Hi everyone, and welcome to Grand X-Ray Slam: Division A, the first of a 2-part Kaggle hackathon series hosted by myself under my company Blue and Gold Healthcare Inc.
\nThis competition challenges you to build machine learning models that can detect 14 thoracic conditions from chest X-ray images. Your models will be tested on real-world clinical complexity, so accuracy and robustness are key.
\nWe’re excited to see your approaches and solutions.
\nGood luck!
— The Grand X-Ray Slam Team
", "messages": [ { "id": 3272575, "postDate": "2025-08-21T08:39:31.070Z", "content": "Hi everyone, and welcome to Grand X-Ray Slam: Division A, the first of a 2-part Kaggle hackathon series hosted by myself under my company Blue and Gold Healthcare Inc.
\nThis competition challenges you to build machine learning models that can detect 14 thoracic conditions from chest X-ray images. Your models will be tested on real-world clinical complexity, so accuracy and robustness are key.
\nWe’re excited to see your approaches and solutions.
\nGood luck!
— The Grand X-Ray Slam Team
", "rawMarkdown": "Hi everyone, and welcome to **Grand X-Ray Slam: Division A**, the first of a 2-part Kaggle hackathon series hosted by myself under my company Blue and Gold Healthcare Inc. \n\nThis competition challenges you to build machine learning models that can detect **14 thoracic conditions** from chest X-ray images. Your models will be tested on real-world clinical complexity, so accuracy and robustness are key. \n\n## Prizes \n- **1st Place:** $750 \n- **2nd Place:** $500 \n- **3rd Place:** $250 \n- **Grand Slam Bonus:** $2,500 (shared among top 3 performers across Division A & B) \n\n## Tips \n- Start by exploring the dataset and trying out the sample submission. \n- Share your ideas, notebooks, and questions here in the discussion forum. \n- Ask for help and helps others! Collaboration helps everyone grow. \n\nWe’re excited to see your approaches and solutions. \nGood luck! \n\n— *The Grand X-Ray Slam Team*\n", "votes": 3 }, { "id": 3275147, "postDate": "2025-08-26T02:09:34.890Z", "content": "I have one question: would it be permitted to write a research paper on this dataset, based on my solution, after the competition ends? @guntasdhanjal
", "rawMarkdown": "I have one question: would it be permitted to write a research paper on this dataset, based on my solution, after the competition ends? @guntasdhanjal ", "votes": 1, "replies": [ { "id": 3275276, "postDate": "2025-08-26T08:54:38.690Z", "content": "yes sure
", "rawMarkdown": "yes sure \n", "votes": 1 }, { "id": 3312599, "postDate": "2025-11-07T14:08:15.240Z", "content": "hi! I am a student of data science and computer science.I daily work on kaggle post dataset and notebook. @ibrahimqasimi view my profile and vote my dataset and notebook as special vote dataset related to youtube tranding top 100 music. And Help @ibrahimqasimi me in different dataset and notebook sir please.
", "rawMarkdown": "hi! I am a student of data science and computer science.I daily work on kaggle post dataset and notebook. @ibrahimqasimi view my profile and vote my dataset and notebook as special vote dataset related to youtube tranding top 100 music. And Help @ibrahimqasimi me in different dataset and notebook sir please.\n", "votes": 1, "replies": [ { "id": 3319458, "postDate": "2025-11-11T19:06:45.023Z", "content": "sure same for u
", "rawMarkdown": "sure same for u " } ] } ] }, { "id": 3281716, "postDate": "2025-09-04T18:33:19.603Z", "content": "can we use the same model architecture for both division A and B, but use the provided dataset in every division, like making the same the model but for each division i use only its dataset @guntasdhanjal
", "rawMarkdown": "can we use the same model architecture for both division A and B, but use the provided dataset in every division, like making the same the model but for each division i use only its dataset @guntasdhanjal ", "replies": [ { "id": 3281919, "postDate": "2025-09-05T08:27:40.573Z", "content": "Yes, you’re free to use the same architecture across both divisions. The only requirement is that each division’s model must be trained on its respective dataset
", "rawMarkdown": "Yes, you’re free to use the same architecture across both divisions. The only requirement is that each division’s model must be trained on its respective dataset", "votes": 1 } ] }, { "id": 3277472, "postDate": "2025-08-28T02:36:08.123Z", "content": "I have another question:
\nIn the rules section, it stated \"The use of external data and models is acceptable unless specifically prohibited by the Host\". Therefore, I believe it is okay to use pretrain weights/models.
