{
  "id": 118214,
  "title": "Please post your LB score without leak",
  "url": "/competitions/ashrae-energy-prediction/discussion/118214",
  "author_name": "Tim Yee",
  "post_date": "2019-11-20T05:24:56.033000",
  "votes": 15,
  "comment_count": 41,
  "views": 0,
  "content": "<p>Since there is currently no visibility of where you stand on LB without data leak. I am sharing my current LB score without any leaks. I hope others will share as well so that we can re-create a limited pre-leak LB. Also, if anyone is training on leaked data - how do you know it will affect private LB?</p>\n\n<p>Current LB score without leak - 1.07</p>\n\n<p>Update: 1.04 (closer to 1.05)</p>\n\n<p>Update - 12/13: 1.043 (estimated) </p>\n\n<p>Update - 12/16: 1.040 (estimated) </p>\n\n<p>Update - 12/18: 1.038 (estimated) </p>\n\n<p>Update - 12/19: 1.032 (estimated)</p>",
  "messages": [
    {
      "id": 677395,
      "postDate": "2019-11-20T05:24:56.033Z",
      "content": "<p>Since there is currently no visibility of where you stand on LB without data leak. I am sharing my current LB score without any leaks. I hope others will share as well so that we can re-create a limited pre-leak LB. Also, if anyone is training on leaked data - how do you know it will affect private LB?</p>\n\n<p>Current LB score without leak - 1.07</p>\n\n<p>Update: 1.04 (closer to 1.05)</p>\n\n<p>Update - 12/13: 1.043 (estimated) </p>\n\n<p>Update - 12/16: 1.040 (estimated) </p>\n\n<p>Update - 12/18: 1.038 (estimated) </p>\n\n<p>Update - 12/19: 1.032 (estimated)</p>",
      "rawMarkdown": "Since there is currently no visibility of where you stand on LB without data leak. I am sharing my current LB score without any leaks. I hope others will share as well so that we can re-create a limited pre-leak LB. Also, if anyone is training on leaked data - how do you know it will affect private LB?\n\nCurrent LB score without leak - 1.07\n\nUpdate: 1.04 (closer to 1.05)\n\nUpdate - 12/13: 1.043 (estimated) \n\nUpdate - 12/16: 1.040 (estimated) \n\nUpdate - 12/18: 1.038 (estimated) \n\nUpdate - 12/19: 1.032 (estimated)",
      "votes": 15
    },
    {
      "id": 696878,
      "postDate": "2019-12-17T07:20:44.943Z",
      "content": "<p>1.048 single model, no leaks</p>",
      "rawMarkdown": "1.048 single model, no leaks",
      "votes": 3
    },
    {
      "id": 695987,
      "postDate": "2019-12-16T00:59:23.980Z",
      "content": "<p>Single model(no leak) cv = 1.068, lb = 1.057</p>",
      "rawMarkdown": "Single model(no leak) cv = 1.068, lb = 1.057",
      "votes": 3,
      "replies": [
        {
          "id": 695995,
          "postDate": "2019-12-16T01:27:21.290Z",
          "content": "<p><a href=\"/harshit92\">@harshit92</a> what's the 3rd decimal place on the LB?</p>",
          "rawMarkdown": "@harshit92 what's the 3rd decimal place on the LB?"
        },
        {
          "id": 696006,
          "postDate": "2019-12-16T01:57:12.173Z",
          "content": "<p>Ohh Sorry, updated to third decimal. Expecting .01-.02 improvement though in new version. </p>",
          "rawMarkdown": "Ohh Sorry, updated to third decimal. Expecting .01-.02 improvement though in new version. "
        },
        {
          "id": 696512,
          "postDate": "2019-12-16T17:42:47.090Z",
          "content": "<p>Very impressive!</p>",
          "rawMarkdown": "Very impressive!"
