Hi, I have a Late submission private score of 0.66281. Does it have any significance now as it seems to be a second best score as per the leaderboard.
\nUpdate: with bit of a tuning, I got the score 0.65705 which is better than the 1st place on leaderboard.
\nThanks!
", "messages": [ { "id": 1984928, "postDate": "2022-10-13T01:25:53.317Z", "content": "Here is the complete pipeline:
\nData preprocessing
\nModel Training
\nSubmission
Please vote if you find it useful. Thanks!
", "rawMarkdown": "Here is the complete pipeline:\n[Data preprocessing](https://www.kaggle.com/vijaybj/0-65705-data-preprocessing)\n[Model Training](https://www.kaggle.com/code/vijaybj/0-65705-model-training)\n[Submission](https://www.kaggle.com/vijaybj/0-65705-submission)\n\nPlease vote if you find it useful. Thanks!", "votes": 3, "replies": [ { "id": 1985894, "postDate": "2022-10-13T16:51:18.153Z", "content": "@vijaybj congrats on your results, you seem to have a few very good scoring submissions :)
\nCould you explain, as short as you like, the logic behind concatenating successive Conv2D layer outputs? Was is to avoid rapid dimension reduction in your model?
", "rawMarkdown": "@vijaybj congrats on your results, you seem to have a few very good scoring submissions :)\n\nCould you explain, as short as you like, the logic behind concatenating successive Conv2D layer outputs? Was is to avoid rapid dimension reduction in your model? ", "votes": 1 }, { "id": 1986295, "postDate": "2022-10-14T02:02:40.683Z", "content": "@td-iceman thank you for those kind words. Actually those successive Conv2D layers are with different kernel size. One is with kernel size 1 and other with kernel size 3. But ultimately it's a SqueezeNet architecture.
\nSqueezeNet Architecture
@vijaybj thanks for the link, this is interesting stuff.
", "rawMarkdown": "@vijaybj thanks for the link, this is interesting stuff." } ] }, { "id": 1979948, "postDate": "2022-10-09T20:44:01.737Z", "content": "Hi, I have a Late submission private score of 0.66281. Does it have any significance now as it seems to be a second best score as per the leaderboard.
\nUpdate: with bit of a tuning, I got the score 0.65705 which is better than the 1st place on leaderboard.
\nThanks!
", "rawMarkdown": "Hi, I have a Late submission private score of 0.66281. Does it have any significance now as it seems to be a second best score as per the leaderboard.\n\nUpdate: with bit of a tuning, I got the score 0.65705 which is better than the 1st place on leaderboard.\n\nThanks!", "votes": 4 }, { "id": 1983779, "postDate": "2022-10-12T08:22:08.800Z", "content": "Since the winners are already announced, I don't think that the leaderboard would be affected with your late submission.
\nCongratulations on the score….your solution could be a great source of knowledge to all.
I have shared it because I thought my efforts will not go waste if it can be useful to kagglers in some way. Thanks!
", "rawMarkdown": "I have shared it because I thought my efforts will not go waste if it can be useful to kagglers in some way. Thanks!", "votes": 1 } ] }, { "id": 1979979, "postDate": "2022-10-09T22:13:31.480Z", "content": "@vijaybj your efforts definitely have significance, especially given you score close to the best solution. I would be very interested to know your solution in detail, and I'm sure the hosts wouldn't mind learning extra valuable information for sure. It's probably the LB that might not be affected, but we know and will appreciate your work nonetheless :)
", "rawMarkdown": "@vijaybj your efforts definitely have significance, especially given you score close to the best solution. I would be very interested to know your solution in detail, and I'm sure the hosts wouldn't mind learning extra valuable information for sure. It's probably the LB that might not be affected, but we know and will appreciate your work nonetheless :)", "votes": 1, "replies": [ { "id": 1980173, "postDate": "2022-10-10T02:42:50.763Z", "content": "Thanks for the reply… I am going to share the complete pipeline in couple of days.
", "rawMarkdown": "Thanks for the reply... I am going to share the complete pipeline in couple of days.", "votes": 1 } ] }, { "id": 1979981, "postDate": "2022-10-09T22:17:36.850Z", "content": "I agree with the other comment. This kind of submission can bring great insights both to other kagglers and the organizers, I would suggest you to publish it (if you can), I think many participants would love to see a solution code that would have lead to a second place.
", "rawMarkdown": "I agree with the other comment. This kind of submission can bring great insights both to other kagglers and the organizers, I would suggest you to publish it (if you can), I think many participants would love to see a solution code that would have lead to a second place.", "votes": 2, "replies": [ { "id": 1980175, "postDate": "2022-10-10T02:44:09.227Z", "content": "True, thanks for the reply. I will be sharing it soon.
