For this competition, we are offering multiple opportunities to earn TPU quota and/or cash prizes for making valued contributions to our growing TPU developer community!
\nWhat can I win?
\nEarn up to 80 hours of additional TPU quota, awarded throughout the competition
\nWin one of 3 \"TPU Star\" cash prizes of $5,000 at the close of the competition by being one of the most knowledgeable and helpful TPU experts in the community.
How can I enter?
\nSubmit a qualified notebook here anytime during the competition to be considered for either prizes.
Extra quota earners will be evaluated and awarded throughout the competition. For more details on how quota earners are evaluated, review the submission page.
\nTPU stars will be evaluated at the close of the competition.
\nGetting Started!
\nWe’ve provided the dataset converted to TFRecords and the associated scripts for creating the files and loading the data:
For this competition, we are offering multiple opportunities to earn TPU quota and/or cash prizes for making valued contributions to our growing TPU developer community!
\nWhat can I win?
\nEarn up to 80 hours of additional TPU quota, awarded throughout the competition
\nWin one of 3 \"TPU Star\" cash prizes of $5,000 at the close of the competition by being one of the most knowledgeable and helpful TPU experts in the community.
How can I enter?
\nSubmit a qualified notebook here anytime during the competition to be considered for either prizes.
Extra quota earners will be evaluated and awarded throughout the competition. For more details on how quota earners are evaluated, review the submission page.
\nTPU stars will be evaluated at the close of the competition.
\nGetting Started!
\nWe’ve provided the dataset converted to TFRecords and the associated scripts for creating the files and loading the data:
The notebooks are unavailable @philculliton
", "rawMarkdown": "The notebooks are unavailable @philculliton", "votes": null }, { "id": "1001174", "postDate": "09/07/2020 05:45:45", "content": "The 2nd and 3rd links of \"Getting Started!\" are unavailable. Please fix them, thanks!
", "rawMarkdown": "The 2nd and 3rd links of \"Getting Started!\" are unavailable. Please fix them, thanks!", "votes": null }, { "id": "1003400", "postDate": "09/08/2020 23:41:31", "content": "Can the data be updates with history_positions not equal to zero
The links should now work - sorry about that!
", "rawMarkdown": "The links should now work - sorry about that!", "votes": null }, { "id": "1005907", "postDate": "09/10/2020 21:24:21", "content": "history_positions is very helpful. plz, update it with value >0. Can you update it with 5 or 10?
", "rawMarkdown": "history_positions is very helpful. plz, update it with value >0. Can you update it with 5 or 10?", "votes": null }, { "id": "1042240", "postDate": "10/08/2020 06:24:39", "content": "@philculliton is TFRecords for full external training dataset 70+GB ? or the training dataset in this competition 22GB
", "rawMarkdown": "philculliton is TFRecords for full external training dataset 70+GB ? or the training dataset in this competition 22GB", "votes": null }, { "id": "1043007", "postDate": "10/08/2020 16:01:23", "content": "@seshurajup You can review the TFRecords here. It contains both the training and validation sets.
", "rawMarkdown": "seshurajup You can review the [TFRecords here](https://www.kaggle.com/ryanholbrook/lyft-motion-prediction-tfrecords). It contains both the training and validation sets.", "votes": null }, { "id": "1043029", "postDate": "10/08/2020 16:26:00", "content": "@juliaelliott is these tfrecords generated from
\nhttps://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/data ( 22 GB ) -- Competition Data Only
\nor
\nhttps://www.kaggle.com/philculliton/lyft-full-training-set ( 71.7 GB ) -- Full training Set also part of these tfrecords ?
Never used TPUs. Is it possible to use TPUs with pytorch?
", "rawMarkdown": "Never used TPUs. Is it possible to use TPUs with pytorch?", "votes": null }, { "id": "1050776", "postDate": "10/15/2020 18:00:45", "content": "pytroch xla (https://github.com/pytorch/xla) it is possible to use
", "rawMarkdown": "pytroch xla (https://github.com/pytorch/xla) it is possible to use", "votes": null } ], "comments": [ { "id": 998445, "author_name": "jpbremer", "author_url": "", "post_date": "09/04/2020 18:21:16", "content": "The notebooks are unavailable @philculliton
", "votes": null, "replies": [] }, { "id": 1001174, "author_name": "hardworkingkaggler", "author_url": "", "post_date": "09/07/2020 05:45:45", "content": "The 2nd and 3rd links of \"Getting Started!\" are unavailable. Please fix them, thanks!
", "votes": null, "replies": [] }, { "id": 1003400, "author_name": "srinivasgopalkrishna", "author_url": "", "post_date": "09/08/2020 23:41:31", "content": "Can the data be updates with history_positions not equal to zero
The links should now work - sorry about that!
