已安装torch但无法安装auto-gptq,提示需先安装torch
安装auto-gptq时提示“torch未安装”但实际已安装的问题解决
问题描述
我尝试在本地安装auto-gptq时遇到以下错误:
Collecting auto-gptq Using cached auto_gptq-0.3.2.tar.gz (63 kB) Installing build dependencies ... done Getting requirements to build wheel ... error error: subprocess-exited-with-error × Getting requirements to build wheel did not run successfully. │ exit code: 4294967295 ╰─> [1 lines of output] torch is not installed, please install torch first! [end of output] note: This error originates from a subprocess, and is likely not a problem with pip. error: subprocess-exited-with-error × Getting requirements to build wheel did not run successfully. │ exit code: 4294967295 ╰─> See above for output. note: This error originates from a subprocess, and is likely not a problem with pip.
我已安装torch,执行安装命令时显示:
PS > C:\...\python.exe -m pip install torch Requirement already satisfied: torch in c:\... (2.0.1) Requirement already satisfied: filelock in c:\... (from torch) (3.12.2) Requirement already satisfied: typing-extensions in c:\... (from torch) (4.7.1) Requirement already satisfied: sympy in c:\... (from torch) (1.12) Requirement already satisfied: networkx in c:\... (from torch) (3.1) Requirement already satisfied: jinja2 in c:\... (from torch) (3.1.2) Requirement already satisfied: MarkupSafe>=2.0 in c:\... (from jinja2->torch) (2.1.3) Requirement already satisfied: mpmath>=0.19 in c:\... (from sympy->torch) (1.3.0)
但仍出现上述错误,请问这是为什么?
可能的原因及解决方法
Python环境不匹配:安装auto-gptq时使用的Python环境,与安装torch的环境不一致。比如默认终端调用的是系统Python,但torch装在虚拟环境里。解决方式:激活torch所在的虚拟环境后再执行安装命令,或者直接用torch所在的Python解释器运行安装:
C:\...\python.exe -m pip install auto-gptq替换路径为你实际的Python可执行文件路径。
版本兼容性问题:auto-gptq 0.3.2对torch版本有特定要求,torch 2.0.1可能不在支持范围内。可以核对auto-gptq的版本兼容说明,要么升级/降级torch到兼容版本,要么安装对应兼容的auto-gptq版本。
pip缓存干扰:pip缓存可能导致依赖检测失效。先清除缓存再重新安装:
pip cache purge之后再执行auto-gptq的安装命令。
CUDA依赖缺失:如果安装的是CPU版torch,但auto-gptq默认需要GPU版torch支持,也会触发该错误。确认安装的是带CUDA的torch版本,或者安装时指定auto-gptq的CPU兼容版本(如果官方支持)。
内容的提问来源于stack exchange,提问作者Daniel Burmester
相关产品推荐
相关产品推荐

