![\"image\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F1723677%2Fb858f6500a16f9a0e3e23ee9126b0069%2Fimg1.png?generation=1602506040379499&alt=media\")
In addition to this competition forum, you can continue the discussion in our official Kaggle Discord Server here: discord.gg/kaggle
\nThe Discord is a great place to ask getting started questions, chat about the nuances of this competition, and connect with potential team mates. Learn more about Discord at our announcement here. Here are a few things to keep in mind though:
\n1. Discord Competition Channels are 'Public'
\nDiscord channels for specific competitions are considered 'public' spaces where you are allowed to talk about competition details (it will not count as private sharing).
2. Discord Competition Channels are Not Monitored by Staff
\nKaggle Staff and Hosts running competitions will not monitor Discord or be available to answer questions in Discord. Always post important questions in the forums.
3. Keep the Good Stuff on the Forums
\nPlease keep important questions, insights, writeups, and other valuable conversation on the Kaggle forums. Discord is intended to be a more casual space to discuss competitions and help each other, we want to keep all the best information on the forums.
4. Remember to never privately share competition code or data
\nPlease remember that private sharing of competition code or data is, as always, not permitted. Code sharing must always be done publicly through the Kaggle forums/notebooks.
I hope you’ll join us to chat on Discord soon!
", "messages": [ { "id": 3176081, "postDate": "2025-04-11T00:48:17.270Z", "content": "Great work—your effort truly stands out!
", "rawMarkdown": "Great work—your effort truly stands out!", "votes": 2 }, { "id": 3181081, "postDate": "2025-04-17T12:49:28.347Z", "content": "This competition is surely challenging, yet the knowledge those challenges provided is immense in the field. They would help implementation in other challenges too.
", "rawMarkdown": "This competition is surely challenging, yet the knowledge those challenges provided is immense in the field. They would help implementation in other challenges too.", "votes": 1 }, { "id": 2751657, "postDate": "2024-04-14T11:56:02.930Z", "content": "Someone asked the question if it is really possible to get score above 0.998 without using whole dataset? So here is my opinion.
\nAchieving a score above 0.998 on the MNIST dataset without using a model trained on the complete dataset requires advanced techniques. Strategies such as data augmentation, optimized model architecture, and hyperparameter tuning are crucial. Additionally, leveraging transfer learning and exploring advanced techniques like generative adversarial networks or capsule networks can further enhance performance. While challenging, these strategies offer promising avenues for improving accuracy.
\n
This is a very helpful opinion, I will implement it in subsequent versions of notebooks. Thanks!
", "rawMarkdown": "This is a very helpful opinion, I will implement it in subsequent versions of notebooks. Thanks!", "votes": 2 }, { "id": 2890178, "postDate": "2024-06-26T01:41:15.863Z", "content": "Thanks for your valuable suggestion.
", "rawMarkdown": "Thanks for your valuable suggestion.", "votes": 2 }, { "id": 3116767, "postDate": "2025-02-06T09:37:36.900Z", "content": "thanks! for this wonderful suggestion
", "rawMarkdown": "thanks! for this wonderful suggestion" }, { "id": 3212390, "postDate": "2025-05-29T18:46:40.697Z", "content": "A good way to verify if your model architecture is achieving it's full potential
", "rawMarkdown": "A good way to verify if your model architecture is achieving it's full potential" } ] }, { "id": 2378717, "postDate": "2023-08-07T19:03:37.900Z", "content": "In addition to this competition forum, you can continue the discussion in our official Kaggle Discord Server here: discord.gg/kaggle
\nThe Discord is a great place to ask getting started questions, chat about the nuances of this competition, and connect with potential team mates. Learn more about Discord at our announcement here. Here are a few things to keep in mind though:
\n1. Discord Competition Channels are 'Public'
\nDiscord channels for specific competitions are considered 'public' spaces where you are allowed to talk about competition details (it will not count as private sharing).
2. Discord Competition Channels are Not Monitored by Staff
\nKaggle Staff and Hosts running competitions will not monitor Discord or be available to answer questions in Discord. Always post important questions in the forums.
