/13/2020 06:17:38",
      "content": "<p>Hi Adarsh, good experimentation! As an alternative you can maybe try training for more epochs even at the cost of depth. It worked for me. Let us know if you try it out and whether that worked for you too :)</p>",
      "rawMarkdown": "Hi Adarsh, good experimentation! As an alternative you can maybe try training for more epochs even at the cost of depth. It worked for me. Let us know if you try it out and whether that worked for you too :)",
      "votes": null
    },
    {
      "id": "723128",
      "postDate": "01/19/2020 14:40:09",
      "content": "<p>keep up the good work!  upvoted..</p>",
      "rawMarkdown": "keep up the good work!  upvoted..",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 716071,
      "author_name": "bibek777",
      "author_url": "",
      "post_date": "01/11/2020 07:52:09",
      "content": "<p>This is an interesting thread!! I'm also curious about how deep we need to go. So far, deeper models have performed better in my experiments; the best one being <code>serex50</code></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 717455,
      "author_name": "mightyrains",
      "author_url": "",
      "post_date": "01/13/2020 06:17:38",
      "content": "<p>Hi Adarsh, good experimentation! As an alternative you can maybe try training for more epochs even at the cost of depth. It worked for me. Let us know if you try it out and whether that worked for you too :)</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 723128,
      "author_name": "",
      "author_url": "",
      "post_date": "01/19/2020 14:40:09",
      "content": "<p>keep up the good work!  upvoted..</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "715983": "Hi Kagglers!!\n\nI was trying out ResNet with various depth (than using default from Keras)!!, I tired with 20, 40, 50 layers deep!! till now 50 got me the highest LB of 0.96 trained for 25 epochs, How deep did you go to get your LB??\nthanks in advance",
    "716071": "This is an interesting thread!! I'm also curious about how deep we need to go. So far, deeper models have performed better in my experiments; the best one being `serex50`",
    "717455": "Hi Adarsh, good experimentation! As an alternative you can maybe try training for more epochs even at the cost of depth. It worked for me. Let us know if you try it out and whether that worked for you too :)",
    "723128": "keep up the good work!  upvoted.."
  },
  "source": "meta"
}