\nHowever in the overview, it says differently:
\nParticipation Rules\n Teams: Solo or up to 4 members\n External data: \n Use: Competition & research only\nSo which one should we follow? So sorry for bothering. @guntasdhanjal
", "rawMarkdown": "I have another question: \n\nIn the rules section, it stated \"The use of external data and models is acceptable unless specifically prohibited by the Host\". Therefore, I believe it is okay to use pretrain weights/models.\n\nHowever in the overview, it says differently: \n```md\nParticipation Rules\n- Teams: Solo or up to 4 members\n- External data: **Not allowed**\n- Use: Competition & research only\n```\nSo which one should we follow? So sorry for bothering. @guntasdhanjal ", "replies": [ { "id": 3278717, "postDate": "2025-08-30T12:33:58.023Z", "content": "Thanks for catching that! You’re right, the two sections aren’t aligned. To clarify: please follow the Participation Rules in the overview external data is not allowed. The line in the rules section about external models was a leftover and we’ll fix it to avoid confusion. Really appreciate you double-checking instead of assuming!
", "rawMarkdown": "Thanks for catching that! You’re right, the two sections aren’t aligned. To clarify: please follow the Participation Rules in the overview external data is not allowed. The line in the rules section about external models was a leftover and we’ll fix it to avoid confusion. Really appreciate you double-checking instead of assuming!", "votes": 1 } ] }, { "id": 3276923, "postDate": "2025-08-27T02:52:08.227Z", "content": "Am I allowed to join and submit to both Division A and B?
\nI see that this hackathon has 2 parts, Division A and Division B. However, the rules doesn't explicitly state that a player is not allowed to join both parts.
\nSo does it mean that I can reuse my model from division A for division B? If so, what are the differences between division A and B?
\nThanks a lot!
Yes you can join both competitions.
", "rawMarkdown": "Yes you can join both competitions.", "replies": [ { "id": 3277781, "postDate": "2025-08-28T16:34:40.397Z", "content": "Can I use these two datasets and develop one model to submit both competitions?
", "rawMarkdown": "Can I use these two datasets and develop one model to submit both competitions? " }, { "id": 3278502, "postDate": "2025-08-29T20:54:59.913Z", "content": "I also had the same doubt about using both datasets to train the model, but I figured in that case having the two separate competitions would makes no sense.
", "rawMarkdown": "I also had the same doubt about using both datasets to train the model, but I figured in that case having the two separate competitions would makes no sense." }, { "id": 3278718, "postDate": "2025-08-30T12:35:15.913Z", "content": "Good question! Yes, you’re welcome to participate in both Division A and Division B. The only condition is that you should not reuse the exact same model across both — each division is meant to showcase different approaches.
\nIn short:
\nYou can join both.
\nBut you need to submit a different model for Division A vs. Division B.
The differences between the two divisions are mainly in their focus and evaluation setup (A is , B is [← you can fill in specifics]). This way, participants get to explore distinct problem angles rather than just recycling a single solution.
\nThanks for asking glad you brought it up!
", "rawMarkdown": "Good question! Yes, you’re welcome to participate in both Division A and Division B. The only condition is that you should not reuse the exact same model across both — each division is meant to showcase different approaches.\n\nIn short:\n\nYou can join both.\nBut you need to submit a different model for Division A vs. Division B.\n\nThe differences between the two divisions are mainly in their focus and evaluation setup (A is ___, B is ___ [← you can fill in specifics]). This way, participants get to explore distinct problem angles rather than just recycling a single solution.\n\nThanks for asking glad you brought it up!" }, { "id": 3278719, "postDate": "2025-08-30T12:35:38.450Z", "content": "Thanks for asking! No, you can’t use a single model for both divisions. Each division is designed to test different aspects, so you’ll need to develop and submit separate models for Division A and Division B.
", "rawMarkdown": "Thanks for asking! No, you can’t use a single model for both divisions. Each division is designed to test different aspects, so you’ll need to develop and submit separate models for Division A and Division B." }, { "id": 3278776, "postDate": "2025-08-30T13:33:47.730Z", "content": "Hi!