        }
      ]
    },
    {
      "id": 691781,
      "postDate": "2019-12-10T13:49:13.863Z",
      "content": "<p>1.05, single model</p>",
      "rawMarkdown": "1.05, single model",
      "votes": 4
    },
    {
      "id": 685092,
      "postDate": "2019-12-01T00:33:07.663Z",
      "content": "<p>1.05 -- with lots of internal blending but no public kernels or leak data.</p>",
      "rawMarkdown": "1.05 -- with lots of internal blending but no public kernels or leak data.",
      "votes": 4,
      "replies": [
        {
          "id": 685219,
          "postDate": "2019-12-01T07:30:44.343Z",
          "content": "<p><a href=\"/purist1024\">@purist1024</a> and merging all available leaks onto your blended 1.05 0,1,2,4,15 give you 0.1 boost on LB?</p>",
          "rawMarkdown": "@purist1024 and merging all available leaks onto your blended 1.05 0,1,2,4,15 give you 0.1 boost on LB?"
        },
        {
          "id": 685263,
          "postDate": "2019-12-01T09:49:46.543Z",
          "content": "<p>Correct, except that for my 0.95 submission I actually tried dropping site 4, so it's just 0,1,2,15. However, I got comparable results on an earlier kernel where I did blend in 4 as well. It does seem that the leaks are worth right around 0.1 in general.</p>",
          "rawMarkdown": "Correct, except that for my 0.95 submission I actually tried dropping site 4, so it's just 0,1,2,15. However, I got comparable results on an earlier kernel where I did blend in 4 as well. It does seem that the leaks are worth right around 0.1 in general."
        },
        {
          "id": 695406,
          "postDate": "2019-12-15T04:57:29.133Z",
          "content": "<p>Updated 12/14: 1.048 blended; 1.054 single strategy.</p>",
          "rawMarkdown": "Updated 12/14: 1.048 blended; 1.054 single strategy.\n"
        }
      ]
    },
    {
      "id": 677522,
      "postDate": "2019-11-20T09:19:08.993Z",
      "content": "<p>LB 1.05 (but I think very close to 1.06 rather than 1.04)</p>",
      "rawMarkdown": "LB 1.05 (but I think very close to 1.06 rather than 1.04)",
      "votes": 4
    },
    {
      "id": 692147,
      "postDate": "2019-12-11T00:35:32.763Z",
      "content": "<p>Current LB score without leak - 1.08</p>",
      "rawMarkdown": "Current LB score without leak - 1.08",
      "votes": 1,
      "replies": [
        {
          "id": 694449,
          "postDate": "2019-12-13T16:19:56.550Z",
          "content": "<p>1.07 now</p>",
          "rawMarkdown": "1.07 now",
          "votes": 1
        },
        {
          "id": 695373,
          "postDate": "2019-12-15T03:03:48.097Z",
          "content": "<p><a href=\"/roydatascience\">@roydatascience</a> your LB score is 0.953\nhow to get so much improvement?\nonly replace prediction by leakage data?</p>",
          "rawMarkdown": "@roydatascience your LB score is 0.953\nhow to get so much improvement?\nonly replace prediction by leakage data?"
        },
        {
          "id": 695893,
          "postDate": "2019-12-15T19:17:25.200Z",
          "content": "<p>I am not using any leak data. Just using multiple models and using weights to blend them</p>",
          "rawMarkdown": "I am not using any leak data. Just using multiple models and using weights to blend them",
          "votes": 2
        }
      ]
    },
    {
      "id": 678034,
      "postDate": "2019-11-20T23:10:13.703Z",
      "content": "<p>1.07</p>",
      "rawMarkdown": "1.07",
      "votes": 1,
      "replies": [
        {
          "id": 687804,
          "postDate": "2019-12-04T21:09:18.340Z",
          "content": "<p>update: 1.05 now</p>",
          "rawMarkdown": "update: 1.05 now"
        }
      ]
    },
    {
      "id": 677860,
      "postDate": "2019-11-20T17:14:31.547Z",
      "content": "<p>1.07</p>",
      "rawMarkdown": "1.07",
      "votes": 1
    },
    {
      "id": 677853,
      "postDate": "2019-11-20T17:01:27.207Z",
      "content": "<p>1.08 nearly close to 1.07</p>",
      "rawMarkdown": "1.08 nearly close to 1.07",
      "votes": 1,
      "replies": [
        {
          "id": 686218,
          "postDate": "2019-12-02T23:16:35.303Z",
          "content": "<p>update 1.06</p>",
          "rawMarkdown": "update 1.06",
          "votes": 1
        }
