這篇文章主要分為兩部分：
(1)離線環境下如何部署Ollama；
(2)在離線環境下如何配置大模型，其中這一步又分為：
?1)部署完整的deepseek大模型，如：deepseek-r1:32B;
?2)部署蒸餾版模型，如：DeepSeek-R1-Distill-Llama-32B-Q5_K_M.gguf。
(deepseek-r1:32B是完整的32B(320億參數)的DeepSeek-R1模型，DeepSeek-R1-Distill-Llama-32B-Q5_K_M.gguf經過量化優化，運行效率更高，適用于設備資源有限(如GPU顯存較小)的情況。如果想要最高的推理質量，可以嘗試完整的deepseek-r1:32B,但可能需要更高端的硬件。具體顯卡要求可以查看這篇博客：https://blog.csdn.net/xiangerersheng/article/details/145640259。

(1)離線環境下部署ollama

一、通過瀏覽器訪問ollama官方安裝腳本，獲取腳本內容，復制里面的所有內容。并在linux中執行vi install.sh，將復制的內容粘貼進去。
（https://ollama.com/install.sh）。
二、修改install.sh腳本，下面是完整的腳本，也可以忽略第一步直接復制下面的所有代碼寫入install.sh。這里主要的修改是：連網獲取ollama的部分改為從本地獲取。

