看 Yi 官方文档，一开始摸不着头脑，不知道从哪里入手。 网上找了一些资料，查到了苏洋的博客，先把环境搭建起来。
为了在 Windows11 机器方便使用GPU，以及开源很多工程都提供docker入门，但WSL2慢，考虑本地已经搞了一个WSL1了会不会冲突，同时虚拟机里面也安装不了WSL2，VMWare桌面虚拟机的话直接使用GPU没有很好的方式等等，纠结了一天，最终还是选了安装 WSL2+Docker Desktop。
跟着文章，你将会了解Windows+WLS2+Docker怎么跑GPU模型，以及在国内怎么下载模型文件。
使用WSL2
在 启用或关闭Windows功能 中选择 虚拟机平台。
1
2
wsl --update
wsl --set-default-version 2
然后在微软商店Microsoft Store里面安装 Ubuntu-20.04 （版本选20或者22）的系统（通过应用商店的话就规避了可能安装同一个的Linux的问题：已经安装 在应用商店的按钮不是[获取]是[打开]）。
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
winse@DESKTOP-BR4MG38:~$ cat /etc/os-release
NAME="Ubuntu"
VERSION="20.04.6 LTS (Focal Fossa)"
ID=ubuntu
ID_LIKE=debian
PRETTY_NAME="Ubuntu 20.04.6 LTS"
VERSION_ID="20.04"
HOME_URL="https://www.ubuntu.com/"
SUPPORT_URL="https://help.ubuntu.com/"
BUG_REPORT_URL="https://bugs.launchpad.net/ubuntu/"
PRIVACY_POLICY_URL="https://www.ubuntu.com/legal/terms-and-policies/privacy-policy"
VERSION_CODENAME=focal
UBUNTU_CODENAME=focal
winse@DESKTOP-BR4MG38:~$ uname -a
Linux DESKTOP-BR4MG38 5.15.133.1-microsoft-standard-WSL2 #1 SMP Thu Oct 5 21:02:42 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux
对比WSL1，WSL2的 ip a ，WSL2还是干净很多，把宿主机的一些信息合并到linux里面了（如：hosts）。
Docker Desktop + WSL2
通过exe安装，安装过程中选择使用WSL2，装好后wsl显示多出了两个linux。
1
2
3
4
5
6
C:\Users\P15>wsl -l -v
NAME STATE VERSION
* Ubuntu Stopped 1
Ubuntu-20.04 Running 2
docker-desktop-data Running 2
docker-desktop Running 2
1
2
3
4
winse@DESKTOP-BR4MG38:~$ su -
root@DESKTOP-BR4MG38:~# echo "winse ALL=(ALL:ALL) NOPASSWD: ALL" >>/etc/sudoers
root@DESKTOP-BR4MG38:~# sed -i.bak -e 's|archive.ubuntu.com/ubuntu/|mirrors.aliyun.com/ubuntu/|' -e 's|security.ubuntu.com/ubuntu/|mirrors.aliyun.com/ubuntu/|' /etc/apt/sources.list
在 WSL-Ubuntu 里面可以直接用 Win11 的程序，直接查看docker的信息：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
winse@DESKTOP-BR4MG38:~$ docker version
Client: Docker Engine - Community
Cloud integration: v1.0.35+desktop.5
Version: 24.0.7
API version: 1.43
Go version: go1.20.10
Git commit: afdd53b
Built: Thu Oct 26 09:08:17 2023
OS/Arch: linux/amd64
Context: default
Server: Docker Desktop
Engine:
Version: 24.0.7
API version: 1.43 (minimum version 1.12)
Go version: go1.20.10
Git commit: 311b9ff
Built: Thu Oct 26 09:08:02 2023
OS/Arch: linux/amd64
Experimental: false
containerd:
Version: 1.6.25
GitCommit: d8f198a4ed8892c764191ef7b3b06d8a2eeb5c7f
runc:
Version: 1.1.10
GitCommit: v1.1.10-0-g18a0cb0
docker-init:
Version: 0.19.0
GitCommit: de40ad0
winse@DESKTOP-BR4MG38:~$ which docker
/usr/bin/docker
winse@DESKTOP-BR4MG38:~$ ll /usr/bin/docker
lrwxrwxrwx 1 root root 48 Jan 13 11:03 /usr/bin/docker -> /mnt/wsl/docker-desktop/cli-tools/usr/bin/docker*
其实用的就是windows的docker
镜像加速
保存会重启docker，再查看docker的信息，确认Registry Mirrors：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
winse@DESKTOP-BR4MG38:~$ docker info
Client: Docker Engine - Community
Version: 24.0.7
Context: default
Debug Mode: false
Plugins:
buildx: Docker Buildx (Docker Inc.)
