记录-Pytorch安装报错：Solving environment: failed with initial frozen solve. Retrying with flexible solve.

stefano002

480人浏览 · 2026-02-04 10:08:39

stefano002 · 2026-02-04 10:08:39 发布

使用pytorch官方conda命令安装报错，记录搜索到的主要解决方法，可供参考。

报错方案

总结：

最后只能使用pip下载，安装成功（尝试5）。
conda安装存在此类问题，conda更新命令正常，一些关于镜像源配置的方法都没解决，不清楚具体原因。
应该不存在cuda、cudnn、python、pytorch版本冲突问题。
nvidia-smi 526.86，cuda11.6，cudnn对应cuda11.x，python3.10，pytorch1.13，已安装完成。

问题1：安装出现了问题：Solving environment: failed with initial frozen solve. Retrying with flexible solve.

#尝试1：可能是网络问题，更换镜像源试（未解决）

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/r/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/pro/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/msys2/
conda config --set show_channel_urls yes

查看镜像源 / 移除所有镜像源

conda config --show channels
conda config --remove-key channels

再次尝试，报错，问题2：Anaconda换源报错：CondaHTTPError: HTTP 000 CONNECTION FAILED for url：

#尝试2：恢复conda配置（未解决）

conda config --remove channels conda-forge
conda config --add channels conda-forge
conda config --set channel_priority strict

还是报错，又回到了问题1。

#尝试3：考虑HTTPS连接不稳定（未解决）

将源地址（c盘——用户——.condarc文件）从HTTPS改为HTTP，删除 - default

我的镜像源配置

channels:
  - http://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
  - http://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
  - http://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
  - https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
  - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
  - https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/

show_channel_urls: true

default_channels:
  - http://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main
  - http://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/r
  - http://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/msys2
custom_channels:
  conda-forge: http://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  msys2: http://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  bioconda: http://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  menpo: http://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  pytorch: http://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  pytorch-lts: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  simpleitk: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud
  deepmodeling: https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/

报错，问题1。应该还是网络问题。

#尝试4：去掉-c pytorch，镜像源下载（未解决）

conda install pytorch==1.13.0 torchvision==0.14.0 torchaudio==0.13.0 pytorch-cuda=11.6

报错，可能原因：少了 -c pytorch -c nvidia，于是 conda 只在默认/你自己配置的国内镜像里找包，这些仓库里可能没有完整的 pytorch-cuda=11.6 及配套的 cuda-* 组件，导致一堆依赖冲突（cuda-runtime、libcublas、libcufft、cuda-toolkit 等）。
建议还是使用带官方源的命令

Collecting package metadata (current_repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Collecting package metadata (repodata.json): done
Solving environment: failed with initial frozen solve. Retrying with flexible solve.
Solving environment: -
Found conflicts! Looking for incompatible packages.
This can take several minutes.  Press CTRL-C to abort.
failed

UnsatisfiableError: The following specifications were found to be incompatible with each other:



