I just wanted to express my gratitude @nbroad for being so sharing and open in this competition. Thanks to your explanations, code and discussions I was able to train a Matcha and a Donut model which helped me compete in this competition and secure a silver place. Sharing is what makes Kaggle such a good learning platform!
\nThanks a lot and congrats to all the competitors!
", "messages": [ { "id": "2309868", "postDate": "06/20/2023 00:23:47", "content": "I just wanted to express my gratitude @nbroad for being so sharing and open in this competition. Thanks to your explanations, code and discussions I was able to train a Matcha and a Donut model which helped me compete in this competition and secure a silver place. Sharing is what makes Kaggle such a good learning platform!
\nThanks a lot and congrats to all the competitors!
", "rawMarkdown": "I just wanted to express my gratitude @nbroad for being so sharing and open in this competition. Thanks to your explanations, code and discussions I was able to train a Matcha and a Donut model which helped me compete in this competition and secure a silver place. Sharing is what makes Kaggle such a good learning platform!\n\nThanks a lot and congrats to all the competitors!", "votes": null }, { "id": "2309870", "postDate": "06/20/2023 00:27:36", "content": "Thank you! I'm glad that others found my code and comments useful! Congrats on your own silver medal!
", "rawMarkdown": "# 😊\n\nThank you! I'm glad that others found my code and comments useful! Congrats on your own silver medal!", "votes": null }, { "id": "2309878", "postDate": "06/20/2023 00:34:55", "content": "Thank you for your work.
", "rawMarkdown": "Thank you for your work.", "votes": null }, { "id": "2309882", "postDate": "06/20/2023 00:42:32", "content": "Thanks a lot @nbroad! I tried pix2struct in the first few weeks of the comp, but couldn't train it at all. Your sharing on Donut and next highlighting the issues with pix2struct training were very helpful. The Donut training and inference codes were high quality & detail oriented (e.g. catching tokenization issue with 1)!
", "rawMarkdown": "Thanks a lot @nbroad! I tried pix2struct in the first few weeks of the comp, but couldn't train it at all. Your sharing on Donut and next highlighting the issues with pix2struct training were very helpful. The Donut training and inference codes were high quality & detail oriented (e.g. catching tokenization issue with 1)!", "votes": null }, { "id": "2309994", "postDate": "06/20/2023 03:44:16", "content": "I thought I would share that the basis of this competition is well outside of my primary skillset and I originally passed up on joining in on the competition. After seeing the way @nbroad broke things down in his original donut notebook I decided to give it a try. I am generally new to Kaggle but this is one of the many reasons I have fallen in love with the platform. Thank you Nicholas I learned a lot!
", "rawMarkdown": "I thought I would share that the basis of this competition is well outside of my primary skillset and I originally passed up on joining in on the competition. After seeing the way @nbroad broke things down in his original donut notebook I decided to give it a try. I am generally new to Kaggle but this is one of the many reasons I have fallen in love with the platform. Thank you Nicholas I learned a lot!", "votes": null }, { "id": "2310036", "postDate": "06/20/2023 04:38:38", "content": "Thanks Nicholas, without you sharing ideas on how to train Matcha it would have been a whole different scenario.
", "rawMarkdown": "Thanks Nicholas, without you sharing ideas on how to train Matcha it would have been a whole different scenario.", "votes": null } ], "comments": [ { "id": 2309870, "author_name": "nbroad", "author_url": "", "post_date": "06/20/2023 00:27:36", "content": "Thank you! I'm glad that others found my code and comments useful! Congrats on your own silver medal!
", "votes": null, "replies": [ { "id": 2309878, "author_name": "catchbat", "author_url": "", "post_date": "06/20/2023 00:34:55", "content": "Thank you for your work.
", "votes": null, "replies": [] }, { "id": 2310036, "author_name": "alejopaullier", "author_url": "", "post_date": "06/20/2023 04:38:38", "content": "Thanks Nicholas, without you sharing ideas on how to train Matcha it would have been a whole different scenario.
", "votes": null, "replies": [] } ] }, { "id": 2309882, "author_name": "conjuring92", "author_url": "", "post_date": "06/20/2023 00:42:32", "content": "Thanks a lot @nbroad! I tried pix2struct in the first few weeks of the comp, but couldn't train it at all. Your sharing on Donut and next highlighting the issues with pix2struct training were very helpful. The Donut training and inference codes were high quality & detail oriented (e.g. catching tokenization issue with 1)!
", "votes": null, "replies": [] }, { "id": 2309994, "author_name": "cody11null", "author_url": "", "post_date": "06/20/2023 03:44:16", "content": "I thought I would share that the basis of this competition is well outside of my primary skillset and I originally passed up on joining in on the competition. After seeing the way @nbroad broke things down in his original donut notebook I decided to give it a try. I am generally new to Kaggle but this is one of the many reasons I have fallen in love with the platform. Thank you Nicholas I learned a lot!
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2309868": "I just wanted to express my gratitude @nbroad for being so sharing and open in this competition. Thanks to your explanations, code and discussions I was able to train a Matcha and a Donut model which helped me compete in this competition and secure a silver place. Sharing is what makes Kaggle such a good learning platform!\n\nThanks a lot and congrats to all the competitors!", "2309870": "# 😊\n\nThank you! I'm glad that others found my code and comments useful! Congrats on your own silver medal!", "2309878": "Thank you for your work.", "2309882": "Thanks a lot @nbroad! I tried pix2struct in the first few weeks of the comp, but couldn't train it at all. Your sharing on Donut and next highlighting the issues with pix2struct training were very helpful. The Donut training and inference codes were high quality & detail oriented (e.g. catching tokenization issue with 1)!", "2309994": "I thought I would share that the basis of this competition is well outside of my primary skillset and I originally passed up on joining in on the competition. After seeing the way @nbroad broke things down in his original donut notebook I decided to give it a try. I am generally new to Kaggle but this is one of the many reasons I have fallen in love with the platform. Thank you Nicholas I learned a lot!", "2310036": "Thanks Nicholas, without you sharing ideas on how to train Matcha it would have been a whole different scenario." }, "source": "meta" }