      ]
    },
    {
      "id": 677837,
      "postDate": "2019-11-20T16:34:33.620Z",
      "content": "<p>1.06</p>",
      "rawMarkdown": "1.06",
      "votes": 1
    },
    {
      "id": 677531,
      "postDate": "2019-11-20T09:36:59.453Z",
      "content": "<p>1.056</p>",
      "rawMarkdown": "1.056",
      "votes": 1
    },
    {
      "id": 677525,
      "postDate": "2019-11-20T09:28:16.753Z",
      "content": "<p>LB 1.08，but very close 1.09</p>",
      "rawMarkdown": "LB 1.08，but very close 1.09",
      "votes": 1
    },
    {
      "id": 677524,
      "postDate": "2019-11-20T09:23:23.770Z",
      "content": "<p>1.07  =_=</p>",
      "rawMarkdown": "1.07  =_=",
      "votes": 1
    },
    {
      "id": 677517,
      "postDate": "2019-11-20T09:09:03.483Z",
      "content": "<p>1.08 </p>",
      "rawMarkdown": "1.08 ",
      "votes": 1
    },
    {
      "id": 677511,
      "postDate": "2019-11-20T09:00:37.003Z",
      "content": "<p>1.06</p>",
      "rawMarkdown": "1.06",
      "votes": 1
    },
    {
      "id": 677502,
      "postDate": "2019-11-20T08:46:31.447Z",
      "content": "<p>1.08 iirc</p>",
      "rawMarkdown": "1.08 iirc",
      "votes": 1
    },
    {
      "id": 677465,
      "postDate": "2019-11-20T07:49:39.703Z",
      "content": "<p>I managed 1.09 so far.</p>",
      "rawMarkdown": "I managed 1.09 so far.",
      "votes": 1
    },
    {
      "id": 691517,
      "postDate": "2019-12-10T08:13:35.137Z",
      "content": "<p>1.04，but i use leak data as train data.</p>",
      "rawMarkdown": "1.04，but i use leak data as train data.",
      "votes": 2
    },
    {
      "id": 677521,
      "postDate": "2019-11-20T09:13:55.377Z",
      "content": "<p>CV: 0.977       LB: 1.062</p>",
      "rawMarkdown": "CV: 0.977       LB: 1.062",
      "votes": 2,
      "replies": [
        {
          "id": 677892,
          "postDate": "2019-11-20T18:11:43.257Z",
          "content": "<p>Let us know how you manage to see the 3rd digit in the public LB 😉</p>",
          "rawMarkdown": "Let us know how you manage to see the 3rd digit in the public LB 😉",
          "votes": 7
        },
        {
          "id": 677915,
          "postDate": "2019-11-20T18:56:15.293Z",
          "content": "<p>Reserving a seat here 😄 </p>",
          "rawMarkdown": "Reserving a seat here 😄 "
        },
        {
          "id": 677916,
          "postDate": "2019-11-20T18:58:12.727Z",
          "content": "<p>It is a guesstimate: 1.06 + rank/bincount; with rank being my rank in the [1.06, 1.07] bin of the LB and bincount being the total number of submissions falling in that same bin. I am fortunate enough to not have submitted models involving leaked data; I would have lost this ability to 'see' the 3rd digit (I think).</p>",
          "rawMarkdown": "It is a guesstimate: 1.06 + rank/bincount; with rank being my rank in the [1.06, 1.07] bin of the LB and bincount being the total number of submissions falling in that same bin. I am fortunate enough to not have submitted models involving leaked data; I would have lost this ability to 'see' the 3rd digit (I think).",
          "votes": 3
        }
      ]
    },
    {
      "id": 696981,
      "postDate": "2019-12-17T10:15:56.280Z",
      "content": "<p>1.049 single model  without leak</p>",
      "rawMarkdown": "1.049 single model  without leak"
    },
    {
      "id": 695685,
      "postDate": "2019-12-15T13:09:13.763Z",
      "content": "<p>1.074 no leak</p>",
      "rawMarkdown": "1.074 no leak",
      "replies": [
        {
          "id": 696011,
          "postDate": "2019-12-16T02:04:29.980Z",
          "content": "<p>Update 1.065</p>",
          "rawMarkdown": "Update 1.065"
        }
      ]
    },
    {
      "id": 691577,
      "postDate": "2019-12-10T09:44:31.590Z",
      "content": "<p><a href=\"/teeyee314\">@teeyee314</a> </p>\n\n<blockquote>\n  <p>Update: 1.04 (closer to 1.05)</p>\n</blockquote>\n\n<p>Single model or blend?</p>",
      "rawMarkdown": "@teeyee314 \n&gt; Update: 1.04 (closer to 1.05)\n\nSingle model or blend?",
      "replies": [
        {
          "id": 692130,
          "postDate": "2019-12-10T23:38:07.827Z",
          "content": "<p>Blend. 1.06 single.</p>",
          "rawMarkdown": "Blend. 1.06 single."
