{
  "id": 585273,
  "title": "He jumped from 25.1 to 16.1👏",
  "url": "/competitions/waveform-inversion/discussion/585273",
  "author_name": "",
  "post_date": "2025-06-19T06:49:23.852592300Z",
  "votes": 20,
  "comment_count": 30,
  "views": 0,
  "content": "<p>Huge applause to greySnow!<br>\nAnd Harshit Sheoran, you haven't submitted for nearly 3 weaks, are you planning  to jump to sub 10? :)</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6624777%2F295e191241ff0a1bf8040267e4ed0b89%2F2025-06-19%2015.47.29.png?generation=1750315661185941&amp;alt=media\" alt=\"\"></p>",
  "messages": [
    {
      "id": "3227650",
      "postDate": "06/19/2025 06:49:23",
      "content": "<p>Huge applause to greySnow!<br>\nAnd Harshit Sheoran, you haven't submitted for nearly 3 weaks, are you planning  to jump to sub 10? :)</p>\n<p><img src=\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6624777%2F295e191241ff0a1bf8040267e4ed0b89%2F2025-06-19%2015.47.29.png?generation=1750315661185941&amp;alt=media\" alt=\"\"></p>",
      "rawMarkdown": "Huge applause to greySnow!\nAnd Harshit Sheoran, you haven't submitted for nearly 3 weaks, are you planning  to jump to sub 10? :)\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6624777%2F295e191241ff0a1bf8040267e4ed0b89%2F2025-06-19%2015.47.29.png?generation=1750315661185941&alt=media)",
      "votes": null
    },
    {
      "id": "3227661",
      "postDate": "06/19/2025 06:54:57",
      "content": "<p>Good job, greySnow! </p>",
      "rawMarkdown": "Good job, greySnow!",
      "votes": null
    },
    {
      "id": "3227662",
      "postDate": "06/19/2025 06:56:41",
      "content": "<p>Yup, amazing performance by greySnow! like a week ago he was ~550th, with lots of white noise between him and gold, he has cleared it all :D</p>\n<p>Training ends tonight! Sub incoming in next 24 hours :D, I hope my LB follows my CV 🤞</p>",
      "rawMarkdown": "Yup, amazing performance by greySnow! like a week ago he was ~550th, with lots of white noise between him and gold, he has cleared it all :D\n\nTraining ends tonight! Sub incoming in next 24 hours :D, I hope my LB follows my CV 🤞",
      "votes": null
    },
    {
      "id": "3227666",
      "postDate": "06/19/2025 06:58:44",
      "content": "<p>Thx. Knowing you, you will probably appear from thin air in gold range on last day :)</p>",
      "rawMarkdown": "Thx. Knowing you, you will probably appear from thin air in gold range on last day :)",
      "votes": null
    },
    {
      "id": "3227671",
      "postDate": "06/19/2025 07:00:14",
      "content": "<p>Nah, I was just too lazy to submit lol<br>\nYou going to jump to sub-10? :D</p>",
      "rawMarkdown": "Nah, I was just too lazy to submit lol\nYou going to jump to sub-10? :D",
      "votes": null
    },
    {
      "id": "3227674",
      "postDate": "06/19/2025 07:07:32",
      "content": "<p>We all knew greySnow wasn't here to just post usefull discussions.</p>\n<p>Looking so forward to your next sub!</p>",
      "rawMarkdown": "We all knew greySnow wasn't here to just post usefull discussions.\n\nLooking so forward to your next sub!",
      "votes": null
    },
    {
      "id": "3227675",
      "postDate": "06/19/2025 07:08:42",
      "content": "<p>Honestly, even if my LB follows my CV like it did for 17.1 (CV 16.1), it's going to be a close call for the sub-10/single digit score and it's probably not going to be quite sub-10.</p>\n<p>My training is going to take even more time, Sub <em>probably not</em> incoming tommorow</p>",
      "rawMarkdown": "Honestly, even if my LB follows my CV like it did for 17.1 (CV 16.1), it's going to be a close call for the sub-10/single digit score and it's probably not going to be quite sub-10.\n\nMy training is going to take even more time, Sub *probably not* incoming tommorow",
      "votes": null
    },
    {
      "id": "3227677",
      "postDate": "06/19/2025 07:13:15",
      "content": "<p>I am also too lazy to submit, but i am still stuggling with gold line.</p>",
      "rawMarkdown": "I am also too lazy to submit, but i am still stuggling with gold line.",
      "votes": null
    },
    {
      "id": "3227956",
      "postDate": "06/19/2025 13:31:02",
      "content": "<p>I can't help but wonder — what kind of machines are you all running? Surely not P100s, right? 😅 Feels like there's some secret hardware magic going on here!</p>",
      "rawMarkdown": "I can't help but wonder — what kind of machines are you all running? Surely not P100s, right? 😅 Feels like there's some secret hardware magic going on here!",