", "rawMarkdown": "True, thanks for the reply. I will be sharing it soon." } ] }, { "id": 2365385, "postDate": "2023-07-30T07:52:58.250Z", "content": "Hey Kishore, Wonderful you atleast made a submission. Great job!
", "rawMarkdown": "Hey Kishore, Wonderful you atleast made a submission. Great job!" } ], "comments": [ { "id": 1984928, "author_name": "Vijay Jadhav", "author_url": "", "post_date": "2022-10-13T01:25:53.317000", "content": "Here is the complete pipeline:
\nData preprocessing
\nModel Training
\nSubmission
Please vote if you find it useful. Thanks!
", "votes": 3, "replies": [ { "id": 1985894, "author_name": "tdiceman", "author_url": "", "post_date": "2022-10-13T16:51:18.153000", "content": "@vijaybj congrats on your results, you seem to have a few very good scoring submissions :)
\nCould you explain, as short as you like, the logic behind concatenating successive Conv2D layer outputs? Was is to avoid rapid dimension reduction in your model?
", "votes": 1, "replies": [] }, { "id": 1986295, "author_name": "Vijay Jadhav", "author_url": "", "post_date": "2022-10-14T02:02:40.683000", "content": "@td-iceman thank you for those kind words. Actually those successive Conv2D layers are with different kernel size. One is with kernel size 1 and other with kernel size 3. But ultimately it's a SqueezeNet architecture.
\nSqueezeNet Architecture
@vijaybj thanks for the link, this is interesting stuff.
", "votes": 0, "replies": [] } ] }, { "id": 1983779, "author_name": "YugalKishore", "author_url": "", "post_date": "2022-10-12T08:22:08.800000", "content": "Since the winners are already announced, I don't think that the leaderboard would be affected with your late submission.
\nCongratulations on the score….your solution could be a great source of knowledge to all.
I have shared it because I thought my efforts will not go waste if it can be useful to kagglers in some way. Thanks!
", "votes": 1, "replies": [] } ] }, { "id": 1979979, "author_name": "tdiceman", "author_url": "", "post_date": "2022-10-09T22:13:31.480000", "content": "@vijaybj your efforts definitely have significance, especially given you score close to the best solution. I would be very interested to know your solution in detail, and I'm sure the hosts wouldn't mind learning extra valuable information for sure. It's probably the LB that might not be affected, but we know and will appreciate your work nonetheless :)
", "votes": 1, "replies": [ { "id": 1980173, "author_name": "Vijay Jadhav", "author_url": "", "post_date": "2022-10-10T02:42:50.763000", "content": "Thanks for the reply… I am going to share the complete pipeline in couple of days.
", "votes": 1, "replies": [] } ] }, { "id": 1979981, "author_name": "Pietro Maldini", "author_url": "", "post_date": "2022-10-09T22:17:36.850000", "content": "I agree with the other comment. This kind of submission can bring great insights both to other kagglers and the organizers, I would suggest you to publish it (if you can), I think many participants would love to see a solution code that would have lead to a second place.
", "votes": 2, "replies": [ { "id": 1980175, "author_name": "Vijay Jadhav", "author_url": "", "post_date": "2022-10-10T02:44:09.227000", "content": "True, thanks for the reply. I will be sharing it soon.
", "votes": 0, "replies": [] } ] }, { "id": 2365385, "author_name": "Maria Kalva", "author_url": "", "post_date": "2023-07-30T07:52:58.250000", "content": "Hey Kishore, Wonderful you atleast made a submission. Great job!
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1984928": "Here is the complete pipeline:\n[Data preprocessing](https://www.kaggle.com/vijaybj/0-65705-data-preprocessing)\n[Model Training](https://www.kaggle.com/code/vijaybj/0-65705-model-training)\n[Submission](https://www.kaggle.com/vijaybj/0-65705-submission)\n\nPlease vote if you find it useful. Thanks!", "1979948": "Hi, I have a Late submission private score of 0.66281. Does it have any significance now as it seems to be a second best score as per the leaderboard.\n\nUpdate: with bit of a tuning, I got the score 0.65705 which is better than the 1st place on leaderboard.\n\nThanks!", "1983779": "Since the winners are already announced, I don't think that the leaderboard would be affected with your late submission.\nCongratulations on the score....your solution could be a great source of knowledge to all.", "1979979": "@vijaybj your efforts definitely have significance, especially given you score close to the best solution. I would be very interested to know your solution in detail, and I'm sure the hosts wouldn't mind learning extra valuable information for sure. It's probably the LB that might not be affected, but we know and will appreciate your work nonetheless :)", "1979981": "I agree with the other comment. This kind of submission can bring great insights both to other kagglers and the organizers, I would suggest you to publish it (if you can), I think many participants would love to see a solution code that would have lead to a second place.", "2365385": "Hey Kishore, Wonderful you atleast made a submission. Great job!" } }