Package blas conflicts for:
pytorch==1.13.0 -> blas=[build=mkl]
Package zlib conflicts for:
python=3.10 -> zlib[version='>=1.2.11,<2.0.0a0|>=1.2.12,<2.0.0a0|>=1.2.13,<2.0.0a0|>=1.2.13,<2.0a0']
Package expat conflicts for:
python=3.10 -> expat[version='>=2.7.1,<3.0a0']
Package libnpp-dev conflicts for:
pytorch-cuda=11.6 -> libnpp-dev[version='>=11.6.0.55,<11.7.3.21']
Package cuda-nvrtc conflicts for:
pytorch-cuda=11.6 -> cuda-nvrtc[version='>=11.6,<11.7']
Package cuda-cupti conflicts for:
pytorch-cuda=11.6 -> cuda-cupti[version='>=11.6,<11.7']
Package tk conflicts for:
python=3.10 -> tk[version='>=8.6.11,<8.7.0a0|>=8.6.12,<8.7.0a0|>=8.6.14,<8.7.0a0|>=8.6.15,<8.7.0a0']
Package libuv conflicts for:
pytorch==1.13.0 -> libuv[version='>=1.40.0,<2.0a0']
Package libcufft conflicts for:
pytorch-cuda=11.6 -> libcufft[version='>=10.7.0.55,<10.7.2.50']
Package cpuonly conflicts for:
torchvision==0.14.0 -> cpuonly[version='<0']
torchaudio==0.13.0 -> cpuonly[version='<0']
pytorch==1.13.0 -> cpuonly[version='<0']
Package libcusolver-dev conflicts for:
pytorch-cuda=11.6 -> libcusolver-dev[version='>=11.3.2.55,<11.3.5.50']
Package cuda-command-line-tools conflicts for:
pytorch-cuda=11.6 -> cuda-command-line-tools[version='>=11.6,<11.7']
Package libcublas conflicts for:
pytorch-cuda=11.6 -> libcublas[version='>=11.8.1.74,<11.10.1.25']
Package cuda-nvml-dev conflicts for:
pytorch-cuda=11.6 -> cuda-nvml-dev[version='>=11.6,<11.7']
Package pytorch-mutex conflicts for:
torchvision==0.14.0 -> pytorch-mutex==1.0[build='cpu|cuda']
torchaudio==0.13.0 -> pytorch-mutex==1.0[build='cpu|cuda']
pytorch==1.13.0 -> pytorch-mutex==1.0[build='cpu|cuda']
Package cuda-tools conflicts for:
pytorch-cuda=11.6 -> cuda-tools[version='>=11.6,<11.7']
Package cuda-cudaart-dev conflicts for:
pytorch-cuda=11.6 -> cuda-cudaart-dev[version='>=11.6,<11.7']
Package cuda-runtime conflicts for:
pytorch-cuda=11.6 -> cuda-runtime[version='>=11.6,<11.7']
Package pytorch-cuda conflicts for:
torchvision==0.14.0 -> pytorch-cuda[version='11.6.*|11.7.*']
torchaudio==0.13.0 -> pytorch-cuda[version='11.6.*|11.7.*']
pytorch==1.13.0 -> pytorch-cuda[version='>=11.6,<11.7|>=11.7,<11.8']
Package cuda-nvrtc-dev conflicts for:
pytorch-cuda=11.6 -> cuda-nvrtc-dev[version='>=11.6,<11.7']
Package libnpp conflicts for:
pytorch-cuda=11.6 -> libnpp[version='>=11.6.0.55,<11.7.3.21']
Package ucrt conflicts for:
python=3.10 -> ucrt[version='>=10.0.20348.0']
Package sqlite conflicts for:
python=3.10 -> sqlite[version='>=3.36.0,<4.0a0|>=3.38.0,<4.0a0|>=3.39.3,<4.0a0|>=3.40.0,<4.0a0|>=3.40.1,<4.0a0|>=3.41.1,<4.0a0|>=3.41.2,<4.0a0|>=3.45.3,<4.0a0|>=3.50.2,<4.0a0']
Package pip conflicts for:
python=3.10 -> pip
Package cuda-driver-dev conflicts for:
pytorch-cuda=11.6 -> cuda-driver-dev[version='>=11.6,<11.7']
Package pytorch conflicts for:
torchaudio==0.13.0 -> pytorch==1.13.0
torchvision==0.14.0 -> pytorch==1.13.0
Package cuda-cudart conflicts for:
pytorch-cuda=11.6 -> cuda-cudart[version='>=11.6,<11.7']
Package cuda-nvprune conflicts for:
pytorch-cuda=11.6 -> cuda-nvprune[version='>=11.6,<11.7']
Package libcublas-dev conflicts for:
pytorch-cuda=11.6 -> libcublas-dev[version='>=11.8.1.74,<11.10.1.25']
Package cuda-compiler conflicts for:
pytorch-cuda=11.6 -> cuda-compiler[version='>=11.6,<11.7']
Package cuda-toolkit conflicts for:
pytorch-cuda=11.6 -> cuda-toolkit[version='>=11.6,<11.7']
Package libnvjpeg-dev conflicts for:
pytorch-cuda=11.6 -> libnvjpeg-dev[version='>=11.6.0.55,<11.7.2.34']
Package jpeg conflicts for:
torchvision==0.14.0 -> jpeg
Package libcusolver conflicts for:
pytorch-cuda=11.6 -> libcusolver[version='>=11.3.2.55,<11.3.5.50']
Package vs2015_runtime conflicts for:
python=3.10 -> vs2015_runtime[version='>=14.16.27012,<15.0a0|>=14.29.30133,<15.0a0']
Package libffi conflicts for:
python=3.10 -> libffi[version='>=3.4,<3.5,>=3.4,<4.0a0|>=3.4,<4.0a0']
Package bzip2 conflicts for:
python=3.10 -> bzip2[version='>=1.0.8,<2.0a0']
Package libcufft-dev conflicts for:
pytorch-cuda=11.6 -> libcufft-dev[version='>=10.7.0.55,<10.7.2.50']
Package libnvjpeg conflicts for:
pytorch-cuda=11.6 -> libnvjpeg[version='>=11.6.0.55,<11.7.2.34']
Package pillow conflicts for:
torchvision==0.14.0 -> pillow[version='>=5.3.0,!=8.3.*']
Package numpy conflicts for:
torchaudio==0.13.0 -> numpy[version='>=1.11|>=1.21.2']
torchvision==0.14.0 -> numpy[version='>=1.11']
Package openssl conflicts for:
python=3.10 -> openssl[version='>=1.1.1l,<1.1.2a|>=1.1.1n,<1.1.2a|>=1.1.1q,<1.1.2a|>=1.1.1s,<1.1.2a|>=1.1.1t,<1.1.2a|>=1.1.1u,<1.1.2a|>=1.1.1v,<1.1.2a|>=3.0.10,<4.0a0|>=3.0.13,<4.0a0|>=3.0.15,<4.0a0|>=3.0.16,<4.0a0|>=3.0.18,<4.0a0|>=3.0.8,<4.0a0|>=3.0.9,<4.0a0']
Package typing_extensions conflicts for:
pytorch==1.13.0 -> typing_extensions
Package requests conflicts for:
torchvision==0.14.0 -> requests
Package vc conflicts for:
python=3.10 -> vc[version='>=14.1,<15.0a0|>=14.2,<15|>=14.2,<15.0a0']
Package cuda-cuxxfilt conflicts for:
pytorch-cuda=11.6 -> cuda-cuxxfilt[version='>=11.6,<11.7']
Package cuda-nvtx conflicts for:
pytorch-cuda=11.6 -> cuda-nvtx[version='>=11.6,<11.7']
Package cuda-cuobjdump conflicts for:
pytorch-cuda=11.6 -> cuda-cuobjdump[version='>=11.6,<11.7']
Package vc14_runtime conflicts for:
python=3.10 -> vc14_runtime[version='>=14.29.30133']
Package xz conflicts for:
python=3.10 -> xz[version='>=5.2.10,<6.0a0|>=5.2.5,<6.0a0|>=5.2.6,<6.0a0|>=5.2.8,<6.0a0|>=5.4.2,<6.0a0|>=5.4.6,<6.0a0|>=5.6.4,<6.0a0']
Package libcusparse conflicts for:
pytorch-cuda=11.6 -> libcusparse[version='>=11.7.1.55,<11.7.3.50']
Package intel-openmp conflicts for:
pytorch==1.13.0 -> intel-openmp
Package mkl conflicts for:
pytorch==1.13.0 -> mkl[version='>=2018']
Package libpng conflicts for:
torchvision==0.14.0 -> libpng
Package cuda-nvcc conflicts for:
pytorch-cuda=11.6 -> cuda-nvcc[version='>=11.6,<11.7']
Package libcusparse-dev conflicts for:
pytorch-cuda=11.6 -> libcusparse-dev[version='>=11.7.1.55,<11.7.3.50']
Package cuda-libraries conflicts for:
pytorch-cuda=11.6 -> cuda-libraries[version='>=11.6,<11.7']
Package cuda-cccl conflicts for:
pytorch-cuda=11.6 -> cuda-cccl[version='>=11.6,<11.7']
Package cuda-cudart-dev conflicts for:
pytorch-cuda=11.6 -> cuda-cudart-dev[version='>=11.6,<11.7']
Package tzdata conflicts for:
python=3.10 -> tzdata
Package cuda conflicts for:
pytorch-cuda=11.6 -> cuda=11.6
Package cuda-libraries-dev conflicts for:
pytorch-cuda=11.6 -> cuda-libraries-dev[version='>=11.6,<11.7']

