I am using GPU clusters, and for whatever reason I am not able to run Large Language Model, Flan xl T5. I traced the problem and found out that the version of Pytorch must be higher than 2.2. However, the thing that surprises me the most is that I have Pytorch version 2.2.1 installed via pip. I know this because I used the "pip list" command and got torch==2.2.1. note: there are more packages I have installed but it is too long.
<my username>:~/Q$ pip list
WARNING: Ignoring invalid distribution - (/mnt/lime/homes/sw2780/.local/lib/python3.8/site-packages)
WARNING: Ignoring invalid distribution -ransformers (/mnt/lime/homes/sw2780/.local/lib/python3.8/site-packages)
Package Version
---------------------------- ---------------------
torch 2.2.1
However, the weird thing is that when I run the command "python3" then do torch.version, it comes out as 1.11.1+cu113.
<my username>:~/Q$ python3
Python 3.8.10 (default, May 26 2023, 14:05:08)
[GCC 9.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> torch.__version__
'1.11.0+cu113'
>>>
I can't remember whether I installed this version or not. but I am sure I did not use Conda to install it. When I am trying to "pip uninstall", it is trying to uninstall 2.2.1 version instead of 1.11.1+cu113. I really want to fix this problem as fast as possible. Could anyone help me please? Thanks in Advance.