I am very grateful for this competition for allowing me to learn a lot of knowledge. I'm a new kaggler, and this is my first time to participate in kaggle competition. I also paid a lot of effort and time for this, although I got zero score in the end.
\nI am also very grateful for the idea of deepflash2, which makes me feel amazing. I forked the v14 notebook https://www.kaggle.com/matjes/hubmap-efficient-sampling-deepflash2-sub and made some changes to train a new model as part of my inference. In addition, I trained my own models as another part of the model ensemble. According to all the submissions, it is true that the part of deeplash2 did not get private scores, but still thanks to the ideas provided by deepflash2.
\nThe only regret is that I didn't check whether there would be any problems again and again before the final submission, which may be caused by my lack of experience. Generally speaking, there are more gains than regrets.
", "messages": [ { "id": "1305854", "postDate": "05/13/2021 14:13:31", "content": "I am very grateful for this competition for allowing me to learn a lot of knowledge. I'm a new kaggler, and this is my first time to participate in kaggle competition. I also paid a lot of effort and time for this, although I got zero score in the end.
\nI am also very grateful for the idea of deepflash2, which makes me feel amazing. I forked the v14 notebook https://www.kaggle.com/matjes/hubmap-efficient-sampling-deepflash2-sub and made some changes to train a new model as part of my inference. In addition, I trained my own models as another part of the model ensemble. According to all the submissions, it is true that the part of deeplash2 did not get private scores, but still thanks to the ideas provided by deepflash2.
\nThe only regret is that I didn't check whether there would be any problems again and again before the final submission, which may be caused by my lack of experience. Generally speaking, there are more gains than regrets.
", "rawMarkdown": "I am very grateful for this competition for allowing me to learn a lot of knowledge. I'm a new kaggler, and this is my first time to participate in kaggle competition. I also paid a lot of effort and time for this, although I got zero score in the end.\n\nI am also very grateful for the idea of deepflash2, which makes me feel amazing. I forked the v14 notebook [https://www.kaggle.com/matjes/hubmap-efficient-sampling-deepflash2-sub](url) and made some changes to train a new model as part of my inference. In addition, I trained my own models as another part of the model ensemble. According to all the submissions, it is true that the part of deeplash2 did not get private scores, but still thanks to the ideas provided by deepflash2.\n\nThe only regret is that I didn't check whether there would be any problems again and again before the final submission, which may be caused by my lack of experience. Generally speaking, there are more gains than regrets.", "votes": null }, { "id": "1306207", "postDate": "05/13/2021 17:04:02", "content": "Thank you for your kind words.
\nWe are sad, that not everybody saw our late changes and discussions about the code.
@matjes and I wish you all the best for future competitions.
", "rawMarkdown": "Thank you for your kind words. \nWe are sad, that not everybody saw our late changes and discussions about the code. \n \n@matjes and I wish you all the best for future competitions.", "votes": null } ], "comments": [ { "id": 1306207, "author_name": "theudas", "author_url": "", "post_date": "05/13/2021 17:04:02", "content": "Thank you for your kind words.
\nWe are sad, that not everybody saw our late changes and discussions about the code.
@matjes and I wish you all the best for future competitions.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1305854": "I am very grateful for this competition for allowing me to learn a lot of knowledge. I'm a new kaggler, and this is my first time to participate in kaggle competition. I also paid a lot of effort and time for this, although I got zero score in the end.\n\nI am also very grateful for the idea of deepflash2, which makes me feel amazing. I forked the v14 notebook [https://www.kaggle.com/matjes/hubmap-efficient-sampling-deepflash2-sub](url) and made some changes to train a new model as part of my inference. In addition, I trained my own models as another part of the model ensemble. According to all the submissions, it is true that the part of deeplash2 did not get private scores, but still thanks to the ideas provided by deepflash2.\n\nThe only regret is that I didn't check whether there would be any problems again and again before the final submission, which may be caused by my lack of experience. Generally speaking, there are more gains than regrets.", "1306207": "Thank you for your kind words. \nWe are sad, that not everybody saw our late changes and discussions about the code. \n \n@matjes and I wish you all the best for future competitions." }, "source": "meta" }