3. Keep the Good Stuff on the Forums
\nPlease keep important questions, insights, writeups, and other valuable conversation on the Kaggle forums. Discord is intended to be a more casual space to discuss competitions and help each other, we want to keep all the best information on the forums.
4. Remember to never privately share competition code or data
\nPlease remember that private sharing of competition code or data is, as always, not permitted. Code sharing must always be done publicly through the Kaggle forums/notebooks.
I hope you’ll join us to chat on Discord soon!
", "rawMarkdown": "In addition to this competition forum, you can continue the discussion in our official Kaggle Discord Server here: [discord.gg/kaggle](http://discord.gg/kaggle)\n\nThe Discord is a great place to ask getting started questions, chat about the nuances of this competition, and connect with potential team mates. Learn more about Discord at our [announcement here](https://www.kaggle.com/discussions/general/429933). Here are a few things to keep in mind though:\n\n**1. Discord Competition Channels are 'Public'**\nDiscord channels for specific competitions are considered 'public' spaces where you are allowed to talk about competition details (it will not count as private sharing).\n\n**2. Discord Competition Channels are Not Monitored by Staff**\nKaggle Staff and Hosts running competitions will not monitor Discord or be available to answer questions in Discord. Always post important questions in the forums.\n\n**3. Keep the Good Stuff on the Forums**\nPlease keep important questions, insights, writeups, and other valuable conversation on the Kaggle forums. Discord is intended to be a more casual space to discuss competitions and help each other, we want to keep all the best information on the forums.\n\n**4. Remember to never privately share competition code or data**\nPlease remember that private sharing of competition code or data is, as always, not permitted. Code sharing must always be done publicly through the Kaggle forums/notebooks.\n\nI hope you’ll join us to chat on Discord soon!", "votes": 40 }, { "id": 2846700, "postDate": "2024-05-31T08:06:21.253Z", "content": "Hi everyone, this competition was a bit challenging , but I enjoyed it
", "rawMarkdown": "Hi everyone, this competition was a bit challenging , but I enjoyed it ", "votes": 1 }, { "id": 3231172, "postDate": "2025-06-24T03:15:31.403Z", "content": "This a a great beginner friendly competition, and it's improvements to implementation can be made throughout the self-learning process.
", "rawMarkdown": "This a a great beginner friendly competition, and it's improvements to implementation can be made throughout the self-learning process." }, { "id": 3214565, "postDate": "2025-05-31T19:34:53.487Z", "content": "Classic CV task—perfect for applying CNNs to reshaped pixel data. Excited to dive in!
", "rawMarkdown": "Classic CV task—perfect for applying CNNs to reshaped pixel data. Excited to dive in!" }, { "id": 3191087, "postDate": "2025-05-01T11:26:59.093Z", "content": "Хорошая работа, спасибо
", "rawMarkdown": "Хорошая работа, спасибо" }, { "id": 3121465, "postDate": "2025-02-11T16:13:31.830Z", "content": "Easy and beginner friendly
", "rawMarkdown": "Easy and beginner friendly" }, { "id": 2954504, "postDate": "2024-08-09T18:38:36.590Z", "content": "Beginner friendly competition, thank you!
", "rawMarkdown": "Beginner friendly competition, thank you!" }, { "id": 2904705, "postDate": "2024-07-04T14:27:39.410Z", "content": "What is considered to be a good score on this competition? Does 99% and above show your proficiency in such tasks?
", "rawMarkdown": "What is considered to be a good score on this competition? Does 99% and above show your proficiency in such tasks?" }, { "id": 2821768, "postDate": "2024-05-18T08:32:57.960Z", "content": "hello everyone!
", "rawMarkdown": "hello everyone!" }, { "id": 2740425, "postDate": "2024-04-07T19:12:05.407Z", "content": "helloi helloi helloi
", "rawMarkdown": "helloi helloi helloi" }, { "id": 2740424, "postDate": "2024-04-07T19:11:21.683Z", "content": "helloi helloi helloi
", "rawMarkdown": "helloi helloi helloi" }, { "id": 2727558, "postDate": "2024-04-01T20:09:59.790Z", "content": "hiiiiiiiiiiiiiiii
", "rawMarkdown": "hiiiiiiiiiiiiiiii" }, { "id": 2726224, "postDate": "2024-04-01T03:59:19.503Z", "content": "Wonderful
", "rawMarkdown": "Wonderful " }, { "id": 2708642, "postDate": "2024-03-21T06:22:06.857Z", "content": "Wonderful. Will start working on it
", "rawMarkdown": "Wonderful. Will start working on it" }, { "id": 2702648, "postDate": "2024-03-17T18:51:00.943Z", "content": "Lol. This dataset is too easy.
