{ "id": 131930, "title": "Worth seeing Kernels and Posts?", "url": "/competitions/flower-classification-with-tpus/discussion/131930", "author_name": "", "post_date": "2020-02-22T18:06:15.793829700Z", "votes": 3, "comment_count": 10, "views": 0, "content": "

Modelling

\n\n\n\n

Augmentation

\n\n\n\n

Visualisation

\n\n\n\n

Discussions

\n\n\n\n

More kernels and discussions will be added soon

", "messages": [ { "id": "753847", "postDate": "02/22/2020 18:06:15", "content": "

Modelling

\n\n\n\n

Augmentation

\n\n\n\n

Visualisation

\n\n\n\n

Discussions

\n\n\n\n

More kernels and discussions will be added soon

", "rawMarkdown": "**Modelling**\n\n- [getting-started-with-100-flowers-on-tpu](https://www.kaggle.com/mgornergoogle/getting-started-with-100-flowers-on-tpu)\n- [Concise Efficient Net starter](https://www.kaggle.com/xhlulu/flowers-tpu-concise-efficientnet-b7)\n- [Ensemble of efficient net and Densenet 201- best score](https://www.kaggle.com/wrrosa/tpu-enet-b7-densenet)\n- [Eda and baseline kernel](https://www.kaggle.com/dimitreoliveira/flower-classification-with-tpus-eda-and-baseline)\n- [Densenet 201 kernel](https://www.kaggle.com/ratan123/densenet201-flower-classification-with-tpus)\n- [Rotations Augmentations GPU/TPU](https://www.kaggle.com/cdeotte/rotation-augmentation-gpu-tpu-0-96)\n- [What's EfficientNet](https://www.kaggle.com/kurianbenoy/introduction-kernel-what-s-efficientnet/)\n- [ENetB7 + Xception+Densenet Ensemble](https://www.kaggle.com/anyexiezouqu/tpu-enet-b7-xception-densnet201)\n\n**Augmentation**\n\n- [Rotations Augmentations GPU/TPU](https://www.kaggle.com/cdeotte/rotation-augmentation-gpu-tpu-0-96)\n- [CutMix and Mixup on GPU/TPU](https://www.kaggle.com/cdeotte/cutmix-and-mixup-on-gpu-tpu)\n- [GridMask augmentation with Tensorflow](https://www.kaggle.com/xiejialun/gridmask-data-augmentation-with-tensorflow)\n- [Customised data augmentations with Tensorflow](https://www.kaggle.com/xiejialun/gridmask-data-augmentation-with-tensorflow)\n- [Making Chris data augmentation faster in GPU/TPU](https://www.kaggle.com/yihdarshieh/make-chris-deotte-s-data-augmentation-faster)\n\n**Visualisation**\n\n- [104 Flowers: Garden of Eden Starter Kernel](https://www.kaggle.com/msheriey/104-flowers-garden-of-eden-starter-kernel)\n\n**Discussions**\n\n- [TPU extreme optimizations](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/135443) by Martin Gorner\n- [How does TensorFlow use multiple devices?](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/130199)\n- https://codelabs.developers.google.com/codelabs/keras-flowers-tpu/#0\n- [Data Balancing with TPU's]( https://www.kaggle.com/c/flower-classification-with-tpus/discussion/130272)\n- [How to use mixup and Cutmix](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/132935)\n- [How to do rotations and augmentations](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/132191)\n- [TPU idle time in Kaggle](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/133447)\n- [External Data should be disallowed](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/134701)\n- [Ok I get it... TPUs are awesome! So what are the main reasons they're not ubiquitous?](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/137983)\n\nMore kernels and discussions will be added soon", "votes": null }, { "id": "753878", "postDate": "02/22/2020 19:10:34", "content": "

Hmm, can't people just sort topics by Hotness or Votes? No offense but this topic is either useless or it implies that the upvoting system is broken and can't be used to accurately evaluate discussions and kernels.

", "rawMarkdown": "Hmm, can't people just sort topics by Hotness or Votes? No offense but this topic is either useless or it implies that the upvoting system is broken and can't be used to accurately evaluate discussions and kernels.", "votes": null }, { "id": "753917", "postDate": "02/22/2020 20:12:29", "content": "

\"Flower Classification with TPUs - EDA\" by Dimitre Oliveira (with SIX oddsome Confusion Matrix), the Nine Kaggle Notebooks from Hieu Phung and Martin G.Notebook for obvious starter reasons. Every one that take a shot in that Comp will receive my upvotes. People have preferences. That are mines. But I won't make a topic with my \"favorites\" since my choices involve partiality.

