Tip by Kirderf:
\n\"How it works: He saved the codes for every solution in an own python file which he run separately in isolated memory. He also take advantages of the new Kaggle env. feature that checks if we are in the hidden test running phase or only in the submitting phase, and use this by saving a submission sample in submitting phase and when in the phase of running the hidden private test phase, using the real code. This use only seconds of the GPU quota.\"
\nhttps://www.kaggle.com/competitions/mayo-clinic-strip-ai/discussion/349200
\nAnd Kirderf's code: Mayo inference, memory and GPU quota efficient
\nhttps://www.kaggle.com/code/kirderf/mayo-inference-memory-and-gpu-quota-efficient
I'm still trying to get those phases mentioned above (hidden test running phase/submitting phase/hidden private test phase/real code?) 😁
\nIt's too much for a beginner like me. Though I'm pretty sure that it'll help you .
\n\"All experiment results are only made possible by the Z8 by HP workstation with 2x A6000 48GB Nvidia GPU cards. With fast and large GPU cards, you can enjoy designing models and sleep early without worrying about out-of-memory errors :) \"
\nhttps://www.kaggle.com/competitions/rsna-breast-cancer-detection/discussion/370333
", "messages": [ { "id": 2473139, "postDate": "2023-10-08T01:40:21.927Z", "content": "Tip by Kirderf:
\n\"How it works: He saved the codes for every solution in an own python file which he run separately in isolated memory. He also take advantages of the new Kaggle env. feature that checks if we are in the hidden test running phase or only in the submitting phase, and use this by saving a submission sample in submitting phase and when in the phase of running the hidden private test phase, using the real code. This use only seconds of the GPU quota.\"
\nhttps://www.kaggle.com/competitions/mayo-clinic-strip-ai/discussion/349200
\nAnd Kirderf's code: Mayo inference, memory and GPU quota efficient
\nhttps://www.kaggle.com/code/kirderf/mayo-inference-memory-and-gpu-quota-efficient
I'm still trying to get those phases mentioned above (hidden test running phase/submitting phase/hidden private test phase/real code?) 😁
\nIt's too much for a beginner like me. Though I'm pretty sure that it'll help you .
\n\"All experiment results are only made possible by the Z8 by HP workstation with 2x A6000 48GB Nvidia GPU cards. With fast and large GPU cards, you can enjoy designing models and sleep early without worrying about out-of-memory errors :) \"
\nhttps://www.kaggle.com/competitions/rsna-breast-cancer-detection/discussion/370333
", "rawMarkdown": "#Save Memory. And GPU quota\n\nTip by Kirderf:\n\n\"How it works: He saved the codes for every solution in an own python file which he run separately in isolated memory. He also take advantages of the new Kaggle env. feature that checks if we are in the hidden test running phase or only in the submitting phase, and use this by saving a submission sample in submitting phase and when in the phase of running the hidden private test phase, using the real code. This use only seconds of the GPU quota.\"\n\nhttps://www.kaggle.com/competitions/mayo-clinic-strip-ai/discussion/349200\n\nAnd Kirderf's code: Mayo inference, memory and GPU quota efficient\nhttps://www.kaggle.com/code/kirderf/mayo-inference-memory-and-gpu-quota-efficient\n\nI'm still trying to get those phases mentioned above (hidden test running phase/submitting phase/hidden private test phase/real code?) 😁\n\nIt's too much for a beginner like me. Though I'm pretty sure that it'll help you .\n\n# Hardware to train large models. By HengCK23\n\n\"All experiment results are only made possible by the Z8 by HP workstation with 2x A6000 48GB Nvidia GPU cards. With fast and large GPU cards, you can enjoy designing models and sleep early without worrying about out-of-memory errors :) \"\n\nhttps://www.kaggle.com/competitions/rsna-breast-cancer-detection/discussion/370333", "votes": 22 }, { "id": 2473741, "postDate": "2023-10-08T14:36:58.657Z", "content": "Beginner 😂😂, very funny !!!
", "rawMarkdown": "Beginner 😂😂, very funny !!!", "votes": 1, "replies": [ { "id": 2473788, "postDate": "2023-10-08T15:24:26.357Z", "content": "Indeed, Yusuph
\nI'm still a beginner which is beyond lame. 😊
\nThough, everyday I start another lesson on Kaggle trying to learn new stuff.
", "rawMarkdown": "Indeed, Yusuph\n\nI'm still a beginner which is beyond lame. 😊\n\nThough, everyday I start another lesson on Kaggle trying to learn new stuff.", "votes": 3, "replies": [ { "id": 2474164, "postDate": "2023-10-09T03:26:32.890Z", "content": "Thank you Marília for continuously sharing what you are learning with other beginners like myself❤️
", "rawMarkdown": "Thank you Marília for continuously sharing what you are learning with other beginners like myself❤️", "votes": 1 }, { "id": 2474778, "postDate": "2023-10-09T13:09:23.533Z", "content": "Since I'm Not able to write like any experienced Kaggler at least I \"Recycle\" some topics published by those amazing, generous professionals.
