virtual environment with torch cpu forces to install torch gpu

[mgathena]

System Info

N/A
(reason: couldn't install)
Python 3.10.12

Information

• The official example scripts
• My own modified scripts

Tasks

• One of the scripts in the examples/ folder of Accelerate or an officially supported no_trainer script in the examples folder of the transformers repo (such as run_no_trainer_glue.py)
• My own task or dataset (give details below)

Reproduction

I am trying to install accelerate via uv. I already have torch==2.4.1+cpu installed. but when i install via uv add accelerate, it starts to download cuda/triton/nccl/torch(GPU). Is there a way to prevent the installation of GPU version of torch? I also tried to pin the version like uv add accelerate==0.14, but it didn't work.

Expected behavior

It shouldn't download the GPU version and continue to work with the CPU version of pytorch