        }
      ]
    },
    {
      "id": 687837,
      "postDate": "2019-12-04T22:46:21.880Z",
      "content": "<p>1.08 - one model. </p>",
      "rawMarkdown": "1.08 - one model. "
    },
    {
      "id": 678512,
      "postDate": "2019-11-21T13:34:59.450Z",
      "content": "<p>1.09</p>",
      "rawMarkdown": "1.09"
    },
    {
      "id": 691529,
      "postDate": "2019-12-10T08:31:29.543Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 696878,
      "author_name": "Gunes Evitan",
      "author_url": "",
      "post_date": "2019-12-17T07:20:44.943000",
      "content": "<p>1.048 single model, no leaks</p>",
      "votes": 3,
      "replies": []
    },
    {
      "id": 695987,
      "author_name": "HarshitMehta",
      "author_url": "",
      "post_date": "2019-12-16T00:59:23.980000",
      "content": "<p>Single model(no leak) cv = 1.068, lb = 1.057</p>",
      "votes": 3,
      "replies": [
        {
          "id": 695995,
          "author_name": "Tim Yee",
          "author_url": "",
          "post_date": "2019-12-16T01:27:21.290000",
          "content": "<p><a href=\"/harshit92\">@harshit92</a> what's the 3rd decimal place on the LB?</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 696006,
          "author_name": "HarshitMehta",
          "author_url": "",
          "post_date": "2019-12-16T01:57:12.173000",
          "content": "<p>Ohh Sorry, updated to third decimal. Expecting .01-.02 improvement though in new version. </p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 696512,
          "author_name": "Ahmet Erdem",
          "author_url": "",
          "post_date": "2019-12-16T17:42:47.090000",
          "content": "<p>Very impressive!</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 691781,
      "author_name": "TASSAN",
      "author_url": "",
      "post_date": "2019-12-10T13:49:13.863000",
      "content": "<p>1.05, single model</p>",
      "votes": 4,
      "replies": []
    },
    {
      "id": 685092,
      "author_name": "Robert Stockton",
      "author_url": "",
      "post_date": "2019-12-01T00:33:07.663000",
      "content": "<p>1.05 -- with lots of internal blending but no public kernels or leak data.</p>",
      "votes": 4,
      "replies": [
        {
          "id": 685219,
          "author_name": "Tim Yee",
          "author_url": "",
          "post_date": "2019-12-01T07:30:44.343000",
          "content": "<p><a href=\"/purist1024\">@purist1024</a> and merging all available leaks onto your blended 1.05 0,1,2,4,15 give you 0.1 boost on LB?</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 685263,
          "author_name": "Robert Stockton",
          "author_url": "",
          "post_date": "2019-12-01T09:49:46.543000",
          "content": "<p>Correct, except that for my 0.95 submission I actually tried dropping site 4, so it's just 0,1,2,15. However, I got comparable results on an earlier kernel where I did blend in 4 as well. It does seem that the leaks are worth right around 0.1 in general.</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 695406,
          "author_name": "Robert Stockton",
          "author_url": "",
          "post_date": "2019-12-15T04:57:29.133000",
          "content": "<p>Updated 12/14: 1.048 blended; 1.054 single strategy.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 677522,
      "author_name": "Hiroyuki Namba",
      "author_url": "",
      "post_date": "2019-11-20T09:19:08.993000",
      "content": "<p>LB 1.05 (but I think very close to 1.06 rather than 1.04)</p>",
      "votes": 4,
      "replies": []
    },
    {
      "id": 692147,
      "author_name": "Ashish Gupta",
      "author_url": "",
      "post_date": "2019-12-11T00:35:32.763000",
      "content": "<p>Current LB score without leak - 1.08</p>",
      "votes": 1,
      "replies": [
        {
          "id": 694449,
          "author_name": "Ashish Gupta",
          "author_url": "",
          "post_date": "2019-12-13T16:19:56.550000",
          "content": "<p>1.07 now</p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 695373,
          "author_name": "Dean",
          "author_url": "",
