Error during install in HPC

[robin-sto]

I am getting an error when trying to install, please help:
error: subprocess-exited-with-error

× python setup.py egg_info did not run successfully.
│ exit code: 1
╰─> [6 lines of output]
Traceback (most recent call last):
File "", line 2, in
File "", line 34, in
File "/tmp/pip-install-fsgv573f/flash-attn_f375291c450540ad8ce1e11cb0a1eef7/setup.py", line 8, in
from packaging.version import parse, Version
ModuleNotFoundError: No module named 'packaging'
[end of output]

[subercui]

Hi, flash-attn 1.0.9 may be too new a version. Can you try pip install "flash-attn<1.0.5". Meanwhile, you may have a look at this one Dao-AILab/flash-attention#453 (comment) . It seems related

[Bec-k]

same here

[subercui]

Hi @blackwidowst, did you solve it? Feel free to let me know if any updates or questions

[subercui]

Hi @Bec-k , would you have a try with the above suggestion and could you share the error messages?

[robin-sto]

It is not working yet, still trying.

[Allamyh]

I got the same error message, solved by installing the "packaging" module. In my case "$ pip install packaging"

[robin-sto]

I was finally able to get it running but now having an issue with the program. When running the sample code available at https://github.com/bowang-lab/scGPT/blob/main/tutorials/Tutorial_Reference_Mapping.ipynb, the function scg.tasks.embed_data() in the code throws an error saying: AttributeError: module 'scgpt.tasks' has no attribute 'embed_data'

Has anyone run into this?

[subercui]

Hi, @blackwidowst

1. If you run the code in the main branch, meaning if you have cloned and updated the repo and run the notebook in the tutorials folder, then the function should be accessible. Here in the the tutorial,
   
   It tries to import the code in the cloned repo and the function is imported here

   scGPT/scgpt/tasks/__init__.py

   Line 2 in ba9b9ef

   from .cell_emb import get_batch_cell_embeddings, embed_data

2. If you try to run the code with the pip installed package, meaning if you just pip installed the package without downloading the repo. Then please update to the newest version by pip install -U scgpt. I just released the new version 0.1.5, and it has this new feature now.

Please let me know how it works