#!/bin/sh
# This script installs Ollama on Linux.
# It detects the current operating system architecture and installs the appropriate version of Ollama.set -eustatus() { echo ">>> $*" >&2; }
error() { echo "ERROR $*"; exit 1; }
warning() { echo "WARNING: $*"; }TEMP_DIR=$(mktemp -d)
cleanup() { rm -rf $TEMP_DIR; }
trap cleanup EXITavailable() { command -v $1 >/dev/null; }
require() {local MISSING=''for TOOL in $*; doif ! available $TOOL; thenMISSING="$MISSING $TOOL"fidoneecho $MISSING
}[ "$(uname -s)" = "Linux" ] || error 'This script is intended to run on Linux only.'ARCH=$(uname -m)
case "$ARCH" inx86_64) ARCH="amd64" ;;aarch64|arm64) ARCH="arm64" ;;*) error "Unsupported architecture: $ARCH" ;;
esacIS_WSL2=falseKERN=$(uname -r)
case "$KERN" in*icrosoft*WSL2 | *icrosoft*wsl2) IS_WSL2=true;;*icrosoft) error "Microsoft WSL1 is not currently supported. Please use WSL2 with 'wsl --set-version <distro> 2'" ;;*) ;;
esacVER_PARAM="${OLLAMA_VERSION:+?version=$OLLAMA_VERSION}"SUDO=
if [ "$(id -u)" -ne 0 ]; then# Running as root, no need for sudoif ! available sudo; thenerror "This script requires superuser permissions. Please re-run as root."fiSUDO="sudo"
fiNEEDS=$(require curl awk grep sed tee xargs)
if [ -n "$NEEDS" ]; thenstatus "ERROR: The following tools are required but missing:"for NEED in $NEEDS; doecho "  - $NEED"doneexit 1
fifor BINDIR in /usr/local/bin /usr/bin /bin; doecho $PATH | grep -q $BINDIR && break || continue
done
OLLAMA_INSTALL_DIR=$(dirname ${BINDIR})status "Installing ollama to $OLLAMA_INSTALL_DIR"
$SUDO install -o0 -g0 -m755 -d $BINDIR
$SUDO install -o0 -g0 -m755 -d "$OLLAMA_INSTALL_DIR"
#if curl -I --silent --fail --location "https://ollama.com/download/ollama-linux-${ARCH}.tgz${VER_PARAM}" >/dev/null ; then
#注釋掉以下代碼
#    status "Downloading Linux ${ARCH} bundle"
#    curl --fail --show-error --location --progress-bar \
#        "https://ollama.com/download/ollama-linux-${ARCH}.tgz${VER_PARAM}" | \
#        $SUDO tar -xzf - -C "$OLLAMA_INSTALL_DIR"
#    BUNDLE=1
#    if [ "$OLLAMA_INSTALL_DIR/bin/ollama" != "$BINDIR/ollama" ] ; then
#        status "Making ollama accessible in the PATH in $BINDIR"
#        $SUDO ln -sf "$OLLAMA_INSTALL_DIR/ollama" "$BINDIR/ollama"
#    fi
#else
#    status "Downloading Linux ${ARCH} CLI"
#    curl --fail --show-error --location --progress-bar -o "$TEMP_DIR/ollama"\
#    "https://ollama.com/download/ollama-linux-${ARCH}${VER_PARAM}"
#    $SUDO install -o0 -g0 -m755 $TEMP_DIR/ollama $OLLAMA_INSTALL_DIR/ollama
#    BUNDLE=0
#    if [ "$OLLAMA_INSTALL_DIR/ollama" != "$BINDIR/ollama" ] ; then
#        status "Making ollama accessible in the PATH in $BINDIR"
#        $SUDO ln -sf "$OLLAMA_INSTALL_DIR/ollama" "$BINDIR/ollama"
#    fi
#fi
#新增以下代碼
LOCAL_OLLAMA_TGZ="./ollama-linux-${ARCH}.tgz${VER_PARAM}"
if [ -f "$LOCAL_OLLAMA_TGZ" ]; thenstatus "Installing from local file $LOCAL_OLLAMA_TGZ"$SUDO tar -xzf "$LOCAL_OLLAMA_TGZ" -C "$OLLAMA_INSTALL_DIR"BUNDLE=1if [ ! -e "$BINDIR/ollama" ]; thenstatus "Making ollama accessible in the PATH in $BINDIR"$SUDO ln -sf "$OLLAMA_INSTALL_DIR/ollama" "$BINDIR/ollama"fi
elseecho "Error: The local file $LOCAL_OLLAMA_TGZ does not exist."exit 1
fiinstall_success() {status 'The Ollama API is now available at 127.0.0.1:11434.'status 'Install complete. Run "ollama" from the command line.'
}
trap install_success EXIT# Everything from this point onwards is optional.configure_systemd() {if ! id ollama >/dev/null 2>&1; thenstatus "Creating ollama user..."$SUDO useradd -r -s /bin/false -U -m -d /usr/share/ollama ollamafiif getent group render >/dev/null 2>&1; thenstatus "Adding ollama user to render group..."$SUDO usermod -a -G render ollamafiif getent group video >/dev/null 2>&1; thenstatus "Adding ollama user to video group..."$SUDO usermod -a -G video ollamafistatus "Adding current user to ollama group..."$SUDO usermod -a -G ollama $(whoami)status "Creating ollama systemd service..."cat <<EOF | $SUDO tee /etc/systemd/system/ollama.service >/dev/null
[Unit]
Description=Ollama Service
After=network-online.target[Service]
ExecStart=$BINDIR/ollama serve
User=ollama
Group=ollama
Restart=always
RestartSec=3
Environment="PATH=$PATH"[Install]
WantedBy=default.target
EOFSYSTEMCTL_RUNNING="$(systemctl is-system-running || true)"case $SYSTEMCTL_RUNNING inrunning|degraded)status "Enabling and starting ollama service..."$SUDO systemctl daemon-reload$SUDO systemctl enable ollamastart_service() { $SUDO systemctl restart ollama; }trap start_service EXIT;;esac
}if available systemctl; thenconfigure_systemd
fi# WSL2 only supports GPUs via nvidia passthrough
# so check for nvidia-smi to determine if GPU is available
if [ "$IS_WSL2" = true ]; thenif available nvidia-smi && [ -n "$(nvidia-smi | grep -o "CUDA Version: [0-9]*\.[0-9]*")" ]; thenstatus "Nvidia GPU detected."fiinstall_successexit 0
fi# Install GPU dependencies on Linux
if ! available lspci && ! available lshw; thenwarning "Unable to detect NVIDIA/AMD GPU. Install lspci or lshw to automatically detect and install GPU dependencies."exit 0
ficheck_gpu() {# Look for devices based on vendor ID for NVIDIA and AMDcase $1 inlspci)case $2 innvidia) available lspci && lspci -d '10de:' | grep -q 'NVIDIA' || return 1 ;;amdgpu) available lspci && lspci -d '1002:' | grep -q 'AMD' || return 1 ;;esac ;;lshw)case $2 innvidia) available lshw && $SUDO lshw -c display -numeric -disable network | grep -q 'vendor: .* \[10DE\]' || return 1 ;;amdgpu) available lshw && $SUDO lshw -c display -numeric -disable network | grep -q 'vendor: .* \[1002\]' || return 1 ;;esac ;;nvidia-smi) available nvidia-smi || return 1 ;;esac
}if check_gpu nvidia-smi; thenstatus "NVIDIA GPU installed."exit 0
fiif ! check_gpu lspci nvidia && ! check_gpu lshw nvidia && ! check_gpu lspci amdgpu && ! check_gpu lshw amdgpu; theninstall_successwarning "No NVIDIA/AMD GPU detected. Ollama will run in CPU-only mode."exit 0