", "rawMarkdown": "\"Flower Classification with TPUs - EDA\" by Dimitre Oliveira (with SIX oddsome Confusion Matrix), the Nine Kaggle Notebooks from Hieu Phung and Martin G.Notebook for obvious starter reasons. Every one that take a shot in that Comp will receive my upvotes. People have preferences. That are mines. But I won't make a topic with my \"favorites\" since my choices involve partiality.", "votes": null }, { "id": "753949", "postDate": "02/22/2020 21:08:58", "content": "

Well said @mpwolke, I would have preferred a topic titled \"Your favorite kernels and Why?\", this way Kernel creators could collect better feedback on what the community likes and improve on it ;)

", "rawMarkdown": "Well said @mpwolke, I would have preferred a topic titled \"Your favorite kernels and Why?\", this way Kernel creators could collect better feedback on what the community likes and improve on it ;)", "votes": null }, { "id": "753981", "postDate": "02/22/2020 22:24:17", "content": "

I agree with you Mourad. That Competition isn't very simple. We read topics from kagglers relating difficulties commiting/submitting or any else. We're longing to the TPU experience. And we have to support any user that try once TPU. If we state by ourselves which are the best what is the use for any one else to take a shot with TPU? Like in EVERY KAGGLE COMP the last Notebooks receive LESS UPVOTES because users have already gone to other Competition and they simply don't read ALL works, with rare exceptions. Kurian your topic is valid to discuss our points of views but NOT to declare which are the best works. That situation occurs in Survey Competition, however it's Kaggle Team (only them) who decide the winners.

", "rawMarkdown": "I agree with you Mourad. That Competition isn't very simple. We read topics from kagglers relating difficulties commiting/submitting or any else. We're longing to the TPU experience. And we have to support any user that try once TPU. If we state by ourselves which are the best what is the use for any one else to take a shot with TPU? Like in EVERY KAGGLE COMP the last Notebooks receive LESS UPVOTES because users have already gone to other Competition and they simply don't read ALL works, with rare exceptions. Kurian your topic is valid to discuss our points of views but NOT to declare which are the best works. That situation occurs in Survey Competition, however it's Kaggle Team (only them) who decide the winners.", "votes": null }, { "id": "754219", "postDate": "02/23/2020 08:29:13", "content": "

Have to agree with you! Who am I to judge someone's work as the best one. And I would like to appologize for my title. I have changed my title, as this post was inspired by similar post on Bengali AI competiton:

\n\n

https://www.kaggle.com/c/bengaliai-cv19/discussion/127976

", "rawMarkdown": "Have to agree with you! Who am I to judge someone's work as the best one. And I would like to appologize for my title. I have changed my title, as this post was inspired by similar post on Bengali AI competiton:\n\nhttps://www.kaggle.com/c/bengaliai-cv19/discussion/127976", "votes": null }, { "id": "754426", "postDate": "02/23/2020 14:38:18", "content": "

The title could be better if you add a question: What's your favorite Flower+TPU Notebook, and Why? Like Mourad suggested. Hegemony is frowned upon because put competitiveness in check. Discourage other competitiors. Even for those in the list. They don't need to optimize anything because they already achieved the top. Dimitre is already in his 37 commit, which means competitors should push harder. Don't stop till they get enough.

", "rawMarkdown": "The title could be better if you add a question: What's your favorite Flower+TPU Notebook, and Why? Like Mourad suggested. Hegemony is frowned upon because put competitiveness in check. Discourage other competitiors. Even for those in the list. They don't need to optimize anything because they already achieved the top. Dimitre is already in his 37 commit, which means competitors should push harder. Don't stop till they get enough.", "votes": null }, { "id": "778720", "postDate": "03/18/2020 17:10:54", "content": "

@kurianbenoy Thank you for putting up this post . I am getting started with this and its soooo helpful

", "rawMarkdown": "kurianbenoy Thank you for putting up this post . I am getting started with this and its soooo helpful", "votes": null }, { "id": "778773", "postDate": "03/18/2020 18:04:08", "content": "

Can you add this one : TPU: extreme optimizations ?

", "rawMarkdown": "Can you add this one : [TPU: extreme optimizations](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/135443) ?", "votes": null }, { "id": "779062", "postDate": "03/19/2020 01:53:46", "content": "

I have added it @mgornergoogle

", "rawMarkdown": "I have added it @mgornergoogle", "votes": null }, { "id": "779116", "postDate": "03/19/2020 03:15:54", "content": "

Thanks, I haven't updated much of the discussions as I haven't been active in this competition for over a week. Will update when I resume back

", "rawMarkdown": "Thanks, I haven't updated much of the discussions as I haven't been active in this competition for over a week. Will update when I resume back", "votes": null } ], "comments": [ { "id": 753878, "author_name": "msheriey", "author_url": "", "post_date": "02/22/2020 19:10:34", "content": "

Hmm, can't people just sort topics by Hotness or Votes? No offense but this topic is either useless or it implies that the upvoting system is broken and can't be used to accurately evaluate discussions and kernels.

