After I commit my kernel, it always died without saying any error or warning. Anyone faced with the same situation? What could be the possible mistake?\nI would appreciate it if anyone who can help me. Thanks.
", "messages": [ { "id": "571271", "postDate": "07/09/2019 12:30:34", "content": "After I commit my kernel, it always died without saying any error or warning. Anyone faced with the same situation? What could be the possible mistake?\nI would appreciate it if anyone who can help me. Thanks.
", "rawMarkdown": "After I commit my kernel, it always died without saying any error or warning. Anyone faced with the same situation? What could be the possible mistake?\nI would appreciate it if anyone who can help me. Thanks.", "votes": null }, { "id": "571273", "postDate": "07/09/2019 12:35:39", "content": "That is most probably because you ran out of RAM. Happens to me often too :(
", "rawMarkdown": "That is most probably because you ran out of RAM. Happens to me often too :(", "votes": null }, { "id": "571388", "postDate": "07/09/2019 14:52:41", "content": "It's definitely Memory Limit Error.\nTry to reduce your inputs, complexity of the model and delete data you won't need anymore from memory.
", "rawMarkdown": "It's definitely Memory Limit Error.\nTry to reduce your inputs, complexity of the model and delete data you won't need anymore from memory.", "votes": null }, { "id": "571423", "postDate": "07/09/2019 15:24:25", "content": "At the start of the competition it is all about trying things out - try using only 50% or even 25% of the training molecules - this will speed up the time it takes to try things out - you should be able to figure out what works and what doesn't - you can then worry about running your models on the entire dataset - worse comes to worse team up with someone with a bigger box in the last month.
", "rawMarkdown": "At the start of the competition it is all about trying things out - try using only 50% or even 25% of the training molecules - this will speed up the time it takes to try things out - you should be able to figure out what works and what doesn't - you can then worry about running your models on the entire dataset - worse comes to worse team up with someone with a bigger box in the last month.", "votes": null }, { "id": "571469", "postDate": "07/09/2019 16:11:45", "content": "I run most of my scripts on local PC in Jupyter kernels. Also run lots in Microsoft Visual Studio Code.
\n\n99% of the time when a Jupyter kernel dies with no error message I can run it in VS Code and find a memory error message. So my bet is 99 to 1 odds that your hitting a memory issue.
", "rawMarkdown": "I run most of my scripts on local PC in Jupyter kernels. Also run lots in Microsoft Visual Studio Code.\n\n99% of the time when a Jupyter kernel dies with no error message I can run it in VS Code and find a memory error message. So my bet is 99 to 1 odds that your hitting a memory issue.", "votes": null }, { "id": "571500", "postDate": "07/09/2019 17:14:04", "content": "Thanks for all you guys who give me advice, I have figured out the problem. \nBased on your's comments, I found that in my occasion is because of run out of the disk(I want to save my generated features in order to try more models), but the csv of generated features is tooooooo big, and the kernel only allowed 4GB for the disk, and this is the reason.\nThanks for those who commented in this topic and gave me ideas which helped me solve the problem. I have up voted all of your comments.
", "rawMarkdown": "Thanks for all you guys who give me advice, I have figured out the problem. \nBased on your's comments, I found that in my occasion is because of run out of the disk(I want to save my generated features in order to try more models), but the csv of generated features is tooooooo big, and the kernel only allowed 4GB for the disk, and this is the reason.\nThanks for those who commented in this topic and gave me ideas which helped me solve the problem. I have up voted all of your comments.", "votes": null }, { "id": "571513", "postDate": "07/09/2019 17:32:27", "content": "I now use df.to_pickle() , it's much better (faster, lighter) than csv
", "rawMarkdown": "I now use df.to_pickle() , it's much better (faster, lighter) than csv", "votes": null }, { "id": "571518", "postDate": "07/09/2019 17:35:51", "content": "@leonshangguan - would you be so kind as to raise this in the product feedback forum\nAt present the feedback sucks for edge cases such as out of memory or disk space.
", "rawMarkdown": "leonshangguan - would you be so kind as to raise this in the product feedback forum\nAt present the feedback sucks for edge cases such as out of memory or disk space.", "votes": null }, { "id": "571521", "postDate": "07/09/2019 17:36:44", "content": "Thanks for your advice, I will try to use it.
", "rawMarkdown": "Thanks for your advice, I will try to use it.", "votes": null }, { "id": "571529", "postDate": "07/09/2019 17:42:16", "content": "Of course, but I don't know where is the 'product feedback forum'? What should I do?
", "rawMarkdown": "Of course, but I don't know where is the 'product feedback forum'? What should I do?", "votes": null }, { "id": "571555", "postDate": "07/09/2019 18:17:31", "content": "@leonshangguan - it is in the global discussions link\nAs an added bonus have a look at https://towardsdatascience.com/the-best-format-to-save-pandas-data-414dca023e0d\nThis will give you a good handle on different ways of storing large data - enjoy!