fiif check_gpu lspci amdgpu || check_gpu lshw amdgpu; thenif [ $BUNDLE -ne 0 ]; thenstatus "Downloading Linux ROCm ${ARCH} bundle"curl --fail --show-error --location --progress-bar \"https://ollama.com/download/ollama-linux-${ARCH}-rocm.tgz${VER_PARAM}" | \$SUDO tar -xzf - -C "$OLLAMA_INSTALL_DIR"install_successstatus "AMD GPU ready."exit 0fi# Look for pre-existing ROCm v6 before downloading the dependenciesfor search in "${HIP_PATH:-''}" "${ROCM_PATH:-''}" "/opt/rocm" "/usr/lib64"; doif [ -n "${search}" ] && [ -e "${search}/libhipblas.so.2" -o -e "${search}/lib/libhipblas.so.2" ]; thenstatus "Compatible AMD GPU ROCm library detected at ${search}"install_successexit 0fidonestatus "Downloading AMD GPU dependencies..."$SUDO rm -rf /usr/share/ollama/lib$SUDO chmod o+x /usr/share/ollama$SUDO install -o ollama -g ollama -m 755 -d /usr/share/ollama/lib/rocmcurl --fail --show-error --location --progress-bar "https://ollama.com/download/ollama-linux-amd64-rocm.tgz${VER_PARAM}" \| $SUDO tar zx --owner ollama --group ollama -C /usr/share/ollama/lib/rocm .install_successstatus "AMD GPU ready."exit 0
fiCUDA_REPO_ERR_MSG="NVIDIA GPU detected, but your OS and Architecture are not supported by NVIDIA.  Please install the CUDA driver manually https://docs.nvidia.com/cuda/cuda-installation-guide-linux/"
# ref: https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#rhel-7-centos-7
# ref: https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#rhel-8-rocky-8
# ref: https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#rhel-9-rocky-9
# ref: https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#fedora
install_cuda_driver_yum() {status 'Installing NVIDIA repository...'case $PACKAGE_MANAGER inyum)$SUDO $PACKAGE_MANAGER -y install yum-utilsif curl -I --silent --fail --location "https://developer.download.nvidia.com/compute/cuda/repos/$1$2/$(uname -m | sed -e 's/aarch64/sbsa/')/cuda-$1$2.repo" >/dev/null ; then$SUDO $PACKAGE_MANAGER-config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/$1$2/$(uname -m | sed -e 's/aarch64/sbsa/')/cuda-$1$2.repoelseerror $CUDA_REPO_ERR_MSGfi;;dnf)if curl -I --silent --fail --location "https://developer.download.nvidia.com/compute/cuda/repos/$1$2/$(uname -m | sed -e 's/aarch64/sbsa/')/cuda-$1$2.repo" >/dev/null ; then$SUDO $PACKAGE_MANAGER config-manager --add-repo https://developer.download.nvidia.com/compute/cuda/repos/$1$2/$(uname -m | sed -e 's/aarch64/sbsa/')/cuda-$1$2.repoelseerror $CUDA_REPO_ERR_MSGfi;;esaccase $1 inrhel)status 'Installing EPEL repository...'# EPEL is required for third-party dependencies such as dkms and libvdpau$SUDO $PACKAGE_MANAGER -y install https://dl.fedoraproject.org/pub/epel/epel-release-latest-$2.noarch.rpm || true;;esacstatus 'Installing CUDA driver...'if [ "$1" = 'centos' ] || [ "$1$2" = 'rhel7' ]; then$SUDO $PACKAGE_MANAGER -y install nvidia-driver-latest-dkmsfi$SUDO $PACKAGE_MANAGER -y install cuda-drivers
}# ref: https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#ubuntu
# ref: https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#debian
install_cuda_driver_apt() {status 'Installing NVIDIA repository...'if curl -I --silent --fail --location "https://developer.download.nvidia.com/compute/cuda/repos/$1$2/$(uname -m | sed -e 's/aarch64/sbsa/')/cuda-keyring_1.1-1_all.deb" >/dev/null ; thencurl -fsSL -o $TEMP_DIR/cuda-keyring.deb https://developer.download.nvidia.com/compute/cuda/repos/$1$2/$(uname -m | sed -e 's/aarch64/sbsa/')/cuda-keyring_1.1-1_all.debelseerror $CUDA_REPO_ERR_MSGficase $1 indebian)status 'Enabling contrib sources...'$SUDO sed 's/main/contrib/' < /etc/apt/sources.list | $SUDO tee /etc/apt/sources.list.d/contrib.list > /dev/nullif [ -f "/etc/apt/sources.list.d/debian.sources" ]; then$SUDO sed 's/main/contrib/' < /etc/apt/sources.list.d/debian.sources | $SUDO tee /etc/apt/sources.list.d/contrib.sources > /dev/nullfi;;esacstatus 'Installing CUDA driver...'$SUDO dpkg -i $TEMP_DIR/cuda-keyring.deb$SUDO apt-get update[ -n "$SUDO" ] && SUDO_E="$SUDO -E" || SUDO_E=DEBIAN_FRONTEND=noninteractive $SUDO_E apt-get -y install cuda-drivers -q
}if [ ! -f "/etc/os-release" ]; thenerror "Unknown distribution. Skipping CUDA installation."
fi. /etc/os-releaseOS_NAME=$ID
OS_VERSION=$VERSION_IDPACKAGE_MANAGER=
for PACKAGE_MANAGER in dnf yum apt-get; doif available $PACKAGE_MANAGER; thenbreakfi
doneif [ -z "$PACKAGE_MANAGER" ]; thenerror "Unknown package manager. Skipping CUDA installation."
fiif ! check_gpu nvidia-smi || [ -z "$(nvidia-smi | grep -o "CUDA Version: [0-9]*\.[0-9]*")" ]; thencase $OS_NAME incentos|rhel) install_cuda_driver_yum 'rhel' $(echo $OS_VERSION | cut -d '.' -f 1) ;;rocky) install_cuda_driver_yum 'rhel' $(echo $OS_VERSION | cut -c1) ;;fedora) [ $OS_VERSION -lt '39' ] && install_cuda_driver_yum $OS_NAME $OS_VERSION || install_cuda_driver_yum $OS_NAME '39';;amzn) install_cuda_driver_yum 'fedora' '37' ;;debian) install_cuda_driver_apt $OS_NAME $OS_VERSION ;;ubuntu) install_cuda_driver_apt $OS_NAME $(echo $OS_VERSION | sed 's/\.//') ;;*) exit ;;esac
fiif ! lsmod | grep -q nvidia || ! lsmod | grep -q nvidia_uvm; thenKERNEL_RELEASE="$(uname -r)"case $OS_NAME inrocky) $SUDO $PACKAGE_MANAGER -y install kernel-devel kernel-headers ;;centos|rhel|amzn) $SUDO $PACKAGE_MANAGER -y install kernel-devel-$KERNEL_RELEASE kernel-headers-$KERNEL_RELEASE ;;fedora) $SUDO $PACKAGE_MANAGER -y install kernel-devel-$KERNEL_RELEASE ;;debian|ubuntu) $SUDO apt-get -y install linux-headers-$KERNEL_RELEASE ;;*) exit ;;esacNVIDIA_CUDA_VERSION=$($SUDO dkms status | awk -F: '/added/ { print $1 }')if [ -n "$NVIDIA_CUDA_VERSION" ]; then$SUDO dkms install $NVIDIA_CUDA_VERSIONfiif lsmod | grep -q nouveau; thenstatus 'Reboot to complete NVIDIA CUDA driver install.'exit 0fi$SUDO modprobe nvidia$SUDO modprobe nvidia_uvm
fi# make sure the NVIDIA modules are loaded on boot with nvidia-persistenced
if available nvidia-persistenced; then$SUDO touch /etc/modules-load.d/nvidia.confMODULES="nvidia nvidia-uvm"for MODULE in $MODULES; doif ! grep -qxF "$MODULE" /etc/modules-load.d/nvidia.conf; thenecho "$MODULE" | $SUDO tee -a /etc/modules-load.d/nvidia.conf > /dev/nullfidone
fistatus "NVIDIA GPU ready."
install_success

