{
  "id": 104284,
  "title": "Debugging Neural Net by Gradient Visualization",
  "url": "/competitions/recursion-cellular-image-classification/discussion/104284",
  "author_name": "wjsheng",
  "post_date": "2019-08-15T18:53:18.628000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>Hi guys, I wonder how have experts been using Tensorboard to debug or tune their model by visualizing the gradients? Is it anyhow helpful? What are we expected to see? </p>\n\n<p>I notice my <code>conv1</code> layer is similar to a perfect normal distribution and has not changed much every epoch compared to other layers. I wonder whether is it training correctly.</p>\n\n<p>Any links or inputs are appreciated. Thank you! :) </p>",
  "messages": [
    {
      "id": 600175,
      "postDate": "2019-08-15T18:53:18.627Z",
      "content": "<p>Hi guys, I wonder how have experts been using Tensorboard to debug or tune their model by visualizing the gradients? Is it anyhow helpful? What are we expected to see? </p>\n\n<p>I notice my <code>conv1</code> layer is similar to a perfect normal distribution and has not changed much every epoch compared to other layers. I wonder whether is it training correctly.</p>\n\n<p>Any links or inputs are appreciated. Thank you! :) </p>",
      "rawMarkdown": "Hi guys, I wonder how have experts been using Tensorboard to debug or tune their model by visualizing the gradients? Is it anyhow helpful? What are we expected to see? \n\nI notice my `conv1` layer is similar to a perfect normal distribution and has not changed much every epoch compared to other layers. I wonder whether is it training correctly.\n\nAny links or inputs are appreciated. Thank you! :) ",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "600175": "Hi guys, I wonder how have experts been using Tensorboard to debug or tune their model by visualizing the gradients? Is it anyhow helpful? What are we expected to see? \n\nI notice my `conv1` layer is similar to a perfect normal distribution and has not changed much every epoch compared to other layers. I wonder whether is it training correctly.\n\nAny links or inputs are appreciated. Thank you! :) "
  }
}