\nIs it allowed to have two different approaches/models for both compeititons but using the data of both DivisionA+B while training each approach?
", "rawMarkdown": "Hi!\n\nIs it allowed to have two different approaches/models for both compeititons but using the data of both DivisionA+B while training each approach?" }, { "id": 3280150, "postDate": "2025-09-02T08:30:13.443Z", "content": "Great question! For fairness, each Division must be trained only on its respective dataset. Using Division A + B together to train a single model (or to transfer features) is not permitted.
\nHowever, you are free to build different models/approaches for each Division separately.
", "rawMarkdown": "Great question! For fairness, each Division must be trained only on its respective dataset. Using Division A + B together to train a single model (or to transfer features) is not permitted.\n\nHowever, you are free to build different models/approaches for each Division separately.", "votes": 1 } ] }, { "id": 3277782, "postDate": "2025-08-28T16:38:53.997Z", "content": "I think the datasets are different. But I am also curious about why split into two division?
", "rawMarkdown": "I think the datasets are different. But I am also curious about why split into two division?", "replies": [ { "id": 3278503, "postDate": "2025-08-29T20:55:27.780Z", "content": "I imagine the reason was to just verify robustness better.
", "rawMarkdown": "I imagine the reason was to just verify robustness better. \n" }, { "id": 3278720, "postDate": "2025-08-30T12:38:03.520Z", "content": "You’re right the datasets are different. The reason we split the hackathon into two divisions is mainly about accessibility and fairness.
\nThe full dataset is over 1TB, which can be a barrier for people without high-end resources (like large storage or paid GPU access). By creating two divisions, we give more participants the chance to compete — even those who want to work with lighter setups.
\nLater on, for evaluation and deployment, we’ll also be focusing on lighter models, so this split makes sure everyone can participate meaningfully without heavy infrastructure requirements.
", "rawMarkdown": "You’re right the datasets are different. The reason we split the hackathon into two divisions is mainly about accessibility and fairness.\n\nThe full dataset is over 1TB, which can be a barrier for people without high-end resources (like large storage or paid GPU access). By creating two divisions, we give more participants the chance to compete — even those who want to work with lighter setups.\n\nLater on, for evaluation and deployment, we’ll also be focusing on lighter models, so this split makes sure everyone can participate meaningfully without heavy infrastructure requirements.", "votes": 1 } ] } ] }, { "id": 3282056, "postDate": "2025-09-05T14:00:26.183Z", "content": "@guntasdhanjal Are submissions are working I am unable to load the file ?
", "rawMarkdown": "@guntasdhanjal Are submissions are working I am unable to load the file ?", "isDeleted": true, "replies": [ { "id": 3282058, "postDate": "2025-09-05T14:06:01.507Z", "content": "Hi @asteyagaur , thanks for flagging this.
\nPlease note that in this competition, submissions are limited to 2 per day (as mentioned in the rules). If you’ve already submitted 2 times today, further submissions will not go through.
If the error appears even when you’re within the 2-per-day limit, could you please share the exact error message you see? That will help us check if it’s a file formatting issue instead.
", "rawMarkdown": "Hi @asteyagaur , thanks for flagging this.\nPlease note that in this competition, submissions are limited to 2 per day (as mentioned in the rules). If you’ve already submitted 2 times today, further submissions will not go through.\n\nIf the error appears even when you’re within the 2-per-day limit, could you please share the exact error message you see? That will help us check if it’s a file formatting issue instead.", "replies": [ { "id": 3282060, "postDate": "2025-09-05T14:18:04.167Z", "content": "@guntasdhanjal I think it is working now . Sorry I should have removed above comment . Might be some Internet issue from my side .
", "rawMarkdown": "@guntasdhanjal I think it is working now . Sorry I should have removed above comment . Might be some Internet issue from my side .", "isDeleted": true }, { "id": 3282066, "postDate": "2025-09-05T14:36:21.383Z", "content": "No worries @asteyagaur Glad it’s working now. Thanks for the update
", "rawMarkdown": "No worries @asteyagaur Glad it’s working now. Thanks for the update" } ] } ] }, { "id": 3281760, "postDate": "2025-09-04T21:22:39.083Z", "content": "@guntasdhanjal Can we use test data in our training ?