      "votes": null
    },
    {
      "id": "3228183",
      "postDate": "06/19/2025 20:23:56",
      "content": "<p><a href=\"https://www.kaggle.com/tomirol\" target=\"_blank\">@tomirol</a> getting 17.x at first sub is also remarkable.</p>",
      "rawMarkdown": "tomirol getting 17.x at first sub is also remarkable.",
      "votes": null
    },
    {
      "id": "3228185",
      "postDate": "06/19/2025 20:39:40",
      "content": "<p>Hey thank you <a href=\"https://www.kaggle.com/cpmpml\" target=\"_blank\">@cpmpml</a>! I was only able to do that because so many people shared a lot of high quality stuff here (including you), and validation results as well. Honestly makes me feel like I need to share more in other comps too!</p>",
      "rawMarkdown": "Hey thank you @cpmpml! I was only able to do that because so many people shared a lot of high quality stuff here (including you), and validation results as well. Honestly makes me feel like I need to share more in other comps too!",
      "votes": null
    },
    {
      "id": "3228235",
      "postDate": "06/20/2025 00:50:28",
      "content": "<p>I would aspire to do the same but training is so much hahaha </p>",
      "rawMarkdown": "I would aspire to do the same but training is so much hahaha",
      "votes": null
    },
    {
      "id": "3228531",
      "postDate": "06/20/2025 08:58:48",
      "content": "<p>Huge congrats to greySnow for that massive leap — 25.1 to 16.1 is no joke! 🔥 Well deserved gold medal. And yep, curious to see if Harshit is quietly cooking something big for that sub-10 breakthrough 😄</p>",
      "rawMarkdown": "Huge congrats to greySnow for that massive leap — 25.1 to 16.1 is no joke! 🔥 Well deserved gold medal. And yep, curious to see if Harshit is quietly cooking something big for that sub-10 breakthrough 😄",
      "votes": null
    },
    {
      "id": "3228533",
      "postDate": "06/20/2025 08:59:35",
      "content": "<p>This competition will be very hard if you do not have extra GPUs.</p>",
      "rawMarkdown": "This competition will be very hard if you do not have extra GPUs.",
      "votes": null
    },
    {
      "id": "3228562",
      "postDate": "06/20/2025 09:39:44",
      "content": "<p>I Got it! 😁</p>",
      "rawMarkdown": "I Got it! 😁",
      "votes": null
    },
    {
      "id": "3228563",
      "postDate": "06/20/2025 09:41:00",
      "content": "<p>Congratulations!<br>\nI wasn't expecting that.</p>",
      "rawMarkdown": "Congratulations!\nI wasn't expecting that.",
      "votes": null
    },
    {
      "id": "3228567",
      "postDate": "06/20/2025 09:43:17",
      "content": "<p>Hoping to see another jump to sub 5</p>",
      "rawMarkdown": "Hoping to see another jump to sub 5",
      "votes": null
    },
    {
      "id": "3228570",
      "postDate": "06/20/2025 09:47:32",
      "content": "<blockquote>\n  <p>Hoping to see another jump to sub 5   </p>\n</blockquote>\n<p>I second that</p>",
      "rawMarkdown": ">Hoping to see another jump to sub 5   \n\nI second that",
      "votes": null
    },
    {
      "id": "3228573",
      "postDate": "06/20/2025 09:53:50",
      "content": "<p>Thank you, Don't even know how to start on a sub 5, haha</p>\n<p>But let's see if I can find another breakthrough, however unlikely.</p>",
      "rawMarkdown": "Thank you, Don't even know how to start on a sub 5, haha\n\nBut let's see if I can find another breakthrough, however unlikely.",
      "votes": null
    },
    {
      "id": "3228583",
      "postDate": "06/20/2025 10:18:38",
      "content": "<blockquote>\n  <p>Thank you, Don't even know how to start on a sub 5, haha<br>\n  But let's see if I can find another breakthrough, however unlikely.  </p>\n</blockquote>\n<p>Just train longer 🤣</p>",
      "rawMarkdown": ">Thank you, Don't even know how to start on a sub 5, haha\nBut let's see if I can find another breakthrough, however unlikely.  \n\nJust train longer 🤣",
      "votes": null
    },
    {
      "id": "3228589",
      "postDate": "06/20/2025 10:31:29",
      "content": "<p>please, chill…. : )</p>\n<p>I guess your method should be included in some waveform inversion paper already. But also 0 MAE solution is something we all would like to see</p>",
      "rawMarkdown": "please, chill.... : )\n\nI guess your method should be included in some waveform inversion paper already. But also 0 MAE solution is something we all would like to see",
      "votes": null
    },
    {
      "id": "3228593",
      "postDate": "06/20/2025 10:37:54",
