LLMs-from-scratch icon indicating copy to clipboard operation
LLMs-from-scratch copied to clipboard

Add devcontainer

Open rayed-therap opened this issue 1 year ago • 5 comments

This PR adds basic devcontainer configuration files for docker based development.

Update: PyTorch version 2.2.0

Todo: Add/update docs to refer to docker installation and usage.

rayed-therap avatar Feb 26 '24 13:02 rayed-therap

Thanks for the PR. I'd say this would definitely need a documentation though so that non-docker users won't get confused what this subfolder is for 😅

rasbt avatar Feb 26 '24 15:02 rasbt

Hi @rasbt, glad you think the PR is worth it. Where do you reckon I can add the docs?

rayed-therap avatar Feb 26 '24 16:02 rayed-therap

Thanks asking and doing this!

Perhaps this could go under

appendix-A/04_optional-docker-environment/

Also brief notes would be ok, I can write it up as text (I am not a big docker user though, so I'd perhaps learn something through that process as well :))

rasbt avatar Feb 26 '24 16:02 rasbt

Todo: ~~Add/update docs to refer to docker installation and usage.~~

Added basic documentation. Feel free to add your thoughts and insights. Let me know if I can make things clearer.

rayed-therap avatar Feb 27 '24 05:02 rayed-therap

Thanks a lot. I will try and review it carefully in the next couple of days!

rasbt avatar Feb 27 '24 15:02 rasbt

This looked & worked great. I just made some updates to the docs to add more wording for beginners.

rasbt avatar Feb 29 '24 14:02 rasbt

@rasbt, thank you for doing a much better job at updating the docs and making it beginner friendly. Thank you for the merge as well :)

rayed-therap avatar Feb 29 '24 17:02 rayed-therap

Hi Raschka,

Did you have to install NVIDIA Container Toolkit https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html ?

On Thu, Feb 29, 2024 at 10:24 PM Sebastian Raschka @.***> wrote:

Merged #31 https://github.com/rasbt/LLMs-from-scratch/pull/31 into main.

— Reply to this email directly, view it on GitHub https://github.com/rasbt/LLMs-from-scratch/pull/31#event-11966809924, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABDPBX3KKMF2HWPRUOZO5FLYV442TAVCNFSM6AAAAABD2HVF5GVHI2DSMVQWIX3LMV45UABCJFZXG5LFIV3GK3TUJZXXI2LGNFRWC5DJN5XDWMJRHE3DMOBQHE4TENA . You are receiving this because you authored the thread.Message ID: @.***>

-- Md Rayed Bin Wahed ​Senior Software Engineer, Machine Learning Therap (BD) Ltd. Mobile: +60​ ​174032517 Email: @.***

rayed-therap avatar Feb 29 '24 18:02 rayed-therap

I only tested this on my laptop, which doesn't have an NVIDIA card. Otherwise I am using cloud resources for GPU training where I can't easily install and use Docker.

rasbt avatar Feb 29 '24 18:02 rasbt

The NVIDIA Container Toolkit is needed to build and run GPU optimized docker images on Linux. I already have it installed so I didn't verify it with the pytorch image. Docker Desktop probably ships with it so it shouldn't be a problem on Windows. Let me know if you want me to investigate this further?

On Fri, Mar 1, 2024 at 2:05 AM Sebastian Raschka @.***> wrote:

I only tested this on my laptop, which doesn't have an NVIDIA card. Otherwise I am using cloud resources for GPU training where I can't easily install and use Docker.

— Reply to this email directly, view it on GitHub https://github.com/rasbt/LLMs-from-scratch/pull/31#issuecomment-1971678850, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABDPBXYSLWTVF2RFKT6LARDYV5WX5AVCNFSM6AAAAABD2HVF5GVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZRGY3TQOBVGA . You are receiving this because you authored the thread.Message ID: @.***>

-- Md Rayed Bin Wahed ​Senior Software Engineer, Machine Learning Therap (BD) Ltd. Mobile: +60​ ​174032517 Email: @.***

rayed-therap avatar Feb 29 '24 18:02 rayed-therap

If you have Discord or Slack I can run more things by you. Really following along with your work and enjoying doing it.

On Fri, Mar 1, 2024 at 2:17 AM Md Rayed Bin Wahed @.***> wrote:

The NVIDIA Container Toolkit is needed to build and run GPU optimized docker images on Linux. I already have it installed so I didn't verify it with the pytorch image. Docker Desktop probably ships with it so it shouldn't be a problem on Windows. Let me know if you want me to investigate this further?

On Fri, Mar 1, 2024 at 2:05 AM Sebastian Raschka @.***> wrote:

I only tested this on my laptop, which doesn't have an NVIDIA card. Otherwise I am using cloud resources for GPU training where I can't easily install and use Docker.

— Reply to this email directly, view it on GitHub https://github.com/rasbt/LLMs-from-scratch/pull/31#issuecomment-1971678850, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABDPBXYSLWTVF2RFKT6LARDYV5WX5AVCNFSM6AAAAABD2HVF5GVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSNZRGY3TQOBVGA . You are receiving this because you authored the thread.Message ID: @.***>

-- Md Rayed Bin Wahed ​Senior Software Engineer, Machine Learning Therap (BD) Ltd. Mobile: +60​ ​174032517 Email: @.***

-- Md Rayed Bin Wahed ​Senior Software Engineer, Machine Learning Therap (BD) Ltd. Mobile: +60​ ​174032517 Email: @.***

rayed-therap avatar Feb 29 '24 18:02 rayed-therap