[Bug] 'nvidia-cublas-cu13' is deprecated causes the installation of lmdeploy from source using uv to fail. #4116
Description
Checklist
- 1. I have searched related issues but cannot get the expected help.
- 2. The bug has not been fixed in the latest version.
- 3. Please note that if the bug-related issue you submitted lacks corresponding environment info and a minimal reproducible demo, it will be challenging for us to reproduce and resolve the issue, reducing the likelihood of receiving feedback.
Describe the bug
Since the Qwen3-VL support in the main branch has not been released yet, I attempted to install lmdeploy from source using uv, and the console reported the following error:
[stderr] ⚠️ THIS PROJECT 'nvidia-cublas-cu13' IS DEPRECATED. Please use 'nvidia-cublas' instead.
Reproduction
(lmdeploy) root@ubuntu:~/lmdeploy# uv pip install -U git+https://github.com/InternLM/lmdeploy.git
Resolved 104 packages in 2.51s
× Failed to build `nvidia-cublas-cu13==0.0.1`
├─▶ The build backend returned an error
╰─▶ Call to `setuptools.build_meta.build_wheel` failed (exit status: 1)
[stderr]
⚠️ THIS PROJECT 'nvidia-cublas-cu13' IS DEPRECATED.
Please use 'nvidia-cublas' instead.
To install the correct package, use:
pip install nvidia-cublas
For more information, visit: https://pypi.org/project/nvidia-cublas/
hint: This usually indicates a problem with the package or the build environment.
help: `nvidia-cublas-cu13` (v0.0.1) was included because `lmdeploy` (v0.10.2) depends on `nvidia-cublas-cu13`Environment
(lmdeploy) root@ubuntu:~/lmdeploy# python -V
Python 3.12.11
(lmdeploy) root@ubuntu:~/lmdeploy# nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2025 NVIDIA Corporation
Built on Wed_Aug_20_01:58:59_PM_PDT_2025
Cuda compilation tools, release 13.0, V13.0.88
Build cuda_13.0.r13.0/compiler.36424714_0