I started at this competition very, very late. Most of my submissions got the \"Notebook Exceeded Allowed Compute\" error. Last 10 or 12 days, I worked very hard on this. I got very frustrated because nothing was working. Whatever I was doing, I got this error. I got some good advice from @icemantd. From the discussion I learned, filtering out large files might be a good idea. In my last notebook, I tried filtering out files with more than 800 MB but still got that error(other people submitted it a with 1.2GB threshold). Then I send the same notebook to my friend, with a threshold of 700 MB to submit. It worked but got a low score on public LB. Today that notebook came out in 50 th place🔥. What a nice surprise!!
\nI should have submitted my ensemble notebook👀
50th place solution notebook: https://www.kaggle.com/code/bakar31/50th-place-solution
\nYou can check my GitHub repo for this competition: https://github.com/Bakar31/Image-Classification-of-Stroke-Blood-Clot-Origin
I started at this competition very, very late. Most of my submissions got the \"Notebook Exceeded Allowed Compute\" error. Last 10 or 12 days, I worked very hard on this. I got very frustrated because nothing was working. Whatever I was doing, I got this error. I got some good advice from @icemantd. From the discussion I learned, filtering out large files might be a good idea. In my last notebook, I tried filtering out files with more than 800 MB but still got that error(other people submitted it a with 1.2GB threshold). Then I send the same notebook to my friend, with a threshold of 700 MB to submit. It worked but got a low score on public LB. Today that notebook came out in 50 th place🔥. What a nice surprise!!
\nI should have submitted my ensemble notebook👀
50th place solution notebook: https://www.kaggle.com/code/bakar31/50th-place-solution
\nYou can check my GitHub repo for this competition: https://github.com/Bakar31/Image-Classification-of-Stroke-Blood-Clot-Origin
Great work @bakar31, I also agree with this, one may trust simple approaches, they usually do well for most cases. Good result!
", "rawMarkdown": "Great work @bakar31, I also agree with this, one may trust simple approaches, they usually do well for most cases. Good result! ", "votes": 1, "replies": [ { "id": 1974999, "postDate": "2022-10-06T15:13:34.440Z", "content": "Thank you❤️
", "rawMarkdown": "Thank you❤️" } ] }, { "id": 1975002, "postDate": "2022-10-06T15:15:45.293Z", "content": "Great going @bakar31, glad you didn't give up :)
", "rawMarkdown": "Great going @bakar31, glad you didn't give up :)", "replies": [ { "id": 1975063, "postDate": "2022-10-06T15:40:53.967Z", "content": "Hey would love to connect in LinkedIn
", "rawMarkdown": "Hey would love to connect in LinkedIn" }, { "id": 1975116, "postDate": "2022-10-06T15:57:53.027Z", "content": "Sent you a request :)
", "rawMarkdown": "Sent you a request :)" } ] } ], "comments": [ { "id": 1974959, "author_name": "Ravi Ramakrishnan", "author_url": "", "post_date": "2022-10-06T14:56:47.707000", "content": "Great work @bakar31, I also agree with this, one may trust simple approaches, they usually do well for most cases. Good result!
", "votes": 1, "replies": [ { "id": 1974999, "author_name": "Abu Bakar", "author_url": "", "post_date": "2022-10-06T15:13:34.440000", "content": "Thank you❤️
", "votes": 0, "replies": [] } ] }, { "id": 1975002, "author_name": "tdiceman", "author_url": "", "post_date": "2022-10-06T15:15:45.293000", "content": "Great going @bakar31, glad you didn't give up :)
", "votes": 0, "replies": [ { "id": 1975063, "author_name": "Abu Bakar", "author_url": "", "post_date": "2022-10-06T15:40:53.967000", "content": "Hey would love to connect in LinkedIn
", "votes": 0, "replies": [] }, { "id": 1975116, "author_name": "tdiceman", "author_url": "", "post_date": "2022-10-06T15:57:53.027000", "content": "Sent you a request :)
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "1974842": "I started at this competition very, very late. Most of my submissions got the \"Notebook Exceeded Allowed Compute\" error. Last 10 or 12 days, I worked very hard on this. I got very frustrated because nothing was working. Whatever I was doing, I got this error. I got some good advice from @icemantd. From the discussion I learned, filtering out large files might be a good idea. In my last notebook, I tried filtering out files with more than 800 MB but still got that error(other people submitted it a with 1.2GB threshold). Then I send the same notebook to my friend, with a threshold of 700 MB to submit. It worked but got a low score on public LB. Today that notebook came out in **50 th** place🔥. What a nice surprise!! \nI should have submitted my ensemble notebook👀\n\n50th place solution notebook: https://www.kaggle.com/code/bakar31/50th-place-solution\nYou can check my GitHub repo for this competition: https://github.com/Bakar31/Image-Classification-of-Stroke-Blood-Clot-Origin", "1974959": "Great work @bakar31, I also agree with this, one may trust simple approaches, they usually do well for most cases. Good result! ", "1975002": "Great going @bakar31, glad you didn't give up :)" } }