", "rawMarkdown": "@guntasdhanjal Can we use test data in our training ?", "isDeleted": true, "replies": [ { "id": 3281918, "postDate": "2025-09-05T08:27:06.217Z", "content": "No, test data cannot be used for training. The test set is strictly for evaluation purposes. You should only train on the provided training data.
\nFor a starting point, you can check out the CNN Baseline Model Notebook.
I have one question: would it be permitted to write a research paper on this dataset, based on my solution, after the competition ends? @guntasdhanjal
", "votes": 1, "replies": [ { "id": 3275276, "author_name": "Guntas Dhanjal", "author_url": "", "post_date": "2025-08-26T08:54:38.690000", "content": "yes sure
", "votes": 1, "replies": [] }, { "id": 3312599, "author_name": "", "author_url": "", "post_date": "2025-11-07T14:08:15.240000", "content": "hi! I am a student of data science and computer science.I daily work on kaggle post dataset and notebook. @ibrahimqasimi view my profile and vote my dataset and notebook as special vote dataset related to youtube tranding top 100 music. And Help @ibrahimqasimi me in different dataset and notebook sir please.
", "votes": 1, "replies": [ { "id": 3319458, "author_name": "", "author_url": "", "post_date": "2025-11-11T19:06:45.023000", "content": "sure same for u
", "votes": 0, "replies": [] } ] } ] }, { "id": 3281716, "author_name": "Masry1", "author_url": "", "post_date": "2025-09-04T18:33:19.603000", "content": "can we use the same model architecture for both division A and B, but use the provided dataset in every division, like making the same the model but for each division i use only its dataset @guntasdhanjal
", "votes": 0, "replies": [ { "id": 3281919, "author_name": "Guntas Dhanjal", "author_url": "", "post_date": "2025-09-05T08:27:40.573000", "content": "Yes, you’re free to use the same architecture across both divisions. The only requirement is that each division’s model must be trained on its respective dataset
", "votes": 1, "replies": [] } ] }, { "id": 3277472, "author_name": "Hiếu Lê Ngọc", "author_url": "", "post_date": "2025-08-28T02:36:08.123000", "content": "I have another question:
\nIn the rules section, it stated \"The use of external data and models is acceptable unless specifically prohibited by the Host\". Therefore, I believe it is okay to use pretrain weights/models.
\nHowever in the overview, it says differently:
\nParticipation Rules\n Teams: Solo or up to 4 members\n External data: \n Use: Competition & research only\nSo which one should we follow? So sorry for bothering. @guntasdhanjal
", "votes": 0, "replies": [ { "id": 3278717, "author_name": "Guntas Dhanjal", "author_url": "", "post_date": "2025-08-30T12:33:58.023000", "content": "Thanks for catching that! You’re right, the two sections aren’t aligned. To clarify: please follow the Participation Rules in the overview external data is not allowed. The line in the rules section about external models was a leftover and we’ll fix it to avoid confusion. Really appreciate you double-checking instead of assuming!
", "votes": 1, "replies": [] } ] }, { "id": 3276923, "author_name": "Hiếu Lê Ngọc", "author_url": "", "post_date": "2025-08-27T02:52:08.227000", "content": "Am I allowed to join and submit to both Division A and B?
\nI see that this hackathon has 2 parts, Division A and Division B. However, the rules doesn't explicitly state that a player is not allowed to join both parts.
\nSo does it mean that I can reuse my model from division A for division B? If so, what are the differences between division A and B?
\nThanks a lot!
Yes you can join both competitions.
", "votes": 0, "replies": [ { "id": 3277781, "author_name": "Luxuer", "author_url": "", "post_date": "2025-08-28T16:34:40.397000", "content": "Can I use these two datasets and develop one model to submit both competitions?
", "votes": 0, "replies": [] }, { "id": 3278502, "author_name": "Alessandro Ciano", "author_url": "", "post_date": "2025-08-29T20:54:59.913000", "content": "I also had the same doubt about using both datasets to train the model, but I figured in that case having the two separate competitions would makes no sense.
", "votes": 0, "replies": [] }, { "id": 3278718, "author_name": "Guntas Dhanjal", "author_url": "", "post_date": "2025-08-30T12:35:15.913000", "content": "Good question! Yes, you’re welcome to participate in both Division A and Division B. The only condition is that you should not reuse the exact same model across both — each division is meant to showcase different approaches.
\nIn short:
\nYou can join both.