      "content": "<p>start from open source model. there is one good 28.8 model (or even with the initial estimate from public csv)<br>\nyou can just do the enhancement (e.g. new samples or phyisics loss, etc)<br>\nthan will speedup your work.</p>",
      "rawMarkdown": "start from open source model. there is one good 28.8 model (or even with the initial estimate from public csv)\nyou can just do the enhancement (e.g. new samples or phyisics loss, etc)\nthan will speedup your work.",
      "votes": null
    },
    {
      "id": "3228614",
      "postDate": "06/20/2025 11:09:05",
      "content": "<p>Great job! Some teams will give up because of your score, while some teams will stay up until the end of the competition to try to figure out how you did it.</p>",
      "rawMarkdown": "Great job! Some teams will give up because of your score, while some teams will stay up until the end of the competition to try to figure out how you did it.",
      "votes": null
    },
    {
      "id": "3228632",
      "postDate": "06/20/2025 11:29:58",
      "content": "<p>the secret here is patience 🤣</p>",
      "rawMarkdown": "the secret here is patience 🤣",
      "votes": null
    },
    {
      "id": "3228903",
      "postDate": "06/20/2025 16:43:48",
      "content": "<p><a href=\"https://www.kaggle.com/harshitsheoran\" target=\"_blank\">@harshitsheoran</a> Congratulations Harshit. Nice improvement on LB!</p>",
      "rawMarkdown": "harshitsheoran Congratulations Harshit. Nice improvement on LB!",
      "votes": null
    },
    {
      "id": "3228905",
      "postDate": "06/20/2025 16:44:19",
      "content": "<p>Great job <a href=\"https://www.kaggle.com/shlomoron\" target=\"_blank\">@shlomoron</a> Nice improvement on LB!</p>",
      "rawMarkdown": "Great job @shlomoron Nice improvement on LB!",
      "votes": null
    },
    {
      "id": "3228907",
      "postDate": "06/20/2025 16:45:19",
      "content": "<p>It looks like there are some feature engineering tricks in addition to choosing/tuning image model.</p>",
      "rawMarkdown": "It looks like there are some feature engineering tricks in addition to choosing/tuning image model.",
      "votes": null
    },
    {
      "id": "3230507",
      "postDate": "06/23/2025 06:10:45",
      "content": "<p><a href=\"https://www.kaggle.com/harshitsheoran\" target=\"_blank\">@harshitsheoran</a> great score, congrats harshit 💥</p>",
      "rawMarkdown": "harshitsheoran great score, congrats harshit 💥",
      "votes": null
    },
    {
      "id": "3230621",
      "postDate": "06/23/2025 09:00:26",
      "content": "<p>Congrats to greySnow!</p>",
      "rawMarkdown": "Congrats to greySnow!",
      "votes": null
    },
    {
      "id": "3230812",
      "postDate": "06/23/2025 13:05:52",
      "content": "<p>Here i can't cross the 100 val_loss in my models. 😅.<br>\nRemarkable results.⭐</p>",
      "rawMarkdown": "Here i can't cross the 100 val_loss in my models. 😅.\nRemarkable results.⭐",
      "votes": null
    },
    {
      "id": "3230847",
      "postDate": "06/23/2025 14:24:32",
      "content": "<p>great job. congratu..</p>",
      "rawMarkdown": "great job. congratu..",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3227661,
      "author_name": "hydantess",
      "author_url": "",
      "post_date": "06/19/2025 06:54:57",
      "content": "<p>Good job, greySnow! </p>",
      "votes": null,
      "replies": [
        {
          "id": 3227666,
          "author_name": "shlomoron",
          "author_url": "",
          "post_date": "06/19/2025 06:58:44",
          "content": "<p>Thx. Knowing you, you will probably appear from thin air in gold range on last day :)</p>",
          "votes": null,
          "replies": [
            {
              "id": 3227677,
              "author_name": "hydantess",
              "author_url": "",
              "post_date": "06/19/2025 07:13:15",
              "content": "<p>I am also too lazy to submit, but i am still stuggling with gold line.</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 3227662,
      "author_name": "harshitsheoran",
      "author_url": "",
      "post_date": "06/19/2025 06:56:41",
      "content": "<p>Yup, amazing performance by greySnow! like a week ago he was ~550th, with lots of white noise between him and gold, he has cleared it all :D</p>\n<p>Training ends tonight! Sub incoming in next 24 hours :D, I hope my LB follows my CV 🤞</p>",
      "votes": null,
      "replies": [
        {
          "id": 3227671,
          "author_name": "shlomoron",