          "post_date": "2019-12-15T03:03:48.097000",
          "content": "<p><a href=\"/roydatascience\">@roydatascience</a> your LB score is 0.953\nhow to get so much improvement?\nonly replace prediction by leakage data?</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 695893,
          "author_name": "Ashish Gupta",
          "author_url": "",
          "post_date": "2019-12-15T19:17:25.200000",
          "content": "<p>I am not using any leak data. Just using multiple models and using weights to blend them</p>",
          "votes": 2,
          "replies": []
        }
      ]
    },
    {
      "id": 678034,
      "author_name": "Jie Wu",
      "author_url": "",
      "post_date": "2019-11-20T23:10:13.703000",
      "content": "<p>1.07</p>",
      "votes": 1,
      "replies": [
        {
          "id": 687804,
          "author_name": "Jie Wu",
          "author_url": "",
          "post_date": "2019-12-04T21:09:18.340000",
          "content": "<p>update: 1.05 now</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 677860,
      "author_name": "DietHard",
      "author_url": "",
      "post_date": "2019-11-20T17:14:31.547000",
      "content": "<p>1.07</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 677853,
      "author_name": "Hanjoon Choe",
      "author_url": "",
      "post_date": "2019-11-20T17:01:27.207000",
      "content": "<p>1.08 nearly close to 1.07</p>",
      "votes": 1,
      "replies": [
        {
          "id": 686218,
          "author_name": "Hanjoon Choe",
          "author_url": "",
          "post_date": "2019-12-02T23:16:35.303000",
          "content": "<p>update 1.06</p>",
          "votes": 1,
          "replies": []
        }
      ]
    },
    {
      "id": 677837,
      "author_name": "ADITYA KUMAR  (Negi)",
      "author_url": "",
      "post_date": "2019-11-20T16:34:33.620000",
      "content": "<p>1.06</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 677531,
      "author_name": "Davide Stenner",
      "author_url": "",
      "post_date": "2019-11-20T09:36:59.453000",
      "content": "<p>1.056</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 677525,
      "author_name": "FISH",
      "author_url": "",
      "post_date": "2019-11-20T09:28:16.753000",
      "content": "<p>LB 1.08，but very close 1.09</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 677524,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-11-20T09:23:23.770000",
      "content": "<p>1.07  =_=</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 677517,
      "author_name": "Chau Huynh",
      "author_url": "",
      "post_date": "2019-11-20T09:09:03.483000",
      "content": "<p>1.08 </p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 677511,
      "author_name": "Zhang Yunfei",
      "author_url": "",
      "post_date": "2019-11-20T09:00:37.003000",
      "content": "<p>1.06</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 677502,
      "author_name": "Gunes Evitan",
      "author_url": "",
      "post_date": "2019-11-20T08:46:31.447000",
      "content": "<p>1.08 iirc</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 677465,
      "author_name": "Konrad Banachewicz",
      "author_url": "",
      "post_date": "2019-11-20T07:49:39.703000",
      "content": "<p>I managed 1.09 so far.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 691517,
      "author_name": "xckkcxxck",
      "author_url": "",
      "post_date": "2019-12-10T08:13:35.137000",
      "content": "<p>1.04，but i use leak data as train data.</p>",
      "votes": 2,
      "replies": []
    },
    {
      "id": 677521,
      "author_name": "Philippe Lonjoux",
      "author_url": "",
      "post_date": "2019-11-20T09:13:55.377000",
      "content": "<p>CV: 0.977       LB: 1.062</p>",
      "votes": 2,
      "replies": [
        {
          "id": 677892,
          "author_name": "bluetrain",
          "author_url": "",
          "post_date": "2019-11-20T18:11:43.257000",
          "content": "<p>Let us know how you manage to see the 3rd digit in the public LB 😉</p>",
          "votes": 7,
          "replies": []
        },
        {
          "id": 677915,