", "votes": null, "replies": [] }, { "id": 753917, "author_name": "mpwolke", "author_url": "", "post_date": "02/22/2020 20:12:29", "content": "

\"Flower Classification with TPUs - EDA\" by Dimitre Oliveira (with SIX oddsome Confusion Matrix), the Nine Kaggle Notebooks from Hieu Phung and Martin G.Notebook for obvious starter reasons. Every one that take a shot in that Comp will receive my upvotes. People have preferences. That are mines. But I won't make a topic with my \"favorites\" since my choices involve partiality.

", "votes": null, "replies": [ { "id": 753949, "author_name": "msheriey", "author_url": "", "post_date": "02/22/2020 21:08:58", "content": "

Well said @mpwolke, I would have preferred a topic titled \"Your favorite kernels and Why?\", this way Kernel creators could collect better feedback on what the community likes and improve on it ;)

", "votes": null, "replies": [] }, { "id": 753981, "author_name": "mpwolke", "author_url": "", "post_date": "02/22/2020 22:24:17", "content": "

I agree with you Mourad. That Competition isn't very simple. We read topics from kagglers relating difficulties commiting/submitting or any else. We're longing to the TPU experience. And we have to support any user that try once TPU. If we state by ourselves which are the best what is the use for any one else to take a shot with TPU? Like in EVERY KAGGLE COMP the last Notebooks receive LESS UPVOTES because users have already gone to other Competition and they simply don't read ALL works, with rare exceptions. Kurian your topic is valid to discuss our points of views but NOT to declare which are the best works. That situation occurs in Survey Competition, however it's Kaggle Team (only them) who decide the winners.

", "votes": null, "replies": [] }, { "id": 754219, "author_name": "kurianbenoy", "author_url": "", "post_date": "02/23/2020 08:29:13", "content": "

Have to agree with you! Who am I to judge someone's work as the best one. And I would like to appologize for my title. I have changed my title, as this post was inspired by similar post on Bengali AI competiton:

\n\n

https://www.kaggle.com/c/bengaliai-cv19/discussion/127976

", "votes": null, "replies": [] }, { "id": 754426, "author_name": "mpwolke", "author_url": "", "post_date": "02/23/2020 14:38:18", "content": "

The title could be better if you add a question: What's your favorite Flower+TPU Notebook, and Why? Like Mourad suggested. Hegemony is frowned upon because put competitiveness in check. Discourage other competitiors. Even for those in the list. They don't need to optimize anything because they already achieved the top. Dimitre is already in his 37 commit, which means competitors should push harder. Don't stop till they get enough.

", "votes": null, "replies": [] } ] }, { "id": 778720, "author_name": "usharengaraju", "author_url": "", "post_date": "03/18/2020 17:10:54", "content": "

@kurianbenoy Thank you for putting up this post . I am getting started with this and its soooo helpful

", "votes": null, "replies": [ { "id": 779116, "author_name": "kurianbenoy", "author_url": "", "post_date": "03/19/2020 03:15:54", "content": "

Thanks, I haven't updated much of the discussions as I haven't been active in this competition for over a week. Will update when I resume back

", "votes": null, "replies": [] } ] }, { "id": 778773, "author_name": "mgorner", "author_url": "", "post_date": "03/18/2020 18:04:08", "content": "

Can you add this one : TPU: extreme optimizations ?

", "votes": null, "replies": [ { "id": 779062, "author_name": "kurianbenoy", "author_url": "", "post_date": "03/19/2020 01:53:46", "content": "