", "rawMarkdown": "Lol. This dataset is too easy." }, { "id": 2671238, "postDate": "2024-02-27T12:02:26.403Z", "content": "interesting
", "rawMarkdown": "interesting" }, { "id": 2616412, "postDate": "2024-01-23T16:02:57.973Z", "content": "Wow. this has a discord server ?
", "rawMarkdown": "Wow. this has a discord server ?", "replies": [ { "id": 2891058, "postDate": "2024-06-26T13:30:31.990Z", "content": "Yeah, it's strange, to be honest)
", "rawMarkdown": "Yeah, it's strange, to be honest)" } ] }, { "id": 2582200, "postDate": "2024-01-01T12:53:42.533Z", "content": "thanx for the info
", "rawMarkdown": "thanx for the info" }, { "id": 3379378, "postDate": "2025-12-19T17:39:35.100Z", "rawMarkdown": "", "isDeleted": true }, { "id": 3214562, "postDate": "2025-05-31T19:32:10.587Z", "rawMarkdown": "", "isDeleted": true }, { "id": 3307864, "postDate": "2025-10-28T01:51:23.463Z", "content": "thanks for the info bro
", "rawMarkdown": "thanks for the info bro" }, { "id": 3163654, "postDate": "2025-03-31T04:19:04.807Z", "content": "Thanks for info!!
", "rawMarkdown": "Thanks for info!!" }, { "id": 2876181, "postDate": "2024-06-17T15:28:33.953Z", "content": "thank you for making this discussion
", "rawMarkdown": "thank you for making this discussion" }, { "id": 2870894, "postDate": "2024-06-13T20:50:30.160Z", "content": "Very informative, thank you.
", "rawMarkdown": "Very informative, thank you. " }, { "id": 2832948, "postDate": "2024-05-24T01:06:22.517Z", "content": "thank you info !
", "rawMarkdown": "thank you info !" }, { "id": 2736015, "postDate": "2024-04-05T02:00:16.917Z", "content": "thank you for setting this up!
", "rawMarkdown": "thank you for setting this up!\n" }, { "id": 2734091, "postDate": "2024-04-03T23:30:02.563Z", "content": "Thank you!
", "rawMarkdown": "Thank you! " }, { "id": 2684454, "postDate": "2024-03-06T16:11:29.893Z", "content": "thanks! good luck
", "rawMarkdown": "thanks! good luck" }, { "id": 2632366, "postDate": "2024-02-02T10:08:54.147Z", "content": "Thank you bro
", "rawMarkdown": "Thank you bro" }, { "id": 2602319, "postDate": "2024-01-15T04:25:43.980Z", "content": "Thanks for info
", "rawMarkdown": "Thanks for info" }, { "id": 2590858, "postDate": "2024-01-07T13:56:04.433Z", "content": "Thanks for sharing the notes
", "rawMarkdown": "Thanks for sharing the notes" }, { "id": 2566100, "postDate": "2023-12-18T13:02:37.310Z", "content": "Thanks for info👍
", "rawMarkdown": "Thanks for info👍" }, { "id": 2428640, "postDate": "2023-09-08T04:21:18.380Z", "content": "Thanks for sharing this info!!
", "rawMarkdown": "Thanks for sharing this info!!" }, { "id": 3125800, "postDate": "2025-02-16T16:36:31.137Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2747510, "postDate": "2024-04-11T23:51:16.860Z", "content": "Thanks for the information
", "rawMarkdown": "Thanks for the information", "isDeleted": true } ], "comments": [ { "id": 3176081, "author_name": "", "author_url": "", "post_date": "2025-04-11T00:48:17.270000", "content": "Great work—your effort truly stands out!