          "author_url": "",
          "post_date": "06/19/2025 07:00:14",
          "content": "<p>Nah, I was just too lazy to submit lol<br>\nYou going to jump to sub-10? :D</p>",
          "votes": null,
          "replies": [
            {
              "id": 3227675,
              "author_name": "harshitsheoran",
              "author_url": "",
              "post_date": "06/19/2025 07:08:42",
              "content": "<p>Honestly, even if my LB follows my CV like it did for 17.1 (CV 16.1), it's going to be a close call for the sub-10/single digit score and it's probably not going to be quite sub-10.</p>\n<p>My training is going to take even more time, Sub <em>probably not</em> incoming tommorow</p>",
              "votes": null,
              "replies": []
            },
            {
              "id": 3228562,
              "author_name": "harshitsheoran",
              "author_url": "",
              "post_date": "06/20/2025 09:39:44",
              "content": "<p>I Got it! 😁</p>",
              "votes": null,
              "replies": [
                {
                  "id": 3228614,
                  "author_name": "hydantess",
                  "author_url": "",
                  "post_date": "06/20/2025 11:09:05",
                  "content": "<p>Great job! Some teams will give up because of your score, while some teams will stay up until the end of the competition to try to figure out how you did it.</p>",
                  "votes": null,
                  "replies": [
                    {
                      "id": 3228632,
                      "author_name": "andy2709",
                      "author_url": "",
                      "post_date": "06/20/2025 11:29:58",
                      "content": "<p>the secret here is patience 🤣</p>",
                      "votes": null,
                      "replies": []
                    }
                  ]
                }
              ]
            },
            {
              "id": 3228563,
              "author_name": "cnumber",
              "author_url": "",
              "post_date": "06/20/2025 09:41:00",
              "content": "<p>Congratulations!<br>\nI wasn't expecting that.</p>",
              "votes": null,
              "replies": []
            },
            {
              "id": 3228567,
              "author_name": "cnumber",
              "author_url": "",
              "post_date": "06/20/2025 09:43:17",
              "content": "<p>Hoping to see another jump to sub 5</p>",
              "votes": null,
              "replies": [
                {
                  "id": 3228570,
                  "author_name": "shlomoron",
                  "author_url": "",
                  "post_date": "06/20/2025 09:47:32",
                  "content": "<blockquote>\n  <p>Hoping to see another jump to sub 5   </p>\n</blockquote>\n<p>I second that</p>",
                  "votes": null,
                  "replies": [
                    {
                      "id": 3228573,
                      "author_name": "harshitsheoran",
                      "author_url": "",
                      "post_date": "06/20/2025 09:53:50",
                      "content": "<p>Thank you, Don't even know how to start on a sub 5, haha</p>\n<p>But let's see if I can find another breakthrough, however unlikely.</p>",
                      "votes": null,
                      "replies": [
                        {
                          "id": 3228583,
                          "author_name": "shlomoron",
                          "author_url": "",
                          "post_date": "06/20/2025 10:18:38",
                          "content": "<blockquote>\n  <p>Thank you, Don't even know how to start on a sub 5, haha<br>\n  But let's see if I can find another breakthrough, however unlikely.  </p>\n</blockquote>\n<p>Just train longer 🤣</p>",
                          "votes": null,
                          "replies": []
                        },
                        {
                          "id": 3228589,
                          "author_name": "samson8",
                          "author_url": "",
                          "post_date": "06/20/2025 10:31:29",
                          "content": "<p>please, chill…. : )</p>\n<p>I guess your method should be included in some waveform inversion paper already. But also 0 MAE solution is something we all would like to see</p>",
                          "votes": null,
                          "replies": []
                        },
                        {
