Using the link in the documentation gives me a virtual machine without the fastai directory in jupyter notebook.
fastai directory missing:
I created a vm with the template from the Azure DataScienceVM directly: https://github.com/Azure/DataScienceVM/tree/master/Extensions/fastaiv1 which works as expected.
Hi adi93, I was able to follow these steps. But when try to open fastai kernel, I get an error message as Permission denied: Untitled.ipynb
You should try the template in my link:
Everything is configured for you!
@twaldecker, Thanks for your reply. I am using ubuntu 16.04, so I suppose I need an alternative to fix it.
Successfully installed fastai 0.7, but received error when I was trying to import fastai.vision . I am using pip for installations. From troubleshooting information, I realised to upgrade fastai to version 1.0 - still in the process of finding a best solution.
Meanwhile I created a virtual env as suggested in this post, updated python 3.7, pip, torch, etc. But facing lot of issues with it. Any help to find an optimal working solution to install fastai 1.0, will be much appreciated!
Can anyone help me achieve the same result as installing fastai extension without actually using the fastai template with Azure Windows DSVM? The problem is that I can’t use custom template or have access to Azure console.
I can replicate things like cmd.exe conda installs and git commands in bash or cmd but I’m not sure how to do multi-line commands like fixKernelSpec function. If I try to run ps1 file within DSVM I get warning that the file is not signed.
DSVM not provisioning VM correctly. It seems like the DSVM instructions are no longer working. When I use this link http://aka.ms/dsvm/fastai, by default I see this option for the VM size
When I try with this default option I got this error
The requested VM size Standard_NC6_Promo is not available in the current region. The sizes available in the current region are: Standard_B1ls,Standard_B1ms,Standard_B1s,Standard_B2ms,Standard_B2s,Standard_B4ms,Standard_B8ms,Standard_B12ms,Standard_B16ms,Standard_B20ms,Standard_DS1_v2,Standard_DS2_v2,Standard_DS3_v2,Standard_DS4_v2,Standard_DS5_v2,Standard_DS11-1_v2,Standard_DS11_v2,Standard_DS12-1_v2,Standard_DS12-2_v2,Standard_DS12_v2,Standard_DS13-2_v2,Standard_DS13-4_v2,Standard_DS13_v2,Standard_DS14-4_v2,Standard_DS14-8_v2,Standard_DS14_v2,Standard_DS15_v2,Standard_DS2_v2_Promo,Standard_DS3_v2_Promo,Standard_DS4_v2_Promo,Standard_DS5_v2_Promo,Standard_DS11_v2_Promo,Standard_DS12_v2_Promo,Standard_DS13_v2_Promo,Standard_DS14_v2_Promo,Standard_F1s,Standard_F2s,Standard_F4s,Standard_F8s,Standard_F16s,Standard_D2s_v3,Standard_D4s_v3,Standard_D8s_v3,Standard_D16s_v3,Standard_D32s_v3,Standard_A0,Standard_A1,Standard_A2,Standard_A3,Standard_A5,Standard_A4,Standard_A6,Standard_A7,Basic_A0,Basic_A1,Basic_A2,Basic_A3,Basic_A4,Standard_D1_v2,Standard_D2_v2,Standard_D3_v2,Standard_D4_v2,Standard_D5_v2,Standard_D11_v2,Standard_D12_v2,Standard_D13_v2,Standard_D14_v2,Standard_D15_v2,Standard_D2_v2_Promo,Standard_D3_v2_Promo,Standard_D4_v2_Promo,Standard_D5_v2_Promo,Standard_D11_v2_Promo,Standard_D12_v2_Promo,Standard_D13_v2_Promo,Standard_D14_v2_Promo,Standard_F1,Standard_F2,Standard_F4,Standard_F8,Standard_F16,Standard_A1_v2,Standard_A2m_v2,Standard_A2_v2,Standard_A4m_v2,Standard_A4_v2,Standard_A8m_v2,Standard_A8_v2,Standard_D2_v3,Standard_D4_v3,Standard_D8_v3,Standard_D16_v3,Standard_D32_v3,Standard_D48_v3,Standard_D64_v3,Standard_D48s_v3,Standard_D64s_v3,Standard_E2_v3,Standard_E4_v3,Standard_E8_v3,Standard_E16_v3,Standard_E20_v3,Standard_E32_v3,Standard_E48_v3,Standard_E64i_v3,Standard_E64_v3,Standard_E2s_v3,Standard_E4-2s_v3,Standard_E4s_v3,Standard_E8-2s_v3,Standard_E8-4s_v3,Standard_E8s_v3,Standard_E16-4s_v3,Standard_E16-8s_v3,Standard_E16s_v3,Standard_E20s_v3,Standard_E