\nBut you need to submit a different model for Division A vs. Division B.
The differences between the two divisions are mainly in their focus and evaluation setup (A is , B is [← you can fill in specifics]). This way, participants get to explore distinct problem angles rather than just recycling a single solution.
\nThanks for asking glad you brought it up!
", "votes": 0, "replies": [] }, { "id": 3278719, "author_name": "Guntas Dhanjal", "author_url": "", "post_date": "2025-08-30T12:35:38.450000", "content": "Thanks for asking! No, you can’t use a single model for both divisions. Each division is designed to test different aspects, so you’ll need to develop and submit separate models for Division A and Division B.
", "votes": 0, "replies": [] }, { "id": 3278776, "author_name": "Harshit Sheoran", "author_url": "", "post_date": "2025-08-30T13:33:47.730000", "content": "Hi!
\nIs it allowed to have two different approaches/models for both compeititons but using the data of both DivisionA+B while training each approach?
", "votes": 0, "replies": [] }, { "id": 3280150, "author_name": "Guntas Dhanjal", "author_url": "", "post_date": "2025-09-02T08:30:13.443000", "content": "Great question! For fairness, each Division must be trained only on its respective dataset. Using Division A + B together to train a single model (or to transfer features) is not permitted.
\nHowever, you are free to build different models/approaches for each Division separately.
", "votes": 1, "replies": [] } ] }, { "id": 3277782, "author_name": "Luxuer", "author_url": "", "post_date": "2025-08-28T16:38:53.997000", "content": "I think the datasets are different. But I am also curious about why split into two division?
", "votes": 0, "replies": [ { "id": 3278503, "author_name": "Alessandro Ciano", "author_url": "", "post_date": "2025-08-29T20:55:27.780000", "content": "I imagine the reason was to just verify robustness better.
", "votes": 0, "replies": [] }, { "id": 3278720, "author_name": "Guntas Dhanjal", "author_url": "", "post_date": "2025-08-30T12:38:03.520000", "content": "You’re right the datasets are different. The reason we split the hackathon into two divisions is mainly about accessibility and fairness.
\nThe full dataset is over 1TB, which can be a barrier for people without high-end resources (like large storage or paid GPU access). By creating two divisions, we give more participants the chance to compete — even those who want to work with lighter setups.
\nLater on, for evaluation and deployment, we’ll also be focusing on lighter models, so this split makes sure everyone can participate meaningfully without heavy infrastructure requirements.
", "votes": 1, "replies": [] } ] } ] }, { "id": 3282056, "author_name": "", "author_url": "", "post_date": "2025-09-05T14:00:26.183000", "content": "@guntasdhanjal Are submissions are working I am unable to load the file ?
", "votes": 0, "replies": [ { "id": 3282058, "author_name": "Guntas Dhanjal", "author_url": "", "post_date": "2025-09-05T14:06:01.507000", "content": "Hi @asteyagaur , thanks for flagging this.
\nPlease note that in this competition, submissions are limited to 2 per day (as mentioned in the rules). If you’ve already submitted 2 times today, further submissions will not go through.
If the error appears even when you’re within the 2-per-day limit, could you please share the exact error message you see? That will help us check if it’s a file formatting issue instead.
", "votes": 0, "replies": [ { "id": 3282060, "author_name": "", "author_url": "", "post_date": "2025-09-05T14:18:04.167000", "content": "@guntasdhanjal I think it is working now . Sorry I should have removed above comment . Might be some Internet issue from my side .
", "votes": 0, "replies": [] }, { "id": 3282066, "author_name": "Guntas Dhanjal", "author_url": "", "post_date": "2025-09-05T14:36:21.383000", "content": "No worries @asteyagaur Glad it’s working now. Thanks for the update
", "votes": 0, "replies": [] } ] } ] }, { "id": 3281760, "author_name": "", "author_url": "", "post_date": "2025-09-04T21:22:39.083000", "content": "@guntasdhanjal Can we use test data in our training ?
", "votes": 0, "replies": [ { "id": 3281918, "author_name": "Guntas Dhanjal", "author_url": "", "post_date": "2025-09-05T08:27:06.217000", "content": "No, test data cannot be used for training. The test set is strictly for evaluation purposes. You should only train on the provided training data.
\nFor a starting point, you can check out the CNN Baseline Model Notebook.