", "votes": 2, "replies": [] }, { "id": 3181081, "author_name": "Himanshu Singh", "author_url": "", "post_date": "2025-04-17T12:49:28.347000", "content": "This competition is surely challenging, yet the knowledge those challenges provided is immense in the field. They would help implementation in other challenges too.
", "votes": 1, "replies": [] }, { "id": 2751657, "author_name": "Aditya Kumar Tiwari", "author_url": "", "post_date": "2024-04-14T11:56:02.930000", "content": "Someone asked the question if it is really possible to get score above 0.998 without using whole dataset? So here is my opinion.
\nAchieving a score above 0.998 on the MNIST dataset without using a model trained on the complete dataset requires advanced techniques. Strategies such as data augmentation, optimized model architecture, and hyperparameter tuning are crucial. Additionally, leveraging transfer learning and exploring advanced techniques like generative adversarial networks or capsule networks can further enhance performance. While challenging, these strategies offer promising avenues for improving accuracy.
\n
This is a very helpful opinion, I will implement it in subsequent versions of notebooks. Thanks!
", "votes": 2, "replies": [] }, { "id": 2890178, "author_name": "ashik1104", "author_url": "", "post_date": "2024-06-26T01:41:15.863000", "content": "Thanks for your valuable suggestion.
", "votes": 2, "replies": [] }, { "id": 3116767, "author_name": "", "author_url": "", "post_date": "2025-02-06T09:37:36.900000", "content": "thanks! for this wonderful suggestion
", "votes": 0, "replies": [] }, { "id": 3212390, "author_name": "Wisam Hussain", "author_url": "", "post_date": "2025-05-29T18:46:40.697000", "content": "A good way to verify if your model architecture is achieving it's full potential
", "votes": 0, "replies": [] } ] }, { "id": 2846700, "author_name": "Yousef Mohammad", "author_url": "", "post_date": "2024-05-31T08:06:21.253000", "content": "Hi everyone, this competition was a bit challenging , but I enjoyed it
", "votes": 1, "replies": [] }, { "id": 3231172, "author_name": "OrangeByte42", "author_url": "", "post_date": "2025-06-24T03:15:31.403000", "content": "This a a great beginner friendly competition, and it's improvements to implementation can be made throughout the self-learning process.
", "votes": 0, "replies": [] }, { "id": 3214565, "author_name": "Baibhav-Paudel7", "author_url": "", "post_date": "2025-05-31T19:34:53.487000", "content": "Classic CV task—perfect for applying CNNs to reshaped pixel data. Excited to dive in!
", "votes": 0, "replies": [] }, { "id": 3191087, "author_name": "Osman Malsagov", "author_url": "", "post_date": "2025-05-01T11:26:59.093000", "content": "Хорошая работа, спасибо
", "votes": 0, "replies": [] }, { "id": 3121465, "author_name": "Vihan Tiwari", "author_url": "", "post_date": "2025-02-11T16:13:31.830000", "content": "Easy and beginner friendly
", "votes": 0, "replies": [] }, { "id": 2954504, "author_name": "Bader Gorchene", "author_url": "", "post_date": "2024-08-09T18:38:36.590000", "content": "Beginner friendly competition, thank you!
", "votes": 0, "replies": [] }, { "id": 2904705, "author_name": "Arystan Kairbayev", "author_url": "", "post_date": "2024-07-04T14:27:39.410000", "content": "What is considered to be a good score on this competition? Does 99% and above show your proficiency in such tasks?
", "votes": 0, "replies": [] }, { "id": 2821768, "author_name": "Phuoc Loc", "author_url": "", "post_date": "2024-05-18T08:32:57.960000", "content": "hello everyone!