32-8s_v3,Standard_E32-16s_v3,Standard_E32s_v3,Standard_E48s_v3,Standard_E64-16s_v3,Standard_E64-32s_v3,Standard_E64is_v3,Standard_E64s_v3,Standard_D1,Standard_D2,Standard_D3,Standard_D4,Standard_D11,Standard_D12,Standard_D13,Standard_D14,Standard_DS1,Standard_DS2,Standard_DS3,Standard_DS4,Standard_DS11,Standard_DS12,Standard_DS13,Standard_DS14,Standard_L8s_v2,Standard_L16s_v2,Standard_L32s_v2,Standard_L48s_v2,Standard_L64s_v2,Standard_L80s_v2,Standard_G1,Standard_G2,Standard_G3,Standard_G4,Standard_G5,Standard_GS1,Standard_GS2,Standard_GS3,Standard_GS4,Standard_GS4-4,Standard_GS4-8,Standard_GS5,Standard_GS5-8,Standard_GS5-16,Standard_L4s,Standard_L8s,Standard_L16s,Standard_L32s,Standard_F2s_v2,Standard_F4s_v2,Standard_F8s_v2,Standard_F16s_v2,Standard_F32s_v2,Standard_F48s_v2,Standard_F64s_v2,Standard_F72s_v2,Standard_A8,Standard_A9,Standard_A10,Standard_A11,Standard_H8,Standard_H8_Promo,Standard_H16,Standard_H16_Promo,Standard_H8m,Standard_H8m_Promo,Standard_H16m,Standard_H16m_Promo,Standard_H16r,Standard_H16r_Promo,Standard_H16mr,Standard_H16mr_Promo,Standard_NV6s_v2,Standard_NV12s_v2,Standard_NV24s_v2,Standard_NV12s_v3,Standard_NV24s_v3,Standard_NV48s_v3. Find out more on the available VM sizes in each region at https://aka.ms/azure-regions.
I got this same error choosing all of the other NC6 options and also when changing the region to East US and East US 2 according to availability breakdown at http://aka.ms/azure-regions
The other 2 options don’t have GPUs so I’m not sure what else to do given the options. Any help would be appreciated!
DSVM not provisioning VM correctly. It seems like the DSVM instructions are no longer working. When I use this link http://aka.ms/dsvm/fastai, by default I see this option for the VM size
When I try with this default option I got this error
The requested VM size Standard_NC6_Promo is not available in the current region. The sizes available in the current region are: Standard_B1ls,Standard_B1ms,Standard_B1s,Standard_B2ms,Standard_B2s,Standard_B4ms,Standard_B8ms,Standard_B12ms,Standard_B16ms,Standard_B20ms,Standard_DS1_v2,Standard_DS2_v2,Standard_DS3_v2,Standard_DS4_v2,Standard_DS5_v2,Standard_DS11-1_v2,Standard_DS11_v2,Standard_DS12-1_v2,Standard_DS12-2_v2,Standard_DS12_v2,Standard_DS13-2_v2,Standard_DS13-4_v2,Standard_DS13_v2,Standard_DS14-4_v2,Standard_DS14-8_v2,Standard_DS14_v2,Standard_DS15_v2,Standard_DS2_v2_Promo,Standard_DS3_v2_Promo,Standard_DS4_v2_Promo,Standard_DS5_v2_Promo,Standard_DS11_v2_Promo,Standard_DS12_v2_Promo,Standard_DS13_v2_Promo,Standard_DS14_v2_Promo,Standard_F1s,Standard_F2s,Standard_F4s,Standard_F8s,Standard_F16s,Standard_D2s_v3,Standard_D4s_v3,Standard_D8s_v3,Standard_D16s_v3,Standard_D32s_v3,Standard_A0,Standard_A1,Standard_A2,Standard_A3,Standard_A5,Standard_A4,Standard_A6,Standard_A7,Basic_A0,Basic_A1,Basic_A2,Basic_A3,Basic_A4,Standard_D1_v2,Standard_D2_v2,Standard_D3_v2,Standard_D4_v2,Standard_D5_v2,Standard_D11_v2,Standard_D12_v2,Standard_D13_v2,Standard_D14_v2,Standard_D15_v2,Standard_D2_v2_Promo,Standard_D3_v2_Promo,Standard_D4_v2_Promo,Standard_D5_v2_Promo,Standard_D11_v2_Promo,Standard_D12_v2_Promo,Standard_D13_v2_Promo,Standard_D14_v2_Promo,Standard_F1,Standard_F2,Standard_F4,Standard_F8,Standard_F16,Standard_A1_v2,Standard_A2m_v2,Standard_A2_v2,Standard_A4m_v2,Standard_A4_v2,Standard_A8m_v2,Standard_A8_v2,Standard_D2_v3,Standard_D4_v3,Standard_D8_v3,Standard_D16_v3,Standard_D32_v3,Standard_D48_v3,Standard_D64_v3,Standard_D48s_v3,Standard_D64s_v3,Standard_E2_v3,Standard_E4_v3,Standard_E8_v3,Standard_E16_v3,Standard_E20_v3,Standard_E32_v3,Standard_E48_v3,Standard_E64i_v3,Standard_E64_v3,Standard_E2s_v3,Standard_E4-2s_v3,Standard_E4s_v3,Standard_E8-2s_v3,Standard_E8-4s_v3,Standard_E8s_v3,Standard_E16-4s_v3,Standard_E16-8s_v3,Standard_E16s_v3,Standard_E20s_v3,Standard_E32-8s_v3,Standard_E32-16s_v3,Standard_E32s_v3,Standard_E48s_v3,Standard_E64-16s_v3,Standard_E64-32s_v3,Standard_E64is_v3,Standard_E64s_v3,Standard_D1,Standard_D2,Standard_D3,Standard_D4,Standard_D11,Standard_D12,Standard_D13,Standard_D14,Standard_DS1,Standard_DS2,Standard_DS3,Standard_DS4,Standard_DS11,Standard_DS12,Standard_DS13,Standard_DS14,Standard_L8s_v2,Standard_L16s_v2,Standard_L32s_v2,Standard_L48s_v2,Standard_L64s_v2,Standard_L80s_v2,Standard_G1,Standard_G2,Standard_G3,Standard_G4,Standard_G5,Standard_GS1,Standard_GS2,Standard_GS3,Standard_GS4,Standard_GS4-4,Standard_GS4-8,Standard_GS5,Standard_GS5-8,Standard_GS5-16,Standard_L4s,Standard_L8s,Standard_L16s,Standard_L32s,Standard_F2s_v2,Standard_F4s_v2,Standard_F8s_v2,Standard_F16s_v2,Standard_F32s_v2,Standard_F48s_v2,Standard_F64s_v2,Standard_F72s_v2,Standard_A8,Standard_A9,Standard_A10,Standard_A11,Standard_H8,Standard_H8_Promo,Standard_H16,Standard_H16_Promo,Standard_H8m,Standard_H8m_Promo,Standard_H16m,Standard_H16m_Promo,Standard_H16r,Standard_H16r_Promo,Standard_H16mr,Standard_H16mr_Promo,Standard_NV6s_v2,Standard_NV12s_v2,Standard_NV24s_v2,Standard_NV12s_v3,Standard_NV24s_v3,Standard_NV48s_v3. Find out more on the available VM sizes in each region at https://aka.ms/azure-regions.
I got this same error choosing all of the other NC6 options and also when changing the region to East US and East US 2 according to availability breakdown at http://aka.ms/azure-regions
The other 2 options don’t have GPUs so I’m not sure what else to do given the options. Any help would be appreciated!
This could be temporary / transient issue with availability of GPU machines (likely due to increase demand from research community for covid-19). I tried on a MSDN Benefits account today and was able to spin up a VM on Standard_NC6 in East US2. (Standard)_NC6_Promo failed due to lack of availability). You can also check in Usage+Quotas on Portal within “Subscriptions” blade if you have quota set for NC Family in the desired regions. By default you should be able to spin upto 18 cores (3 GPUs) but things may be tight in some regions. At this point, best option is to try different regions (even outside US). You can also try Low Priority (Preemptable).
Hi,
I have a azure VM running. If I am away from my laptop while my code is executing through jupyter notebook for a long time , the code seems to be getting stuck. The progress bar does not show progress. Would anyone know how to debug this ? Currently I am just killing the kernel and restarting which is a waste of time/resources
P.S: the image below is a sample after I restarted my kernel
A few questions: Is that a Data Science VM on Azure? Also are you using Windows or Linux VM? How long was the notebook “idle” before the code appears stuck?
Of all the solutions and scripts I tried, this is the only one that took me where I needed to go to set this up. This resource should be on the website, not the outdated scripts presented there. The old scripts don’t work at all.
Thank you @twaldecker for putting this together.
Hi, I am receiving this error message when trying to use the template. Could anyone help me with this?
{“code”:“InvalidTemplateDeployment”,“details”:[{“code”:“InvalidParameter”,“target”:“imageReference”,“message”:“The following list of images referenced from the deployment template are not found: Publisher: microsoft-dsvm, Offer: dsvm-windows, Sku: server-2016, Version: 19.01.14. Please refer to Find and use marketplace purchase plan information using PowerShell - Azure Virtual Machines | Microsoft Learn for instructions on finding available images.”}],“message”:“The template deployment ‘Microsoft.Template-20231112135655’ is not valid according to the validation procedure. The tracking id is ‘88a8daa8-31c0-4c3e-9bbc-a99643bc26a1’. See inner errors for details.”}
