![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2832287%2F8eab5b121d5594afd86b34cd32e3a08f%2FScreenshot%20from%202023-08-01%2007-15-12.png?generation=1690849211810797&alt=media\")
Is it possible to exceed 0.5? My LB is my best attempt so far😂
", "messages": [ { "id": 2364643, "postDate": "2023-07-29T13:42:53.687Z", "content": "best cv model : lb 0.510
\nbest lb model : lb 0.556
I think there might be a shake-up because of CV-LB not correlated
", "rawMarkdown": "best cv model : lb 0.510\nbest lb model : lb 0.556\n\nI think there might be a shake-up because of CV-LB not correlated", "votes": 3, "replies": [ { "id": 2364653, "postDate": "2023-07-29T13:49:33.217Z", "content": "what is the best lb model if training all data directly? Thanks !
", "rawMarkdown": "what is the best lb model if training all data directly? Thanks !", "replies": [ { "id": 2364669, "postDate": "2023-07-29T14:06:06.217Z", "content": "I use all data in my final models
", "rawMarkdown": "I use all data in my final models", "votes": 1 } ] }, { "id": 2366612, "postDate": "2023-07-31T05:41:09.773Z", "content": "I think there might be a shake-up because of CV-LB not correlated I agree
best lb single model: 0.510 without dialate
\nbest lb ensemble model: 0.538 without dialate
\nbest lb ensemble model: 0.551 with dialate
What about the contribution from blackpink? 😂😂
", "rawMarkdown": "What about the contribution from blackpink? 😂😂", "replies": [ { "id": 2367939, "postDate": "2023-08-01T00:19:36.307Z", "content": ":(( dataset 1 is all I need. Single model with dataset 1, Private 0.543, PB 0.51. My best private is 0.597
", "rawMarkdown": ":(( dataset 1 is all I need. Single model with dataset 1, Private 0.543, PB 0.51. My best private is 0.597\n", "votes": 1 }, { "id": 2367941, "postDate": "2023-08-01T00:20:13.297Z", "content": "
Private 0.597 ![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F2832287%2F18480c0e058ce89e85d3091cedcae340%2FScreenshot%20from%202023-08-01%2007-21-36.png?generation=1690849323767129&alt=media\")
0.48 with Internimage XL on 1024 scale and 0.5 with 5 folds
", "rawMarkdown": "0.48 with Internimage XL on 1024 scale and 0.5 with 5 folds ", "votes": 4, "replies": [ { "id": 2356124, "postDate": "2023-07-23T23:27:18.300Z", "content": "Great idea! I haven't followed and used any of the SOTA so far😂
", "rawMarkdown": "Great idea! I haven't followed and used any of the SOTA so far😂" }, { "id": 2364655, "postDate": "2023-07-29T13:51:58.597Z", "content": "SOTA model do not show higher lb results, maybe data process is import?
", "rawMarkdown": "SOTA model do not show higher lb results, maybe data process is import? " } ] }, { "id": 2356384, "postDate": "2023-07-24T06:38:37.163Z", "content": "mmdet3.1 0.499,1024 scale, singlefold
", "rawMarkdown": "mmdet3.1 0.499,1024 scale, singlefold", "votes": 1 }, { "id": 2355074, "postDate": "2023-07-23T04:35:15.890Z", "content": "Is it possible to exceed 0.5? My LB is my best attempt so far😂
", "rawMarkdown": "Is it possible to exceed 0.5? My LB is my best attempt so far😂", "votes": 2 }, { "id": 2363929, "postDate": "2023-07-29T03:33:18.593Z", "content": "What is a good starter notebook to use and what modifications are needed to achieve 0.5LB?
\nThank you.
", "rawMarkdown": "What is a good starter notebook to use and what modifications are needed to achieve 0.5LB?\n\nThank you." }, { "id": 2356371, "postDate": "2023-07-24T06:12:07.073Z", "content": "I achieved LB 0.505 with single fold mmdet model.
", "rawMarkdown": "I achieved LB 0.505 with single fold mmdet model." }, { "id": 2356155, "postDate": "2023-07-24T00:45:05.410Z", "content": "mmdet3.0? --> 0.483
", "rawMarkdown": "mmdet3.0? --> 0.483" }, { "id": 2356018, "postDate": "2023-07-23T19:32:24.573Z", "content": "My best model so far achieved .515 LB.
", "rawMarkdown": "My best model so far achieved .515 LB.", "replies": [ { "id": 2356125, "postDate": "2023-07-23T23:28:28.097Z", "content": "Nice work, bro. I'll keep trying!
", "rawMarkdown": "Nice work, bro. I'll keep trying!" } ] }, { "id": 2355694, "postDate": "2023-07-23T14:17:43.387Z", "content": "I guess no one will share a notebook with such a high score this close to the deadline. I believe that some in the top have such models. We will know quite soon.
", "rawMarkdown": "I guess no one will share a notebook with such a high score this close to the deadline. I believe that some in the top have such models. We will know quite soon.", "replies": [ { "id": 2355875, "postDate": "2023-07-23T17:33:33.173Z", "content": "Well, I don't think a notebook is necessary for this topic, at first I thought I was just bringing up a topic that has been very common in previous competitions…😂 BTW, the Discussion for this competition doesn't seem to be as \"lively\" as it used to be, a little bit weird😂lol
", "rawMarkdown": "Well, I don't think a notebook is necessary for this topic, at first I thought I was just bringing up a topic that has been very common in previous competitions...😂 BTW, the Discussion for this competition doesn't seem to be as \"lively\" as it used to be, a little bit weird😂lol" } ] }, { "id": 2361418, "postDate": "2023-07-27T11:52:03.290Z", "content": "mmdet-maskrcnn 0.485 with single fold and 0.5 with 5 folds
", "rawMarkdown": "mmdet-maskrcnn 0.485 with single fold and 0.5 with 5 folds", "isDeleted": true } ], "comments": [ { "id": 2364643, "author_name": "atom1231", "author_url": "", "post_date": "2023-07-29T13:42:53.687000", "content": "best cv model : lb 0.510
\nbest lb model : lb 0.556
I think there might be a shake-up because of CV-LB not correlated
", "votes": 3, "replies": [ { "id": 2364653, "author_name": "clwclw", "author_url": "", "post_date": "2023-07-29T13:49:33.217000", "content": "what is the best lb model if training all data directly? Thanks !