三、下載ollama安裝包
下載地址：https://github.com/ollama/ollama/releases
具體需要下載哪個可以通過在命令行輸入 lscpu 查看自己的cpu架構

x86_64 CPU 選擇下載 ollama-linux-amd64.tgz aarch64
arm64 CPU 選擇下載 ollama-linux-arm64.tgz

四、運行install.sh腳本進行安裝
進入install.sh所在目錄：

chmod +x install.sh    #給腳本賦予執行權限
./install.sh
# 如果報錯誤： bash: ./build_android.sh：/bin/sh^M：解釋器錯誤: 沒有那個文件或目錄，執行
sed -i 's/\r$//' install.sh

五、配置ollama
需要配置監聽host、下載路徑、多CUDA使用等變量。
運行：

vim /etc/systemd/system/ollama.service

在里面加入你新創建的路徑：

這里host設置主要是為了后面open webui，可能無法訪問。
cuda設置，使用兩塊gpu，主要是優先使用3號gpu。也可以刪除Environment=“CUDA_VISIBLE_DEVICES=3,2這行來默認使用所有gpu或指定所有gpu，如你的服務器有 4 塊 GPU（ID: 0,1,2,3），則Environment=“CUDA_VISIBLE_DEVICES=0,1,2,3。然后添加：