", "rawMarkdown": "leonshangguan - it is in the global discussions link\nAs an added bonus have a look at https://towardsdatascience.com/the-best-format-to-save-pandas-data-414dca023e0d\nThis will give you a good handle on different ways of storing large data - enjoy!", "votes": null } ], "comments": [ { "id": 571273, "author_name": "somnambwl", "author_url": "", "post_date": "07/09/2019 12:35:39", "content": "That is most probably because you ran out of RAM. Happens to me often too :(
", "votes": null, "replies": [] }, { "id": 571388, "author_name": "stanislavblinov", "author_url": "", "post_date": "07/09/2019 14:52:41", "content": "It's definitely Memory Limit Error.\nTry to reduce your inputs, complexity of the model and delete data you won't need anymore from memory.
", "votes": null, "replies": [] }, { "id": 571423, "author_name": "scirpus", "author_url": "", "post_date": "07/09/2019 15:24:25", "content": "At the start of the competition it is all about trying things out - try using only 50% or even 25% of the training molecules - this will speed up the time it takes to try things out - you should be able to figure out what works and what doesn't - you can then worry about running your models on the entire dataset - worse comes to worse team up with someone with a bigger box in the last month.
", "votes": null, "replies": [] }, { "id": 571469, "author_name": "pcjimmmy", "author_url": "", "post_date": "07/09/2019 16:11:45", "content": "I run most of my scripts on local PC in Jupyter kernels. Also run lots in Microsoft Visual Studio Code.
\n\n99% of the time when a Jupyter kernel dies with no error message I can run it in VS Code and find a memory error message. So my bet is 99 to 1 odds that your hitting a memory issue.
", "votes": null, "replies": [] }, { "id": 571500, "author_name": "leonshangguan", "author_url": "", "post_date": "07/09/2019 17:14:04", "content": "Thanks for all you guys who give me advice, I have figured out the problem. \nBased on your's comments, I found that in my occasion is because of run out of the disk(I want to save my generated features in order to try more models), but the csv of generated features is tooooooo big, and the kernel only allowed 4GB for the disk, and this is the reason.\nThanks for those who commented in this topic and gave me ideas which helped me solve the problem. I have up voted all of your comments.
", "votes": null, "replies": [ { "id": 571513, "author_name": "jmtest", "author_url": "", "post_date": "07/09/2019 17:32:27", "content": "I now use df.to_pickle() , it's much better (faster, lighter) than csv
", "votes": null, "replies": [] }, { "id": 571518, "author_name": "scirpus", "author_url": "", "post_date": "07/09/2019 17:35:51", "content": "@leonshangguan - would you be so kind as to raise this in the product feedback forum\nAt present the feedback sucks for edge cases such as out of memory or disk space.
", "votes": null, "replies": [] }, { "id": 571521, "author_name": "leonshangguan", "author_url": "", "post_date": "07/09/2019 17:36:44", "content": "Thanks for your advice, I will try to use it.
", "votes": null, "replies": [] }, { "id": 571529, "author_name": "leonshangguan", "author_url": "", "post_date": "07/09/2019 17:42:16", "content": "Of course, but I don't know where is the 'product feedback forum'? What should I do?
", "votes": null, "replies": [] }, { "id": 571555, "author_name": "scirpus", "author_url": "", "post_date": "07/09/2019 18:17:31", "content": "@leonshangguan - it is in the global discussions link\nAs an added bonus have a look at https://towardsdatascience.com/the-best-format-to-save-pandas-data-414dca023e0d\nThis will give you a good handle on different ways of storing large data - enjoy!
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "571271": "After I commit my kernel, it always died without saying any error or warning. Anyone faced with the same situation? What could be the possible mistake?\nI would appreciate it if anyone who can help me. Thanks.", "571273": "That is most probably because you ran out of RAM. Happens to me often too :(", "571388": "It's definitely Memory Limit Error.\nTry to reduce your inputs, complexity of the model and delete data you won't need anymore from memory.", "571423": "At the start of the competition it is all about trying things out - try using only 50% or even 25% of the training molecules - this will speed up the time it takes to try things out - you should be able to figure out what works and what doesn't - you can then worry about running your models on the entire dataset - worse comes to worse team up with someone with a bigger box in the last month.", "571469": "I run most of my scripts on local PC in Jupyter kernels. Also run lots in Microsoft Visual Studio Code.\n\n99% of the time when a Jupyter kernel dies with no error message I can run it in VS Code and find a memory error message. So my bet is 99 to 1 odds that your hitting a memory issue.", "571500": "Thanks for all you guys who give me advice, I have figured out the problem. \nBased on your's comments, I found that in my occasion is because of run out of the disk(I want to save my generated features in order to try more models), but the csv of generated features is tooooooo big, and the kernel only allowed 4GB for the disk, and this is the reason.\nThanks for those who commented in this topic and gave me ideas which helped me solve the problem. I have up voted all of your comments.", "571513": "I now use df.to_pickle() , it's much better (faster, lighter) than csv", "571518": "leonshangguan - would you be so kind as to raise this in the product feedback forum\nAt present the feedback sucks for edge cases such as out of memory or disk space.", "571521": "Thanks for your advice, I will try to use it.", "571529": "Of course, but I don't know where is the 'product feedback forum'? What should I do?", "571555": "leonshangguan - it is in the global discussions link\nAs an added bonus have a look at https://towardsdatascience.com/the-best-format-to-save-pandas-data-414dca023e0d\nThis will give you a good handle on different ways of storing large data - enjoy!" }, "source": "meta" }