{
  "id": 109789,
  "title": "Using High-level frameworks is not learner friendly",
  "url": "/competitions/understanding_cloud_organization/discussion/109789",
  "author_name": "",
  "post_date": "2019-09-22T10:37:09.663059600Z",
  "votes": 14,
  "comment_count": 3,
  "views": 0,
  "content": "<p>This is my first hands on with image segmentation and I tried to learn from existing pytorch notebooks. One thing I immediately noticed is Using High level frameworks like catalyst is very convenient but <strong>For learner like me who came here to learn it is difficult to know what is happening under the hood</strong> and that's what we are here for, that is what we want to learn. <br>\nso I wrote <a href=\"https://www.kaggle.com/dhananjay3/image-segmentation-from-scratch-in-pytorch\">this kernel</a>. It does all things from scratch so we can see what's happening.  This is was not going to be my best kernel but I compiled it hoping someone will benefit from it. Posting here because otherwise it will get buried down other high scoring kernels and many beginners don't scroll that down.</p>",
  "messages": [
    {
      "id": "631610",
      "postDate": "09/22/2019 10:37:09",
      "content": "<p>This is my first hands on with image segmentation and I tried to learn from existing pytorch notebooks. One thing I immediately noticed is Using High level frameworks like catalyst is very convenient but <strong>For learner like me who came here to learn it is difficult to know what is happening under the hood</strong> and that's what we are here for, that is what we want to learn. <br>\nso I wrote <a href=\"https://www.kaggle.com/dhananjay3/image-segmentation-from-scratch-in-pytorch\">this kernel</a>. It does all things from scratch so we can see what's happening.  This is was not going to be my best kernel but I compiled it hoping someone will benefit from it. Posting here because otherwise it will get buried down other high scoring kernels and many beginners don't scroll that down.</p>",
      "rawMarkdown": "This is my first hands on with image segmentation and I tried to learn from existing pytorch notebooks. One thing I immediately noticed is Using High level frameworks like catalyst is very convenient but **For learner like me who came here to learn it is difficult to know what is happening under the hood** and that's what we are here for, that is what we want to learn.   \nso I wrote [this kernel](https://www.kaggle.com/dhananjay3/image-segmentation-from-scratch-in-pytorch). It does all things from scratch so we can see what's happening.  This is was not going to be my best kernel but I compiled it hoping someone will benefit from it. Posting here because otherwise it will get buried down other high scoring kernels and many beginners don't scroll that down.",
      "votes": null
    },
    {
      "id": "631727",
      "postDate": "09/22/2019 14:55:06",
      "content": "<p>I think many learners like you and me will benefit from the kernels like this one. Good job, man!</p>",
      "rawMarkdown": "I think many learners like you and me will benefit from the kernels like this one. Good job, man!",
      "votes": null
    },
    {
      "id": "633417",
      "postDate": "09/24/2019 21:57:19",
      "content": "<p>Thanks for your work, I find it useful </p>",
      "rawMarkdown": "Thanks for your work, I find it useful",
      "votes": null
    },
    {
      "id": "635089",
      "postDate": "09/27/2019 06:36:47",
      "content": "<p>Instead of copying writing same code line by line gives good reasoning behind code, It worked for me to understand High level code.</p>",
      "rawMarkdown": "Instead of copying writing same code line by line gives good reasoning behind code, It worked for me to understand High level code.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 631727,
      "author_name": "phunghieu",
      "author_url": "",
      "post_date": "09/22/2019 14:55:06",
      "content": "<p>I think many learners like you and me will benefit from the kernels like this one. Good job, man!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 633417,
      "author_name": "errorinever",
      "author_url": "",
      "post_date": "09/24/2019 21:57:19",
      "content": "<p>Thanks for your work, I find it useful </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 635089,
      "author_name": "yogeshmahawar",
      "author_url": "",
      "post_date": "09/27/2019 06:36:47",
      "content": "<p>Instead of copying writing same code line by line gives good reasoning behind code, It worked for me to understand High level code.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "631610": "This is my first hands on with image segmentation and I tried to learn from existing pytorch notebooks. One thing I immediately noticed is Using High level frameworks like catalyst is very convenient but **For learner like me who came here to learn it is difficult to know what is happening under the hood** and that's what we are here for, that is what we want to learn.   \nso I wrote [this kernel](https://www.kaggle.com/dhananjay3/image-segmentation-from-scratch-in-pytorch). It does all things from scratch so we can see what's happening.  This is was not going to be my best kernel but I compiled it hoping someone will benefit from it. Posting here because otherwise it will get buried down other high scoring kernels and many beginners don't scroll that down.",
    "631727": "I think many learners like you and me will benefit from the kernels like this one. Good job, man!",
    "633417": "Thanks for your work, I find it useful",
    "635089": "Instead of copying writing same code line by line gives good reasoning behind code, It worked for me to understand High level code."
  },
  "source": "meta"
}