", "votes": 0, "replies": [] }, { "id": 2740425, "author_name": "saju s mukundan", "author_url": "", "post_date": "2024-04-07T19:12:05.407000", "content": "helloi helloi helloi
", "votes": 0, "replies": [] }, { "id": 2740424, "author_name": "Qaswin#7356", "author_url": "", "post_date": "2024-04-07T19:11:21.683000", "content": "helloi helloi helloi
", "votes": 0, "replies": [] }, { "id": 2727558, "author_name": "sidhuser", "author_url": "", "post_date": "2024-04-01T20:09:59.790000", "content": "hiiiiiiiiiiiiiiii
", "votes": 0, "replies": [] }, { "id": 2726224, "author_name": "Aditya Kumar", "author_url": "", "post_date": "2024-04-01T03:59:19.503000", "content": "Wonderful
", "votes": 0, "replies": [] }, { "id": 2708642, "author_name": "Hitesh T", "author_url": "", "post_date": "2024-03-21T06:22:06.857000", "content": "Wonderful. Will start working on it
", "votes": 0, "replies": [] }, { "id": 2702648, "author_name": "Julio Torres", "author_url": "", "post_date": "2024-03-17T18:51:00.943000", "content": "Lol. This dataset is too easy.
", "votes": 0, "replies": [] }, { "id": 2671238, "author_name": "Amirali Ebrahim-zadeh", "author_url": "", "post_date": "2024-02-27T12:02:26.403000", "content": "interesting
", "votes": 0, "replies": [] }, { "id": 2616412, "author_name": "Aman Raj", "author_url": "", "post_date": "2024-01-23T16:02:57.973000", "content": "Wow. this has a discord server ?
", "votes": 0, "replies": [ { "id": 2891058, "author_name": "Daniel Gutsuliak", "author_url": "", "post_date": "2024-06-26T13:30:31.990000", "content": "Yeah, it's strange, to be honest)
", "votes": 0, "replies": [] } ] }, { "id": 2582200, "author_name": "Djellal", "author_url": "", "post_date": "2024-01-01T12:53:42.533000", "content": "thanx for the info
", "votes": 0, "replies": [] }, { "id": 3379378, "author_name": "", "author_url": "", "post_date": "2025-12-19T17:39:35.100000", "content": "", "votes": 0, "replies": [] }, { "id": 3214562, "author_name": "", "author_url": "", "post_date": "2025-05-31T19:32:10.587000", "content": "", "votes": 0, "replies": [] }, { "id": 3307864, "author_name": "andreasX", "author_url": "", "post_date": "2025-10-28T01:51:23.463000", "content": "thanks for the info bro
", "votes": 0, "replies": [] }, { "id": 3163654, "author_name": "h3110W0r1d1", "author_url": "", "post_date": "2025-03-31T04:19:04.807000", "content": "Thanks for info!!
", "votes": 0, "replies": [] }, { "id": 2876181, "author_name": "DuckDuckGu", "author_url": "", "post_date": "2024-06-17T15:28:33.953000", "content": "thank you for making this discussion
", "votes": 0, "replies": [] }, { "id": 2870894, "author_name": "Logan", "author_url": "", "post_date": "2024-06-13T20:50:30.160000", "content": "Very informative, thank you.
", "votes": 0, "replies": [] }, { "id": 2832948, "author_name": "daiti", "author_url": "", "post_date": "2024-05-24T01:06:22.517000", "content": "thank you info !
", "votes": 0, "replies": [] }, { "id": 2736015, "author_name": "Yohan Su", "author_url": "", "post_date": "2024-04-05T02:00:16.917000", "content": "thank you for setting this up!
", "votes": 0, "replies": [] }, { "id": 2734091, "author_name": "kimjoey2002", "author_url": "", "post_date": "2024-04-03T23:30:02.563000", "content": "Thank you!