          "author_name": "Mohammed Tayor",
          "author_url": "",
          "post_date": "2019-11-20T18:56:15.293000",
          "content": "<p>Reserving a seat here 😄 </p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 677916,
          "author_name": "Philippe Lonjoux",
          "author_url": "",
          "post_date": "2019-11-20T18:58:12.727000",
          "content": "<p>It is a guesstimate: 1.06 + rank/bincount; with rank being my rank in the [1.06, 1.07] bin of the LB and bincount being the total number of submissions falling in that same bin. I am fortunate enough to not have submitted models involving leaked data; I would have lost this ability to 'see' the 3rd digit (I think).</p>",
          "votes": 3,
          "replies": []
        }
      ]
    },
    {
      "id": 696981,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-12-17T10:15:56.280000",
      "content": "<p>1.049 single model  without leak</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 695685,
      "author_name": "makogarei",
      "author_url": "",
      "post_date": "2019-12-15T13:09:13.763000",
      "content": "<p>1.074 no leak</p>",
      "votes": 0,
      "replies": [
        {
          "id": 696011,
          "author_name": "makogarei",
          "author_url": "",
          "post_date": "2019-12-16T02:04:29.980000",
          "content": "<p>Update 1.065</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 691577,
      "author_name": "Gunes Evitan",
      "author_url": "",
      "post_date": "2019-12-10T09:44:31.590000",
      "content": "<p><a href=\"/teeyee314\">@teeyee314</a> </p>\n\n<blockquote>\n  <p>Update: 1.04 (closer to 1.05)</p>\n</blockquote>\n\n<p>Single model or blend?</p>",
      "votes": 0,
      "replies": [
        {
          "id": 692130,
          "author_name": "Tim Yee",
          "author_url": "",
          "post_date": "2019-12-10T23:38:07.827000",
          "content": "<p>Blend. 1.06 single.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 687837,
      "author_name": "Rafal Bodziony",
      "author_url": "",
      "post_date": "2019-12-04T22:46:21.880000",
      "content": "<p>1.08 - one model. </p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 678512,
      "author_name": "AdityaVikramSingh",
      "author_url": "",
      "post_date": "2019-11-21T13:34:59.450000",
      "content": "<p>1.09</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 691529,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-12-10T08:31:29.543000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "677395": "Since there is currently no visibility of where you stand on LB without data leak. I am sharing my current LB score without any leaks. I hope others will share as well so that we can re-create a limited pre-leak LB. Also, if anyone is training on leaked data - how do you know it will affect private LB?\n\nCurrent LB score without leak - 1.07\n\nUpdate: 1.04 (closer to 1.05)\n\nUpdate - 12/13: 1.043 (estimated) \n\nUpdate - 12/16: 1.040 (estimated) \n\nUpdate - 12/18: 1.038 (estimated) \n\nUpdate - 12/19: 1.032 (estimated)",
    "696878": "1.048 single model, no leaks",
    "695987": "Single model(no leak) cv = 1.068, lb = 1.057",
    "691781": "1.05, single model",
    "685092": "1.05 -- with lots of internal blending but no public kernels or leak data.",
    "677522": "LB 1.05 (but I think very close to 1.06 rather than 1.04)",
    "692147": "Current LB score without leak - 1.08",
    "678034": "1.07",
    "677860": "1.07",
    "677853": "1.08 nearly close to 1.07",
    "677837": "1.06",
    "677531": "1.056",
    "677525": "LB 1.08，but very close 1.09",
    "677524": "1.07  =_=",
    "677517": "1.08 ",
    "677511": "1.06",
    "677502": "1.08 iirc",
    "677465": "I managed 1.09 so far.",
    "691517": "1.04，but i use leak data as train data.",
    "677521": "CV: 0.977       LB: 1.062",
    "696981": "1.049 single model  without leak",
    "695685": "1.074 no leak",
    "691577": "@teeyee314 \n&gt; Update: 1.04 (closer to 1.05)\n\nSingle model or blend?",
    "687837": "1.08 - one model. ",
    "678512": "1.09",
    "691529": ""
  }
}