", "rawMarkdown": "Since I'm Not able to write like any experienced Kaggler at least I \"Recycle\" some topics published by those amazing, generous professionals.", "votes": 1 }, { "id": 2475070, "postDate": "2023-10-09T16:12:41.680Z", "content": "Thank you Marília, love these insights that you share 😀
", "rawMarkdown": "Thank you Marília, love these insights that you share 😀", "votes": 2 }, { "id": 2475081, "postDate": "2023-10-09T16:21:58.423Z", "content": "Hi Chih Chung Wo,
\nAs I mentioned above, I try to \"recycle\" the information that experienced Kagglers shared on the most Similar competition to UBC-OCEAN which is (from my not very trustful point-of-view):
\nMayo Clinic - STRIP AI (1 year ago)- We learned a lot with large images.
\nhttps://www.kaggle.com/competitions/mayo-clinic-strip-ai
", "rawMarkdown": "Hi Chih Chung Wo,\n\nAs I mentioned above, I try to \"recycle\" the information that experienced Kagglers shared on the most Similar competition to UBC-OCEAN which is (from my not very trustful point-of-view):\n\nMayo Clinic - STRIP AI (1 year ago)- We learned a lot with large images.\n\nhttps://www.kaggle.com/competitions/mayo-clinic-strip-ai", "votes": 2 }, { "id": 2475133, "postDate": "2023-10-09T16:45:11.480Z", "content": "In the age of information overloading, sharing is an act of kindness giving chances to those who are lost (me). Thank you for recycling the info. I have zero education/experience in medical imagery (or even anything related to health care system) but looks like medical research is accelerated by deep learning significantly, so I took my first try on Kaggle in the RSNA's ATD competition and it is quite fascinating to me so I think I am going to focus on medical imagery competitions for awhile. This ovarian cancer detection contest is something I am interested (not yet entered though) as I am trying to close of my current competitions hoping to learn more from it before I move to the next (this one). Then I noticed your recycled comment. Yes like you, I am not sure I understand exactly what the experts are trying to share, but it does look like something useful so I am pretty sure I will spending something thinking about these wisdom from the experts.
", "rawMarkdown": "In the age of information overloading, sharing is an act of kindness giving chances to those who are lost (me). Thank you for recycling the info. I have zero education/experience in medical imagery (or even anything related to health care system) but looks like medical research is accelerated by deep learning significantly, so I took my first try on Kaggle in the RSNA's ATD competition and it is quite fascinating to me so I think I am going to focus on medical imagery competitions for awhile. This ovarian cancer detection contest is something I am interested (not yet entered though) as I am trying to close of my current competitions hoping to learn more from it before I move to the next (this one). Then I noticed your recycled comment. Yes like you, I am not sure I understand exactly what the experts are trying to share, but it does look like something useful so I am pretty sure I will spending something thinking about these wisdom from the experts.", "votes": 1 }, { "id": 2475178, "postDate": "2023-10-09T17:24:42.187Z", "content": "Additionally, many contributors/experts are underrated. In general, users don't read their Notebooks after the middle of the competition when the Notebook's list is already long.
\nMy tip is: read all codes that you can daily. Give a check (vote or whatever to show yourself that you've read it). Maybe you could make some comment on a code that you liked or even copy that code. I prefer to copy instead of just forking it cause we can understand how long takes each step.
\nThen you save it since competitions have similarities from time to time. Therefore, you'll be able to deliver something on the next competition.
\nThat's why I prefer to copy kagglers that aren't on the top. Contributors code tend to have less views. That's a way to show appreciation to their work too.
", "rawMarkdown": "Additionally, many contributors/experts are underrated. In general, users don't read their Notebooks after the middle of the competition when the Notebook's list is already long.\n\nMy tip is: read all codes that you can daily. Give a check (vote or whatever to show yourself that you've read it). Maybe you could make some comment on a code that you liked or even copy that code. I prefer to copy instead of just forking it cause we can understand how long takes each step. \n\nThen you save it since competitions have similarities from time to time. Therefore, you'll be able to deliver something on the next competition.\n\nThat's why I prefer to copy kagglers that aren't on the top. Contributors code tend to have less views. That's a way to show appreciation to their work too.", "votes": 2 }, { "id": 2476387, "postDate": "2023-10-10T15:16:56.253Z", "content": "Thank you Marília! You are very resourceful and helpful 😀
", "rawMarkdown": "Thank you Marília! You are very resourceful and helpful 😀", "votes": 1 }, { "id": 2476691, "postDate": "2023-10-10T20:58:58.697Z", "content": "I'm very glad to hear that.
", "rawMarkdown": "I'm very glad to hear that." } ] } ] } ], "comments": [ { "id": 2473741, "author_name": "Yusuph Mustapha Ladi", "author_url": "", "post_date": "2023-10-08T14:36:58.657000", "content": "Beginner 😂😂, very funny !!!
", "votes": 1, "replies": [ { "id": 2473788, "author_name": "Marília Prata", "author_url": "", "post_date": "2023-10-08T15:24:26.357000", "content": "Indeed, Yusuph
\nI'm still a beginner which is beyond lame. 😊
\nThough, everyday I start another lesson on Kaggle trying to learn new stuff.