", "votes": null, "replies": [] }, { "id": 1005907, "author_name": "doanquanvietnamca", "author_url": "", "post_date": "09/10/2020 21:24:21", "content": "history_positions is very helpful. plz, update it with value >0. Can you update it with 5 or 10?
", "votes": null, "replies": [] }, { "id": 1042240, "author_name": "seshurajup", "author_url": "", "post_date": "10/08/2020 06:24:39", "content": "@philculliton is TFRecords for full external training dataset 70+GB ? or the training dataset in this competition 22GB
", "votes": null, "replies": [ { "id": 1043007, "author_name": "juliaelliott", "author_url": "", "post_date": "10/08/2020 16:01:23", "content": "@seshurajup You can review the TFRecords here. It contains both the training and validation sets.
", "votes": null, "replies": [] }, { "id": 1043029, "author_name": "seshurajup", "author_url": "", "post_date": "10/08/2020 16:26:00", "content": "@juliaelliott is these tfrecords generated from
\nhttps://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/data ( 22 GB ) -- Competition Data Only
\nor
\nhttps://www.kaggle.com/philculliton/lyft-full-training-set ( 71.7 GB ) -- Full training Set also part of these tfrecords ?
Never used TPUs. Is it possible to use TPUs with pytorch?
", "votes": null, "replies": [ { "id": 1050776, "author_name": "seshurajup", "author_url": "", "post_date": "10/15/2020 18:00:45", "content": "pytroch xla (https://github.com/pytorch/xla) it is possible to use
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "996829": "###Are you a TPU Star?\nFor this competition, we are offering multiple opportunities to earn **TPU quota and/or cash prizes** for making valued contributions to our growing TPU developer community!\n\n**What can I win?**\nEarn up to 80 hours of **additional TPU quota**, awarded throughout the competition\nWin one of **3 \"TPU Star\" cash prizes of $5,000** at the close of the competition by being one of the most knowledgeable and helpful TPU experts in the community.\n\n**How can I enter?**\n[Submit a qualified notebook here](https://www.kaggle.com/tpu-prize) anytime during the competition to be considered for either prizes.\n\n**Extra quota earners** will be evaluated and awarded throughout the competition. For more details on how quota earners are evaluated, [review the submission page](https://www.kaggle.com/tpu-prize).\n\n**TPU stars** will be evaluated at the close of the competition. \n- Experts from the Google Cloud TPU team will identify TPU stars on the following criteria:\n - Forum and notebook discussion/comment contributions\n - Quality of code samples in public notebooks and/or forums\n - Thoughtful analysis and explainability of code content\n- At a minimum, you must have shared a public notebook which uses Kaggle's TPU integration on this competition's dataset. We will consider both self-nominations through the above link and Kaggle-identified upvoted/active contributors\n- The deadline for submissions is **December 13, 2020**.\n \n**Getting Started!**\nWe’ve provided the dataset converted to TFRecords and the associated scripts for creating the files and loading the data:\n- The training & validation data converted to TFRecords format are here: [https://www.kaggle.com/ryanholbrook/lyft-motion-prediction-tfrecords](https://www.kaggle.com/ryanholbrook/lyft-motion-prediction-tfrecords)\n- A short notebook for loading the TFRecords is here: [https://www.kaggle.com/ryanholbrook/loading-the-lyft-tfrecords](https://www.kaggle.com/ryanholbrook/loading-the-lyft-tfrecords)\n- The script that's creating the TFRecord files is here: [https://www.kaggle.com/ryanholbrook/writing-lyft-tfrecords](https://www.kaggle.com/ryanholbrook/writing-lyft-tfrecords)", "998445": "The notebooks are unavailable @philculliton", "1001174": "The 2nd and 3rd links of \"Getting Started!\" are unavailable. Please fix them, thanks!", "1003400": "Can the data be updates with `history_positions` not equal to zero", "1003480": "The links should now work - sorry about that!", "1005907": "history_positions is very helpful. plz, update it with value >0. Can you update it with 5 or 10?", "1042240": "philculliton is TFRecords for full external training dataset 70+GB ? or the training dataset in this competition 22GB", "1043007": "seshurajup You can review the [TFRecords here](https://www.kaggle.com/ryanholbrook/lyft-motion-prediction-tfrecords). It contains both the training and validation sets.", "1043029": "juliaelliott is these tfrecords generated from \nhttps://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/data ( 22 GB ) -- Competition Data Only\nor \nhttps://www.kaggle.com/philculliton/lyft-full-training-set ( 71.7 GB ) -- Full training Set also part of these tfrecords ?", "1050718": "Never used TPUs. Is it possible to use TPUs with pytorch?", "1050776": "pytroch xla (https://github.com/pytorch/xla) it is possible to use" }, "source": "meta" }