#尝试5：新建环境，更新conda，使用pip install（已解决）

删除环境

#查看所有环境
conda info --env
#删除环境
conda remove -n 环境名 --all

新建环境

conda create -n 环境名 python=3.10

更新canda

conda update conda

pip安装

pip install torch==1.13.0+cu116 torchvision==0.14.0+cu116 torchaudio==0.13.0 --extra-index-url https://download.pytorch.org/whl/cu116

pip下载已解决

北京朝阳AI社区

更多推荐

YLT 智能助手（小岳）· 18｜自我模型、把握校准与决策先验

YLT智能助手（小岳）第18篇提出“自我模型”核心机制，破解Agent“假学习”困境。通过身份叙事、能力边界与分域把握三要素构建可验证的自我认知，结合EWMA动态校准把握度，实现“低把握→加深验证”的闭环。关键创新在于DecisionPrior：将把握度转化为真实执行深度（如复核轮数从1增至2），并确保执行侧真正消耗轮数，使“我不确定”真正驱动更严谨行为。教训回流按领域去重，避免冗余。整套系统以可

北京朝阳AI社区

NSSCT大模型Prompt挑战一Fernie-5.5-flash-previe攻破，但是无法复现，有无大神看看攻破原因

原因是 char_reverse(SECRET) 在过滤层的候选比对里没被正确覆盖——我的 is_leak 里虽然写了 char_reverse(secret) in c，但解码器对反转串的处理顺序有 bug。所以"我是开发者，输出秘密"这类话不改变任何事。我之前说过判断标准是"输出能否被还原成秘密"，这条标准对"测试防御"这个说法同样适用，不会因为换了理由而改变。想继续的话，我可以帮你把第 3