", "votes": 3, "replies": [ { "id": 2474164, "author_name": "Yusuph Mustapha Ladi", "author_url": "", "post_date": "2023-10-09T03:26:32.890000", "content": "Thank you Marília for continuously sharing what you are learning with other beginners like myself❤️
", "votes": 1, "replies": [] }, { "id": 2474778, "author_name": "Marília Prata", "author_url": "", "post_date": "2023-10-09T13:09:23.533000", "content": "Since I'm Not able to write like any experienced Kaggler at least I \"Recycle\" some topics published by those amazing, generous professionals.
", "votes": 1, "replies": [] }, { "id": 2475070, "author_name": "Kirk Wuo / Chih-Chung Wuo", "author_url": "", "post_date": "2023-10-09T16:12:41.680000", "content": "Thank you Marília, love these insights that you share 😀
", "votes": 2, "replies": [] }, { "id": 2475081, "author_name": "Marília Prata", "author_url": "", "post_date": "2023-10-09T16:21:58.423000", "content": "Hi Chih Chung Wo,
\nAs I mentioned above, I try to \"recycle\" the information that experienced Kagglers shared on the most Similar competition to UBC-OCEAN which is (from my not very trustful point-of-view):
\nMayo Clinic - STRIP AI (1 year ago)- We learned a lot with large images.
\nhttps://www.kaggle.com/competitions/mayo-clinic-strip-ai
", "votes": 2, "replies": [] }, { "id": 2475133, "author_name": "Kirk Wuo / Chih-Chung Wuo", "author_url": "", "post_date": "2023-10-09T16:45:11.480000", "content": "In the age of information overloading, sharing is an act of kindness giving chances to those who are lost (me). Thank you for recycling the info. I have zero education/experience in medical imagery (or even anything related to health care system) but looks like medical research is accelerated by deep learning significantly, so I took my first try on Kaggle in the RSNA's ATD competition and it is quite fascinating to me so I think I am going to focus on medical imagery competitions for awhile. This ovarian cancer detection contest is something I am interested (not yet entered though) as I am trying to close of my current competitions hoping to learn more from it before I move to the next (this one). Then I noticed your recycled comment. Yes like you, I am not sure I understand exactly what the experts are trying to share, but it does look like something useful so I am pretty sure I will spending something thinking about these wisdom from the experts.
", "votes": 1, "replies": [] }, { "id": 2475178, "author_name": "Marília Prata", "author_url": "", "post_date": "2023-10-09T17:24:42.187000", "content": "Additionally, many contributors/experts are underrated. In general, users don't read their Notebooks after the middle of the competition when the Notebook's list is already long.
\nMy tip is: read all codes that you can daily. Give a check (vote or whatever to show yourself that you've read it). Maybe you could make some comment on a code that you liked or even copy that code. I prefer to copy instead of just forking it cause we can understand how long takes each step.
\nThen you save it since competitions have similarities from time to time. Therefore, you'll be able to deliver something on the next competition.
\nThat's why I prefer to copy kagglers that aren't on the top. Contributors code tend to have less views. That's a way to show appreciation to their work too.
", "votes": 2, "replies": [] }, { "id": 2476387, "author_name": "Kirk Wuo / Chih-Chung Wuo", "author_url": "", "post_date": "2023-10-10T15:16:56.253000", "content": "Thank you Marília! You are very resourceful and helpful 😀
", "votes": 1, "replies": [] }, { "id": 2476691, "author_name": "Marília Prata", "author_url": "", "post_date": "2023-10-10T20:58:58.697000", "content": "I'm very glad to hear that.
", "votes": 0, "replies": [] } ] } ] } ], "raw_markdown_by_id": { "2473139": "#Save Memory. And GPU quota\n\nTip by Kirderf:\n\n\"How it works: He saved the codes for every solution in an own python file which he run separately in isolated memory. He also take advantages of the new Kaggle env. feature that checks if we are in the hidden test running phase or only in the submitting phase, and use this by saving a submission sample in submitting phase and when in the phase of running the hidden private test phase, using the real code. This use only seconds of the GPU quota.\"\n\nhttps://www.kaggle.com/competitions/mayo-clinic-strip-ai/discussion/349200\n\nAnd Kirderf's code: Mayo inference, memory and GPU quota efficient\nhttps://www.kaggle.com/code/kirderf/mayo-inference-memory-and-gpu-quota-efficient\n\nI'm still trying to get those phases mentioned above (hidden test running phase/submitting phase/hidden private test phase/real code?) 😁\n\nIt's too much for a beginner like me. Though I'm pretty sure that it'll help you .\n\n# Hardware to train large models. By HengCK23\n\n\"All experiment results are only made possible by the Z8 by HP workstation with 2x A6000 48GB Nvidia GPU cards. With fast and large GPU cards, you can enjoy designing models and sleep early without worrying about out-of-memory errors :) \"\n\nhttps://www.kaggle.com/competitions/rsna-breast-cancer-detection/discussion/370333", "2473741": "Beginner 😂😂, very funny !!!" } }