I changed a few parameters and it seems that 0.652 is the highest score.
", "messages": [ { "id": 2822054, "postDate": "2024-05-18T12:18:54.370Z", "content": "I changed a few parameters and it seems that 0.652 is the highest score.
", "rawMarkdown": "I changed a few parameters and it seems that 0.652 is the highest score.", "votes": 5 }, { "id": 2822073, "postDate": "2024-05-18T12:34:48.040Z", "content": "From baroque feature engineering on a multi-tabular dataset to random parameter searches- how has this competition degraded!!!
\nI feel sorry for the hosts @a520hh
You don’t think it was self inflicted by them? If they are allowing metric hack of week num then they should have not hidden it in test set, otherwise of course people got creative to discover these “hacks”, time better spent vs modelling for the LB scores
", "rawMarkdown": "You don’t think it was self inflicted by them? If they are allowing metric hack of week num then they should have not hidden it in test set, otherwise of course people got creative to discover these “hacks”, time better spent vs modelling for the LB scores ", "votes": 4, "replies": [ { "id": 2822082, "postDate": "2024-05-18T12:40:34.510Z", "content": "True my friend @julianmukaj
\nI think all of us will remember this competition for a long time for varied reasons!
Do you think there will be a shake up for this late hack? @ravi20076
", "rawMarkdown": "Do you think there will be a shake up for this late hack? @ravi20076 ", "votes": 1 }, { "id": 2822136, "postDate": "2024-05-18T13:10:21.370Z", "content": "Frankly nothing scientific has worked here for us at least. I don't know about this, but I am somehow enjoying the late drama here @octaviograu
", "rawMarkdown": "Frankly nothing scientific has worked here for us at least. I don't know about this, but I am somehow enjoying the late drama here @octaviograu ", "votes": 3 } ] }, { "id": 2822083, "postDate": "2024-05-18T12:40:55.273Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2822123, "postDate": "2024-05-18T13:02:19.437Z", "content": "indeed @ravi20076
", "rawMarkdown": "indeed @ravi20076 ", "votes": 1 }, { "id": 2822443, "postDate": "2024-05-18T15:40:48.047Z", "content": "@ravi20076 I think the most valuable thing the host will get from this competition is the multiple suggestions of other metrics to assess stability. And a clear reason why the current metric is (probably) not what they want.
", "rawMarkdown": "@ravi20076 I think the most valuable thing the host will get from this competition is the multiple suggestions of other metrics to assess stability. And a clear reason why the current metric is (probably) not what they want. ", "votes": 1 } ] }, { "id": 2822147, "postDate": "2024-05-18T13:15:15.363Z", "content": "The team name \"All You Need Is Eval Metric Hack\" is spot on.
", "rawMarkdown": "The team name \"All You Need Is Eval Metric Hack\" is spot on." }, { "id": 2822135, "postDate": "2024-05-18T13:09:39.117Z", "content": "Now you are 0.653, just by changing the parameters?
", "rawMarkdown": "Now you are 0.653, just by changing the parameters?" } ], "comments": [ { "id": 2822073, "author_name": "Ravi Ramakrishnan", "author_url": "", "post_date": "2024-05-18T12:34:48.040000", "content": "From baroque feature engineering on a multi-tabular dataset to random parameter searches- how has this competition degraded!!!
\nI feel sorry for the hosts @a520hh
You don’t think it was self inflicted by them? If they are allowing metric hack of week num then they should have not hidden it in test set, otherwise of course people got creative to discover these “hacks”, time better spent vs modelling for the LB scores
", "votes": 4, "replies": [ { "id": 2822082, "author_name": "Ravi Ramakrishnan", "author_url": "", "post_date": "2024-05-18T12:40:34.510000", "content": "True my friend @julianmukaj
\nI think all of us will remember this competition for a long time for varied reasons!
Do you think there will be a shake up for this late hack? @ravi20076
", "votes": 1, "replies": [] }, { "id": 2822136, "author_name": "Ravi Ramakrishnan", "author_url": "", "post_date": "2024-05-18T13:10:21.370000", "content": "Frankly nothing scientific has worked here for us at least. I don't know about this, but I am somehow enjoying the late drama here @octaviograu
", "votes": 3, "replies": [] } ] }, { "id": 2822083, "author_name": "", "author_url": "", "post_date": "2024-05-18T12:40:55.273000", "content": "", "votes": 0, "replies": [] }, { "id": 2822123, "author_name": "Octavi Grau", "author_url": "", "post_date": "2024-05-18T13:02:19.437000", "content": "indeed @ravi20076
", "votes": 1, "replies": [] }, { "id": 2822443, "author_name": "Ern711", "author_url": "", "post_date": "2024-05-18T15:40:48.047000", "content": "@ravi20076 I think the most valuable thing the host will get from this competition is the multiple suggestions of other metrics to assess stability. And a clear reason why the current metric is (probably) not what they want.
", "votes": 1, "replies": [] } ] }, { "id": 2822147, "author_name": "Bruce", "author_url": "", "post_date": "2024-05-18T13:15:15.363000", "content": "The team name \"All You Need Is Eval Metric Hack\" is spot on.
", "votes": 0, "replies": [] }, { "id": 2822135, "author_name": "", "author_url": "", "post_date": "2024-05-18T13:09:39.117000", "content": "Now you are 0.653, just by changing the parameters?
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2822054": "I changed a few parameters and it seems that 0.652 is the highest score.", "2822073": "From baroque feature engineering on a multi-tabular dataset to random parameter searches- how has this competition degraded!!!\nI feel sorry for the hosts @a520hh ", "2822147": "The team name \"All You Need Is Eval Metric Hack\" is spot on.", "2822135": "Now you are 0.653, just by changing the parameters?" } }