                          "id": 3228903,
                          "author_name": "cdeotte",
                          "author_url": "",
                          "post_date": "06/20/2025 16:43:48",
                          "content": "<p><a href=\"https://www.kaggle.com/harshitsheoran\" target=\"_blank\">@harshitsheoran</a> Congratulations Harshit. Nice improvement on LB!</p>",
                          "votes": null,
                          "replies": []
                        },
                        {
                          "id": 3230507,
                          "author_name": "nischaydnk",
                          "author_url": "",
                          "post_date": "06/23/2025 06:10:45",
                          "content": "<p><a href=\"https://www.kaggle.com/harshitsheoran\" target=\"_blank\">@harshitsheoran</a> great score, congrats harshit 💥</p>",
                          "votes": null,
                          "replies": []
                        }
                      ]
                    }
                  ]
                }
              ]
            }
          ]
        },
        {
          "id": 3227674,
          "author_name": "cnumber",
          "author_url": "",
          "post_date": "06/19/2025 07:07:32",
          "content": "<p>We all knew greySnow wasn't here to just post usefull discussions.</p>\n<p>Looking so forward to your next sub!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 3227956,
      "author_name": "theexaltedone",
      "author_url": "",
      "post_date": "06/19/2025 13:31:02",
      "content": "<p>I can't help but wonder — what kind of machines are you all running? Surely not P100s, right? 😅 Feels like there's some secret hardware magic going on here!</p>",
      "votes": null,
      "replies": [
        {
          "id": 3228533,
          "author_name": "jijingdeyi",
          "author_url": "",
          "post_date": "06/20/2025 08:59:35",
          "content": "<p>This competition will be very hard if you do not have extra GPUs.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 3228183,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "06/19/2025 20:23:56",
      "content": "<p><a href=\"https://www.kaggle.com/tomirol\" target=\"_blank\">@tomirol</a> getting 17.x at first sub is also remarkable.</p>",
      "votes": null,
      "replies": [
        {
          "id": 3228185,
          "author_name": "tomirol",
          "author_url": "",
          "post_date": "06/19/2025 20:39:40",
          "content": "<p>Hey thank you <a href=\"https://www.kaggle.com/cpmpml\" target=\"_blank\">@cpmpml</a>! I was only able to do that because so many people shared a lot of high quality stuff here (including you), and validation results as well. Honestly makes me feel like I need to share more in other comps too!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 3228235,
      "author_name": "cody11null",
      "author_url": "",
      "post_date": "06/20/2025 00:50:28",
      "content": "<p>I would aspire to do the same but training is so much hahaha </p>",
      "votes": null,
      "replies": [
        {
          "id": 3228593,
          "author_name": "hengck23",
          "author_url": "",
          "post_date": "06/20/2025 10:37:54",
          "content": "<p>start from open source model. there is one good 28.8 model (or even with the initial estimate from public csv)<br>\nyou can just do the enhancement (e.g. new samples or phyisics loss, etc)<br>\nthan will speedup your work.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 3228531,
      "author_name": "harshgupta4444",
      "author_url": "",
      "post_date": "06/20/2025 08:58:48",
      "content": "<p>Huge congrats to greySnow for that massive leap — 25.1 to 16.1 is no joke! 🔥 Well deserved gold medal. And yep, curious to see if Harshit is quietly cooking something big for that sub-10 breakthrough 😄</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3228905,
      "author_name": "cdeotte",
      "author_url": "",
      "post_date": "06/20/2025 16:44:19",
      "content": "<p>Great job <a href=\"https://www.kaggle.com/shlomoron\" target=\"_blank\">@shlomoron</a> Nice improvement on LB!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3228907,
      "author_name": "cdeotte",
      "author_url": "",
      "post_date": "06/20/2025 16:45:19",
      "content": "<p>It looks like there are some feature engineering tricks in addition to choosing/tuning image model.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3230621,