", "votes": 0, "replies": [ { "id": 2364669, "author_name": "atom1231", "author_url": "", "post_date": "2023-07-29T14:06:06.217000", "content": "I use all data in my final models
", "votes": 1, "replies": [] } ] }, { "id": 2366612, "author_name": "Sven", "author_url": "", "post_date": "2023-07-31T05:41:09.773000", "content": "I think there might be a shake-up because of CV-LB not correlated I agree
best lb single model: 0.510 without dialate
\nbest lb ensemble model: 0.538 without dialate
\nbest lb ensemble model: 0.551 with dialate
What about the contribution from blackpink? 😂😂
", "votes": 0, "replies": [ { "id": 2367939, "author_name": "Quan Vu", "author_url": "", "post_date": "2023-08-01T00:19:36.307000", "content": ":(( dataset 1 is all I need. Single model with dataset 1, Private 0.543, PB 0.51. My best private is 0.597
", "votes": 1, "replies": [] }, { "id": 2367941, "author_name": "Quan Vu", "author_url": "", "post_date": "2023-08-01T00:20:13.297000", "content": "
Private 0.597
0.48 with Internimage XL on 1024 scale and 0.5 with 5 folds
", "votes": 4, "replies": [ { "id": 2356124, "author_name": "阳光开朗大男孩", "author_url": "", "post_date": "2023-07-23T23:27:18.300000", "content": "Great idea! I haven't followed and used any of the SOTA so far😂
", "votes": 0, "replies": [] }, { "id": 2364655, "author_name": "clwclw", "author_url": "", "post_date": "2023-07-29T13:51:58.597000", "content": "SOTA model do not show higher lb results, maybe data process is import?
", "votes": 0, "replies": [] } ] }, { "id": 2356384, "author_name": "Ksana", "author_url": "", "post_date": "2023-07-24T06:38:37.163000", "content": "mmdet3.1 0.499,1024 scale, singlefold
", "votes": 1, "replies": [] }, { "id": 2363929, "author_name": "Bo Peng", "author_url": "", "post_date": "2023-07-29T03:33:18.593000", "content": "What is a good starter notebook to use and what modifications are needed to achieve 0.5LB?
\nThank you.
", "votes": 0, "replies": [] }, { "id": 2356371, "author_name": "Araik Tamazian", "author_url": "", "post_date": "2023-07-24T06:12:07.073000", "content": "I achieved LB 0.505 with single fold mmdet model.
", "votes": 0, "replies": [] }, { "id": 2356155, "author_name": "JCW2023", "author_url": "", "post_date": "2023-07-24T00:45:05.410000", "content": "mmdet3.0? --> 0.483
", "votes": 0, "replies": [] }, { "id": 2356018, "author_name": "Abdelfattah Toulaoui", "author_url": "", "post_date": "2023-07-23T19:32:24.573000", "content": "My best model so far achieved .515 LB.
", "votes": 0, "replies": [ { "id": 2356125, "author_name": "阳光开朗大男孩", "author_url": "", "post_date": "2023-07-23T23:28:28.097000", "content": "Nice work, bro. I'll keep trying!
", "votes": 0, "replies": [] } ] }, { "id": 2355694, "author_name": "Yassine Alouini", "author_url": "", "post_date": "2023-07-23T14:17:43.387000", "content": "I guess no one will share a notebook with such a high score this close to the deadline. I believe that some in the top have such models. We will know quite soon.
", "votes": 0, "replies": [ { "id": 2355875, "author_name": "阳光开朗大男孩", "author_url": "", "post_date": "2023-07-23T17:33:33.173000", "content": "Well, I don't think a notebook is necessary for this topic, at first I thought I was just bringing up a topic that has been very common in previous competitions…😂 BTW, the Discussion for this competition doesn't seem to be as \"lively\" as it used to be, a little bit weird😂lol
", "votes": 0, "replies": [] } ] }, { "id": 2361418, "author_name": "", "author_url": "", "post_date": "2023-07-27T11:52:03.290000", "content": "mmdet-maskrcnn 0.485 with single fold and 0.5 with 5 folds
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2364643": "best cv model : lb 0.510\nbest lb model : lb 0.556\n\nI think there might be a shake-up because of CV-LB not correlated", "2367469": "best lb single model: 0.510 without dialate\nbest lb ensemble model: 0.538 without dialate\nbest lb ensemble model: 0.551 with dialate", "2356109": "0.48 with Internimage XL on 1024 scale and 0.5 with 5 folds ", "2356384": "mmdet3.1 0.499,1024 scale, singlefold", "2355074": "Is it possible to exceed 0.5? My LB is my best attempt so far😂", "2363929": "What is a good starter notebook to use and what modifications are needed to achieve 0.5LB?\n\nThank you.", "2356371": "I achieved LB 0.505 with single fold mmdet model.", "2356155": "mmdet3.0? --> 0.483", "2356018": "My best model so far achieved .515 LB.", "2355694": "I guess no one will share a notebook with such a high score this close to the deadline. I believe that some in the top have such models. We will know quite soon.", "2361418": "mmdet-maskrcnn 0.485 with single fold and 0.5 with 5 folds" } }