Conda环境和下载pip包(torch)在8GB A100上很慢
- Dominant language
- Python
- Stars
- 2k
- Forks
- 1.5k
- PR merge metrics
- No merged PRs in 30d
Description
Hello,
In 1.2. 基础关卡, 第 2 关 (the demo which gives 30% A100 computing access), it takes a very long time to set up the "demo" conda environment (approximately 5 minutes); when I install a pip package it takes 25 minutes, for 30 minutes in total.
Mainly, the issue is installing the packages; downloading works okay but when installing the collected packages it takes quite a long time.
If creating the demo conda environment takes time due to the small amount of CPU resources in the virtual machine, maybe it would be a good idea to do either or both of two things:
1. Tell learners in advance to partition some of their time specifically to only download the conda environment and LLM, or
2. Create the demo conda environment on all the 10% A100 machines (which should be possible as the download size is ~5 GB, while the total memory available to users is ~40 GB.)
Do you think this a good idea? Thank you very much!
Contributor guide
No contributing guide indexed for this repository
Research direction
Start with the 1.2 Basic Level, Lesson 2 demo setup described in the issue and measure conda-environment creation and pip installation time on the 8GB and 10% A100 machines. Decide whether the work is learner guidance, pre-creating the demo environment, or both. Done means the chosen approach is implemented and the setup-time expectations or results are documented.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- devops, machine-learning
- Issue type
- Feature
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 35/100