      "author_name": "wellkilo",
      "author_url": "",
      "post_date": "06/23/2025 09:00:26",
      "content": "<p>Congrats to greySnow!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3230812,
      "author_name": "arpit1bansal",
      "author_url": "",
      "post_date": "06/23/2025 13:05:52",
      "content": "<p>Here i can't cross the 100 val_loss in my models. 😅.<br>\nRemarkable results.⭐</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 3230847,
      "author_name": "plkyes",
      "author_url": "",
      "post_date": "06/23/2025 14:24:32",
      "content": "<p>great job. congratu..</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "3227650": "Huge applause to greySnow!\nAnd Harshit Sheoran, you haven't submitted for nearly 3 weaks, are you planning  to jump to sub 10? :)\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F6624777%2F295e191241ff0a1bf8040267e4ed0b89%2F2025-06-19%2015.47.29.png?generation=1750315661185941&alt=media)",
    "3227661": "Good job, greySnow!",
    "3227662": "Yup, amazing performance by greySnow! like a week ago he was ~550th, with lots of white noise between him and gold, he has cleared it all :D\n\nTraining ends tonight! Sub incoming in next 24 hours :D, I hope my LB follows my CV 🤞",
    "3227666": "Thx. Knowing you, you will probably appear from thin air in gold range on last day :)",
    "3227671": "Nah, I was just too lazy to submit lol\nYou going to jump to sub-10? :D",
    "3227674": "We all knew greySnow wasn't here to just post usefull discussions.\n\nLooking so forward to your next sub!",
    "3227675": "Honestly, even if my LB follows my CV like it did for 17.1 (CV 16.1), it's going to be a close call for the sub-10/single digit score and it's probably not going to be quite sub-10.\n\nMy training is going to take even more time, Sub *probably not* incoming tommorow",
    "3227677": "I am also too lazy to submit, but i am still stuggling with gold line.",
    "3227956": "I can't help but wonder — what kind of machines are you all running? Surely not P100s, right? 😅 Feels like there's some secret hardware magic going on here!",
    "3228183": "tomirol getting 17.x at first sub is also remarkable.",
    "3228185": "Hey thank you @cpmpml! I was only able to do that because so many people shared a lot of high quality stuff here (including you), and validation results as well. Honestly makes me feel like I need to share more in other comps too!",
    "3228235": "I would aspire to do the same but training is so much hahaha",
    "3228531": "Huge congrats to greySnow for that massive leap — 25.1 to 16.1 is no joke! 🔥 Well deserved gold medal. And yep, curious to see if Harshit is quietly cooking something big for that sub-10 breakthrough 😄",
    "3228533": "This competition will be very hard if you do not have extra GPUs.",
    "3228562": "I Got it! 😁",
    "3228563": "Congratulations!\nI wasn't expecting that.",
    "3228567": "Hoping to see another jump to sub 5",
    "3228570": ">Hoping to see another jump to sub 5   \n\nI second that",
    "3228573": "Thank you, Don't even know how to start on a sub 5, haha\n\nBut let's see if I can find another breakthrough, however unlikely.",
    "3228583": ">Thank you, Don't even know how to start on a sub 5, haha\nBut let's see if I can find another breakthrough, however unlikely.  \n\nJust train longer 🤣",
    "3228589": "please, chill.... : )\n\nI guess your method should be included in some waveform inversion paper already. But also 0 MAE solution is something we all would like to see",
    "3228593": "start from open source model. there is one good 28.8 model (or even with the initial estimate from public csv)\nyou can just do the enhancement (e.g. new samples or phyisics loss, etc)\nthan will speedup your work.",
    "3228614": "Great job! Some teams will give up because of your score, while some teams will stay up until the end of the competition to try to figure out how you did it.",
    "3228632": "the secret here is patience 🤣",
    "3228903": "harshitsheoran Congratulations Harshit. Nice improvement on LB!",
    "3228905": "Great job @shlomoron Nice improvement on LB!",
    "3228907": "It looks like there are some feature engineering tricks in addition to choosing/tuning image model.",
    "3230507": "harshitsheoran great score, congrats harshit 💥",
    "3230621": "Congrats to greySnow!",
    "3230812": "Here i can't cross the 100 val_loss in my models. 😅.\nRemarkable results.⭐",
    "3230847": "great job. congratu.."
  },
  "source": "meta"
}