", "votes": 0, "replies": [] }, { "id": 2684454, "author_name": "", "author_url": "", "post_date": "2024-03-06T16:11:29.893000", "content": "", "votes": 0, "replies": [] }, { "id": 2632366, "author_name": "", "author_url": "", "post_date": "2024-02-02T10:08:54.147000", "content": "", "votes": 0, "replies": [] }, { "id": 2602319, "author_name": "", "author_url": "", "post_date": "2024-01-15T04:25:43.980000", "content": "", "votes": 0, "replies": [] }, { "id": 2590858, "author_name": "", "author_url": "", "post_date": "2024-01-07T13:56:04.433000", "content": "", "votes": 0, "replies": [] }, { "id": 2566100, "author_name": "", "author_url": "", "post_date": "2023-12-18T13:02:37.310000", "content": "", "votes": 0, "replies": [] }, { "id": 2428640, "author_name": "", "author_url": "", "post_date": "2023-09-08T04:21:18.380000", "content": "", "votes": 0, "replies": [] }, { "id": 3125800, "author_name": "", "author_url": "", "post_date": "2025-02-16T16:36:31.137000", "content": "", "votes": 0, "replies": [] }, { "id": 2747510, "author_name": "", "author_url": "", "post_date": "2024-04-11T23:51:16.860000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "3176081": "Great work—your effort truly stands out!", "3181081": "This competition is surely challenging, yet the knowledge those challenges provided is immense in the field. They would help implementation in other challenges too.", "2751657": "Someone asked the question if it is really possible to get score above 0.998 without using whole dataset? So here is my opinion.\n\n\nAchieving a score above 0.998 on the MNIST dataset without using a model trained on the complete dataset requires advanced techniques. Strategies such as data augmentation, optimized model architecture, and hyperparameter tuning are crucial. Additionally, leveraging transfer learning and exploring advanced techniques like generative adversarial networks or capsule networks can further enhance performance. While challenging, these strategies offer promising avenues for improving accuracy.\n\n", "2378717": "In addition to this competition forum, you can continue the discussion in our official Kaggle Discord Server here: [discord.gg/kaggle](http://discord.gg/kaggle)\n\nThe Discord is a great place to ask getting started questions, chat about the nuances of this competition, and connect with potential team mates. Learn more about Discord at our [announcement here](https://www.kaggle.com/discussions/general/429933). Here are a few things to keep in mind though:\n\n**1. Discord Competition Channels are 'Public'**\nDiscord channels for specific competitions are considered 'public' spaces where you are allowed to talk about competition details (it will not count as private sharing).\n\n**2. Discord Competition Channels are Not Monitored by Staff**\nKaggle Staff and Hosts running competitions will not monitor Discord or be available to answer questions in Discord. Always post important questions in the forums.\n\n**3. Keep the Good Stuff on the Forums**\nPlease keep important questions, insights, writeups, and other valuable conversation on the Kaggle forums. Discord is intended to be a more casual space to discuss competitions and help each other, we want to keep all the best information on the forums.\n\n**4. Remember to never privately share competition code or data**\nPlease remember that private sharing of competition code or data is, as always, not permitted. Code sharing must always be done publicly through the Kaggle forums/notebooks.\n\nI hope you’ll join us to chat on Discord soon!", "2846700": "Hi everyone, this competition was a bit challenging , but I enjoyed it ", "3231172": "This a a great beginner friendly competition, and it's improvements to implementation can be made throughout the self-learning process.", "3214565": "Classic CV task—perfect for applying CNNs to reshaped pixel data. Excited to dive in!", "3191087": "Хорошая работа, спасибо", "3121465": "Easy and beginner friendly", "2954504": "Beginner friendly competition, thank you!", "2904705": "What is considered to be a good score on this competition? Does 99% and above show your proficiency in such tasks?", "2821768": "hello everyone!", "2740425": "helloi helloi helloi", "2740424": "helloi helloi helloi", "2727558": "hiiiiiiiiiiiiiiii", "2726224": "Wonderful ", "2708642": "Wonderful. Will start working on it", "2702648": "Lol. This dataset is too easy.", "2671238": "interesting", "2616412": "Wow. this has a discord server ?", "2582200": "thanx for the info", "3379378": "", "3214562": "", "3307864": "thanks for the info bro", "3163654": "Thanks for info!!", "2876181": "thank you for making this discussion", "2870894": "Very informative, thank you. ", "2832948": "thank you info !", "2736015": "thank you for setting this up!\n", "2734091": "Thank you! ", "2684454": "thanks! good luck", "2632366": "Thank you bro", "2602319": "Thanks for info", "2590858": "Thanks for sharing the notes", "2566100": "Thanks for info👍", "2428640": "Thanks for sharing this info!!", "3125800": "", "2747510": "Thanks for the information" } }