Hey, I am trying to run the baseline shared by Peter with GPU. But it seems the GPU is not working in it, I found the value of torch.cuda.is_available() to be False while my GPU quota is going down consistently.
I tried searching for this issue and found this one, according to it, custom packages are not supported by GPU in Kaggle kernel or we can say not all libraries are going to use GPU directly. Any pointers on how to make GPU run?
\nI am using the kaggle_l5kit by adding it through Add Data->Kernel Output Files, did anyone face the same issue?
Update:
\nHey, I am trying to run the baseline shared by Peter with GPU. But it seems the GPU is not working in it, I found the value of torch.cuda.is_available() to be False while my GPU quota is going down consistently.
I tried searching for this issue and found this one, according to it, custom packages are not supported by GPU in Kaggle kernel or we can say not all libraries are going to use GPU directly. Any pointers on how to make GPU run?
\nI am using the kaggle_l5kit by adding it through Add Data->Kernel Output Files, did anyone face the same issue?
Update:
\nI've just published my temporary solution, details here
", "rawMarkdown": "I've just published my temporary solution, [details here](https://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/discussion/177549)", "votes": 1 }, { "id": 986224, "postDate": "2020-08-26T09:52:20.133Z", "content": "I guess I posted a well known issue, I can't delete the topic now even if I want to, sorry for the trouble guys!
", "rawMarkdown": "I guess I posted a well known issue, I can't delete the topic now even if I want to, sorry for the trouble guys!", "votes": 1 }, { "id": 2259004, "postDate": "2023-05-14T16:02:46.557Z", "content": "Hello! I have a similar problem with you.However,the value oftorch.cuda.is_available()is true.the unilization ratio of CPU is quite high,while the unilization ratio of GPU is 0.Could you help me?
@sarques most of us are facing this issue with GPU and browser freez. At least I did till last night (EST). Yet to test out today. Follow this discussion page for more info. https://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/discussion/177125
\nAnother info/help for you is that. The l5kit must be added as utility script and not as data to kernal output files.
", "rawMarkdown": "@sarques most of us are facing this issue with GPU and browser freez. At least I did till last night (EST). Yet to test out today. Follow this discussion page for more info. https://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/discussion/177125 \n\nAnother info/help for you is that. The l5kit must be added as utility script and not as data to kernal output files. ", "replies": [ { "id": 986218, "postDate": "2020-08-26T09:48:33.980Z", "content": "Thanks Deepak.
", "rawMarkdown": "Thanks Deepak." } ] }, { "id": 986212, "postDate": "2020-08-26T09:43:52.567Z", "content": "You can find more details here
\nKaggle is working on a fix.
", "rawMarkdown": "You can find more details [here](https://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/discussion/177125)\n\nKaggle is working on a fix.", "replies": [ { "id": 986217, "postDate": "2020-08-26T09:48:14.457Z", "content": "Sure, thanks Peter! :)
", "rawMarkdown": "Sure, thanks Peter! :)" } ] }, { "id": 986219, "postDate": "2020-08-26T09:49:41.803Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 986232, "author_name": "Peter", "author_url": "", "post_date": "2020-08-26T09:58:50.380000", "content": "I've just published my temporary solution, details here
", "votes": 1, "replies": [] }, { "id": 986224, "author_name": "Gajendra Saraswat", "author_url": "", "post_date": "2020-08-26T09:52:20.133000", "content": "I guess I posted a well known issue, I can't delete the topic now even if I want to, sorry for the trouble guys!
", "votes": 1, "replies": [] }, { "id": 2259004, "author_name": "Ross.sin", "author_url": "", "post_date": "2023-05-14T16:02:46.557000", "content": "Hello! I have a similar problem with you.However,the value oftorch.cuda.is_available()is true.the unilization ratio of CPU is quite high,while the unilization ratio of GPU is 0.Could you help me?
@sarques most of us are facing this issue with GPU and browser freez. At least I did till last night (EST). Yet to test out today. Follow this discussion page for more info. https://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/discussion/177125
\nAnother info/help for you is that. The l5kit must be added as utility script and not as data to kernal output files.
", "votes": 0, "replies": [ { "id": 986218, "author_name": "Gajendra Saraswat", "author_url": "", "post_date": "2020-08-26T09:48:33.980000", "content": "Thanks Deepak.
", "votes": 0, "replies": [] } ] }, { "id": 986212, "author_name": "Peter", "author_url": "", "post_date": "2020-08-26T09:43:52.567000", "content": "You can find more details here
\nKaggle is working on a fix.
", "votes": 0, "replies": [ { "id": 986217, "author_name": "Gajendra Saraswat", "author_url": "", "post_date": "2020-08-26T09:48:14.457000", "content": "Sure, thanks Peter! :)
", "votes": 0, "replies": [] } ] }, { "id": 986219, "author_name": "", "author_url": "", "post_date": "2020-08-26T09:49:41.803000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "986196": "Hey, I am trying to run the baseline shared by Peter with GPU. But it seems the GPU is not working in it, I found the value of `torch.cuda.is_available()` to be `False` while my GPU quota is going down consistently.\n\nI tried searching for this issue and found this [one](https://www.kaggle.com/general/74529?rvi=1), according to it, custom packages are not supported by GPU in Kaggle kernel or we can say not all libraries are going to use GPU directly. Any pointers on how to make GPU run?\n\nI am using the `kaggle_l5kit` by adding it through `Add Data->Kernel Output Files`, did anyone face the same issue?\n\n****\n\nUpdate: \n\n- The solution to this problem can be found [here](https://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/discussion/177549)\n- I guess I posted a well known issue, I can't delete the topic now even if I want to, sorry for the trouble and please be kind enough to not downvote! :/", "986232": "I've just published my temporary solution, [details here](https://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/discussion/177549)", "986224": "I guess I posted a well known issue, I can't delete the topic now even if I want to, sorry for the trouble guys!", "2259004": "Hello! I have a similar problem with you.However,the value of`torch.cuda.is_available() `is true.the unilization ratio of CPU is quite high,while the unilization ratio of GPU is 0.Could you help me?", "986214": "@sarques most of us are facing this issue with GPU and browser freez. At least I did till last night (EST). Yet to test out today. Follow this discussion page for more info. https://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/discussion/177125 \n\nAnother info/help for you is that. The l5kit must be added as utility script and not as data to kernal output files. ", "986212": "You can find more details [here](https://www.kaggle.com/c/lyft-motion-prediction-autonomous-vehicles/discussion/177125)\n\nKaggle is working on a fix.", "986219": "" } }