{ "id": 436629, "title": "some youtube video on the competition you should not missed!", "url": "/competitions/predict-ai-model-runtime/discussion/436629", "author_name": "", "post_date": "2023-09-03T09:52:14.358000200Z", "votes": 45, "comment_count": 7, "views": 0, "content": "

\"\"

\n

the picture looks familiar!!!

\n

\"\"

\n

https://mangpo.net/talks/2021-xla-autotuning-pact.pdf
\nhttps://arxiv.org/pdf/2008.01040.pdf

\n

\"\"

\n

\"\"

\n

GO is the GO from alphaGO

", "messages": [ { "id": "2421309", "postDate": "09/03/2023 09:52:14", "content": "

\"\"

\n

the picture looks familiar!!!

\n

\"\"

\n

https://mangpo.net/talks/2021-xla-autotuning-pact.pdf
\nhttps://arxiv.org/pdf/2008.01040.pdf

\n

\"\"

\n

\"\"

\n

GO is the GO from alphaGO

", "rawMarkdown": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F391a5d7b600e1c5f06e40241fc8546a6%2FSelection_999(3027).png?generation=1693734714671636&alt=media)\n\nthe picture looks familiar!!!\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2Fe1f134892b2af3d5e7e6be5091922be7%2FSelection_999(3031).png?generation=1693736272473162&alt=media)\n\n\nhttps://mangpo.net/talks/2021-xla-autotuning-pact.pdf\nhttps://arxiv.org/pdf/2008.01040.pdf\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F0b2a2fd70a5fa4f5c9056c0c8f8e65de%2FSelection_999(3029).png?generation=1693735339715297&alt=media)\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2Fe6ef2d1f96e587231d5a6c75f76bf8d6%2FSelection_999(3030).png?generation=1693735932225537&alt=media)\n\nGO is the GO from alphaGO", "votes": null }, { "id": "2421315", "postDate": "09/03/2023 09:53:51", "content": "

the google search keyword are:
\nHLO fusion, XLA compiler if you want to learn more

\n

\"\"

\n

actually you can dump HLO output from kaggle kernel i think …. but i am not sure if that is useful at all

", "rawMarkdown": "the google search keyword are:\nHLO fusion, XLA compiler if you want to learn more\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2Fc6194fa5d2df1a69eb088604e81c8a64%2FSelection_999(3026).png?generation=1693734791256727&alt=media)\n\nactually you can dump HLO output from kaggle kernel i think .... but i am not sure if that is useful at all", "votes": null }, { "id": "2421352", "postDate": "09/03/2023 10:20:16", "content": "

\"\"

\n

i suppose one can find more information on NeurIPS as well …

", "rawMarkdown": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F2f77e52a7cf559ea352228b573418613%2FSelection_999(3032).png?generation=1693736380472029&alt=media)\n\ni suppose one can find more information on NeurIPS as well ...", "votes": null }, { "id": "2421490", "postDate": "09/03/2023 11:57:18", "content": "

@hengck23 thanks for sharing this video

", "rawMarkdown": "hengck23 thanks for sharing this video", "votes": null }, { "id": "2422004", "postDate": "09/03/2023 16:46:49", "content": "

i was wondering how they train large graph ….

\n

\"\"

\n

so they are using segment graph + drop graph

\n

[1] Learning Large Graph Property Prediction via Graph Segment Training - Kaidi Cao, Phitchaya Mangpo Phothilimthana
\nhttps://openreview.net/pdf?id=Emy5KJ4w6J

", "rawMarkdown": "i was wondering how they train large graph ....\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F685327b178ebf7696263f11167e8534d%2FSelection_999(3035).png?generation=1693759585971686&alt=media)\n\nso they are using segment graph + drop graph\n\n[1] Learning Large Graph Property Prediction via Graph Segment Training - Kaidi Cao, Phitchaya Mangpo Phothilimthana\nhttps://openreview.net/pdf?id=Emy5KJ4w6J", "votes": null }, { "id": "2422104", "postDate": "09/03/2023 17:57:48", "content": "

This paper has a bit more details about XLA layout assignment and tiling: https://mangpo.net/papers/xla-autotuning-pact2021.pdf. The dataset is essentially generated from the autotuner described in the paper.

", "rawMarkdown": "This paper has a bit more details about XLA layout assignment and tiling: https://mangpo.net/papers/xla-autotuning-pact2021.pdf. The dataset is essentially generated from the autotuner described in the paper.", "votes": null }, { "id": "2475401", "postDate": "10/09/2023 21:41:36", "content": "

Another talk we just presented recently about the baseline GNN models: https://www.youtube.com/watch?v=VASg2XNgj-4

", "rawMarkdown": "Another talk we just presented recently about the baseline GNN models: https://www.youtube.com/watch?v=VASg2XNgj-4", "votes": null }, { "id": "2511685", "postDate": "11/03/2023 20:28:40", "content": "

Actual link to the first video: https://www.youtube.com/watch?v=vYQ37pBNAFI&ab_channel=CSITNaresuanUniversity
\nAnd second video: https://www.youtube.com/watch?v=esD_zvAf49I&ab_channel=uwsampl