Environment="OLLAMA_PARALLEL=true"

這個參數讓 Ollama 進行 多 GPU 自動并行計算，適用于大模型（如 DeepSeek-R1-32B）。
修改之后保存文件并退出，并重啟ollama服務，運行：

sudo systemctl daemon-reload
sudo systemctl restart ollama

需要注意的是，對于創建的OLLAMA_MODELS路徑文件夾，需要給它777權限，否則可能會啟動失敗：

sudo chmod 777 /XXX

(2)在離線環境下如何配置大模型*

?1)部署完整的deepseek大模型，如：deepseek-r1:32B;
找一個可以上網的主機，執行：

ollama run deepseek-r1:32B

當大模型下載完成后，找到models目錄（應該包含blobs和manifests兩個子文件夾），將整個目錄拷貝到服務器你新創建的路徑下。
重啟ollama，再執行ollama list命令，如果看到有模型了，說明成功。
啟動模型：

ollama run DeepSeek-R1:32B

成功運行后將進入交互式命令行界面。
需要注意的是：默認情況下ollama模型的所在路徑為（但本人嘗試后發現不一定，所以最好在可以上網的主機上安裝ollama時指定一個路徑，那么下載的模型文件就會在你指定的這個路徑下，方便尋找）：

macOS: ~/.ollama/models 
Linux: **/usr/share/ollama/.ollama/models**
Windows: C:Users<username>.ollamamodels

?2)部署蒸餾版模型，如：DeepSeek-R1-Distill-Llama-32B-Q5_K_M.gguf。
一、獲取GGUF格式模型文件，可以在網上下載或通過ModelScope平臺下載量化后的模型文件：

pip install modelscope
modelscope download --model unsloth/DeepSeek-R1-Distill-Llama-70B-GGUF DeepSeek-R1-Distill-Llama-70B-Q5_K_M.gguf --local_dir  /DeepSeek-R1-Distill-Llama-70B-GGUF

二、創建Modelfile配置文件：

vi Modelfile

然后補充配置文件：

# 這里填入下載的gguf文件路徑
FROM /home/DeepSeek-R1-Distill-Llama-32B-GGUF/DeepSeek-R1-Distill-Llama-32B-Q5_K_M.gguf ##替換為你下載的文件路徑和文件# 以下為模型模板配置
TEMPLATE """{{- if .System }}{{ .System }}{{ end }}
{{- range $i, $_ := .Messages }}
{{- $last := eq (len (slice $.Messages $i)) 1}}
{{- if eq .Role "user" }}<｜User｜>{{ .Content }}
{{- else if eq .Role "assistant" }}<｜Assistant｜>{{ .Content }}{{- if not $last }}<｜end▁of▁sentence｜>{{- end }}
{{- end }}
{{- if and $last (ne .Role "assistant") }}<｜Assistant｜>{{- end }}
{{- end }}"""PARAMETER stop "<｜begin▁of▁sentence｜>"
PARAMETER stop "<｜end▁of▁sentence｜>"
PARAMETER stop "<｜User｜>"
PARAMETER stop "<｜Assistant｜>"PARAMETER num_ctx 12800

這里From 模型是必須的，其它的可以自行設置。具體可以參考：https://github.com/ollama/ollama/blob/main/docs/modelfile.md#format

三、模型加載與運行：
導入模型，Modelfile路徑替換為自己前面設置的。

ollama create DeepSeek-R1-Distill-Llama-32B-Q5_K_M -f /home/DeepSeek-R1-Distill-Llama-32B-GGUF/Modelfile

輸入：ollama list如果實現預期輸出，即證明安裝成功。
啟動模型：

ollama run DeepSeek-R1-Distill-Llama-32B-Q5_K_M:latest

成功運行后將進入交互式命令行界面。

目前部署結果是在Linux命令行運行的，交互體驗較差。后續會繼續實現前端可視化界面安裝，提升用戶體驗。

部署過程中參考了以下幾篇博客：

1. linux離線安裝Ollama并完成大模型配置（無網絡）：https://blog.csdn.net/m0_71142057/article/details/143186418
2. ollama 部署 deepseek-r1 70B 模型完整指南：https://zhuanlan.zhihu.com/p/20441592371
3. linux離線部署Ollama+Deepseek r1+open webui ：https://blog.csdn.net/MrIqzd/article/details/145394179