I have added it @mgornergoogle

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "753847": "**Modelling**\n\n- [getting-started-with-100-flowers-on-tpu](https://www.kaggle.com/mgornergoogle/getting-started-with-100-flowers-on-tpu)\n- [Concise Efficient Net starter](https://www.kaggle.com/xhlulu/flowers-tpu-concise-efficientnet-b7)\n- [Ensemble of efficient net and Densenet 201- best score](https://www.kaggle.com/wrrosa/tpu-enet-b7-densenet)\n- [Eda and baseline kernel](https://www.kaggle.com/dimitreoliveira/flower-classification-with-tpus-eda-and-baseline)\n- [Densenet 201 kernel](https://www.kaggle.com/ratan123/densenet201-flower-classification-with-tpus)\n- [Rotations Augmentations GPU/TPU](https://www.kaggle.com/cdeotte/rotation-augmentation-gpu-tpu-0-96)\n- [What's EfficientNet](https://www.kaggle.com/kurianbenoy/introduction-kernel-what-s-efficientnet/)\n- [ENetB7 + Xception+Densenet Ensemble](https://www.kaggle.com/anyexiezouqu/tpu-enet-b7-xception-densnet201)\n\n**Augmentation**\n\n- [Rotations Augmentations GPU/TPU](https://www.kaggle.com/cdeotte/rotation-augmentation-gpu-tpu-0-96)\n- [CutMix and Mixup on GPU/TPU](https://www.kaggle.com/cdeotte/cutmix-and-mixup-on-gpu-tpu)\n- [GridMask augmentation with Tensorflow](https://www.kaggle.com/xiejialun/gridmask-data-augmentation-with-tensorflow)\n- [Customised data augmentations with Tensorflow](https://www.kaggle.com/xiejialun/gridmask-data-augmentation-with-tensorflow)\n- [Making Chris data augmentation faster in GPU/TPU](https://www.kaggle.com/yihdarshieh/make-chris-deotte-s-data-augmentation-faster)\n\n**Visualisation**\n\n- [104 Flowers: Garden of Eden Starter Kernel](https://www.kaggle.com/msheriey/104-flowers-garden-of-eden-starter-kernel)\n\n**Discussions**\n\n- [TPU extreme optimizations](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/135443) by Martin Gorner\n- [How does TensorFlow use multiple devices?](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/130199)\n- https://codelabs.developers.google.com/codelabs/keras-flowers-tpu/#0\n- [Data Balancing with TPU's]( https://www.kaggle.com/c/flower-classification-with-tpus/discussion/130272)\n- [How to use mixup and Cutmix](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/132935)\n- [How to do rotations and augmentations](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/132191)\n- [TPU idle time in Kaggle](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/133447)\n- [External Data should be disallowed](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/134701)\n- [Ok I get it... TPUs are awesome! So what are the main reasons they're not ubiquitous?](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/137983)\n\nMore kernels and discussions will be added soon", "753878": "Hmm, can't people just sort topics by Hotness or Votes? No offense but this topic is either useless or it implies that the upvoting system is broken and can't be used to accurately evaluate discussions and kernels.", "753917": "\"Flower Classification with TPUs - EDA\" by Dimitre Oliveira (with SIX oddsome Confusion Matrix), the Nine Kaggle Notebooks from Hieu Phung and Martin G.Notebook for obvious starter reasons. Every one that take a shot in that Comp will receive my upvotes. People have preferences. That are mines. But I won't make a topic with my \"favorites\" since my choices involve partiality.", "753949": "Well said @mpwolke, I would have preferred a topic titled \"Your favorite kernels and Why?\", this way Kernel creators could collect better feedback on what the community likes and improve on it ;)", "753981": "I agree with you Mourad. That Competition isn't very simple. We read topics from kagglers relating difficulties commiting/submitting or any else. We're longing to the TPU experience. And we have to support any user that try once TPU. If we state by ourselves which are the best what is the use for any one else to take a shot with TPU? Like in EVERY KAGGLE COMP the last Notebooks receive LESS UPVOTES because users have already gone to other Competition and they simply don't read ALL works, with rare exceptions. Kurian your topic is valid to discuss our points of views but NOT to declare which are the best works. That situation occurs in Survey Competition, however it's Kaggle Team (only them) who decide the winners.", "754219": "Have to agree with you! Who am I to judge someone's work as the best one. And I would like to appologize for my title. I have changed my title, as this post was inspired by similar post on Bengali AI competiton:\n\nhttps://www.kaggle.com/c/bengaliai-cv19/discussion/127976", "754426": "The title could be better if you add a question: What's your favorite Flower+TPU Notebook, and Why? Like Mourad suggested. Hegemony is frowned upon because put competitiveness in check. Discourage other competitiors. Even for those in the list. They don't need to optimize anything because they already achieved the top. Dimitre is already in his 37 commit, which means competitors should push harder. Don't stop till they get enough.", "778720": "kurianbenoy Thank you for putting up this post . I am getting started with this and its soooo helpful", "778773": "Can you add this one : [TPU: extreme optimizations](https://www.kaggle.com/c/flower-classification-with-tpus/discussion/135443) ?", "779062": "I have added it @mgornergoogle", "779116": "Thanks, I haven't updated much of the discussions as I haven't been active in this competition for over a week. Will update when I resume back" }, "source": "meta" }