", "rawMarkdown": "Actual link to the first video: https://www.youtube.com/watch?v=vYQ37pBNAFI&ab_channel=CSITNaresuanUniversity \nAnd second video: https://www.youtube.com/watch?v=esD_zvAf49I&ab_channel=uwsampl", "votes": null } ], "comments": [ { "id": 2421315, "author_name": "hengck23", "author_url": "", "post_date": "09/03/2023 09:53:51", "content": "

the google search keyword are:
\nHLO fusion, XLA compiler if you want to learn more

\n

\"\"

\n

actually you can dump HLO output from kaggle kernel i think …. but i am not sure if that is useful at all

", "votes": null, "replies": [] }, { "id": 2421352, "author_name": "hengck23", "author_url": "", "post_date": "09/03/2023 10:20:16", "content": "

\"\"

\n

i suppose one can find more information on NeurIPS as well …

", "votes": null, "replies": [] }, { "id": 2421490, "author_name": "faysalmiah1721758", "author_url": "", "post_date": "09/03/2023 11:57:18", "content": "

@hengck23 thanks for sharing this video

", "votes": null, "replies": [] }, { "id": 2422004, "author_name": "hengck23", "author_url": "", "post_date": "09/03/2023 16:46:49", "content": "

i was wondering how they train large graph ….

\n

\"\"

\n

so they are using segment graph + drop graph

\n

[1] Learning Large Graph Property Prediction via Graph Segment Training - Kaidi Cao, Phitchaya Mangpo Phothilimthana
\nhttps://openreview.net/pdf?id=Emy5KJ4w6J

", "votes": null, "replies": [] }, { "id": 2422104, "author_name": "mangpophothilimthana", "author_url": "", "post_date": "09/03/2023 17:57:48", "content": "

This paper has a bit more details about XLA layout assignment and tiling: https://mangpo.net/papers/xla-autotuning-pact2021.pdf. The dataset is essentially generated from the autotuner described in the paper.

", "votes": null, "replies": [] }, { "id": 2475401, "author_name": "mangpophothilimthana", "author_url": "", "post_date": "10/09/2023 21:41:36", "content": "

Another talk we just presented recently about the baseline GNN models: https://www.youtube.com/watch?v=VASg2XNgj-4

", "votes": null, "replies": [] }, { "id": 2511685, "author_name": "lewington", "author_url": "", "post_date": "11/03/2023 20:28:40", "content": "

Actual link to the first video: https://www.youtube.com/watch?v=vYQ37pBNAFI&ab_channel=CSITNaresuanUniversity
\nAnd second video: https://www.youtube.com/watch?v=esD_zvAf49I&ab_channel=uwsampl

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2421309": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F391a5d7b600e1c5f06e40241fc8546a6%2FSelection_999(3027).png?generation=1693734714671636&alt=media)\n\nthe picture looks familiar!!!\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2Fe1f134892b2af3d5e7e6be5091922be7%2FSelection_999(3031).png?generation=1693736272473162&alt=media)\n\n\nhttps://mangpo.net/talks/2021-xla-autotuning-pact.pdf\nhttps://arxiv.org/pdf/2008.01040.pdf\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F0b2a2fd70a5fa4f5c9056c0c8f8e65de%2FSelection_999(3029).png?generation=1693735339715297&alt=media)\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2Fe6ef2d1f96e587231d5a6c75f76bf8d6%2FSelection_999(3030).png?generation=1693735932225537&alt=media)\n\nGO is the GO from alphaGO", "2421315": "the google search keyword are:\nHLO fusion, XLA compiler if you want to learn more\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2Fc6194fa5d2df1a69eb088604e81c8a64%2FSelection_999(3026).png?generation=1693734791256727&alt=media)\n\nactually you can dump HLO output from kaggle kernel i think .... but i am not sure if that is useful at all", "2421352": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F2f77e52a7cf559ea352228b573418613%2FSelection_999(3032).png?generation=1693736380472029&alt=media)\n\ni suppose one can find more information on NeurIPS as well ...", "2421490": "hengck23 thanks for sharing this video", "2422004": "i was wondering how they train large graph ....\n\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F113660%2F685327b178ebf7696263f11167e8534d%2FSelection_999(3035).png?generation=1693759585971686&alt=media)\n\nso they are using segment graph + drop graph\n\n[1] Learning Large Graph Property Prediction via Graph Segment Training - Kaidi Cao, Phitchaya Mangpo Phothilimthana\nhttps://openreview.net/pdf?id=Emy5KJ4w6J", "2422104": "This paper has a bit more details about XLA layout assignment and tiling: https://mangpo.net/papers/xla-autotuning-pact2021.pdf. The dataset is essentially generated from the autotuner described in the paper.", "2475401": "Another talk we just presented recently about the baseline GNN models: https://www.youtube.com/watch?v=VASg2XNgj-4", "2511685": "Actual link to the first video: https://www.youtube.com/watch?v=vYQ37pBNAFI&ab_channel=CSITNaresuanUniversity \nAnd second video: https://www.youtube.com/watch?v=esD_zvAf49I&ab_channel=uwsampl" }, "source": "meta" }