Version: v0.12.0-desktop.2
Path: /usr/local/lib/docker/cli-plugins/docker-buildx
compose: Docker Compose (Docker Inc.)
Version: v2.23.3-desktop.2
Path: /usr/local/lib/docker/cli-plugins/docker-compose
dev: Docker Dev Environments (Docker Inc.)
Version: v0.1.0
Path: /usr/local/lib/docker/cli-plugins/docker-dev
extension: Manages Docker extensions (Docker Inc.)
Version: v0.2.21
Path: /usr/local/lib/docker/cli-plugins/docker-extension
feedback: Provide feedback, right in your terminal! (Docker Inc.)
Version: 0.1
Path: /usr/local/lib/docker/cli-plugins/docker-feedback
init: Creates Docker-related starter files for your project (Docker Inc.)
Version: v0.1.0-beta.10
Path: /usr/local/lib/docker/cli-plugins/docker-init
sbom: View the packaged-based Software Bill Of Materials (SBOM) for an image (Anchore Inc.)
Version: 0.6.0
Path: /usr/local/lib/docker/cli-plugins/docker-sbom
scan: Docker Scan (Docker Inc.)
Version: v0.26.0
Path: /usr/local/lib/docker/cli-plugins/docker-scan
scout: Docker Scout (Docker Inc.)
Version: v1.2.0
Path: /usr/local/lib/docker/cli-plugins/docker-scout
Server:
Containers: 1
Running: 1
Paused: 0
Stopped: 0
Images: 5
Server Version: 24.0.7
Storage Driver: overlay2
Backing Filesystem: extfs
Supports d_type: true
Using metacopy: false
Native Overlay Diff: true
userxattr: false
Logging Driver: json-file
Cgroup Driver: cgroupfs
Cgroup Version: 1
Plugins:
Volume: local
Network: bridge host ipvlan macvlan null overlay
Log: awslogs fluentd gcplogs gelf journald json-file local logentries splunk syslog
Swarm: inactive
Runtimes: io.containerd.runc.v2 runc
Default Runtime: runc
Init Binary: docker-init
containerd version: d8f198a4ed8892c764191ef7b3b06d8a2eeb5c7f
runc version: v1.1.10-0-g18a0cb0
init version: de40ad0
Security Options:
seccomp
Profile: unconfined
Kernel Version: 5.15.133.1-microsoft-standard-WSL2
Operating System: Docker Desktop
OSType: linux
Architecture: x86_64
CPUs: 16
Total Memory: 31.26GiB
Name: docker-desktop
ID: 340fee1c-e22a-485c-a973-f0e26d7535c9
Docker Root Dir: /var/lib/docker
Debug Mode: false
HTTP Proxy: http.docker.internal:3128
HTTPS Proxy: http.docker.internal:3128
No Proxy: hubproxy.docker.internal
Experimental: false
Insecure Registries:
hubproxy.docker.internal:5555
127.0.0.0/8
Registry Mirrors:
https://us69kjun.mirror.aliyuncs.com/
https://docker.mirrors.ustc.edu.cn/
https://hub-mirror.c.163.com/
https://mirror.baidubce.com/
Live Restore Enabled: false
WARNING: No blkio throttle.read_bps_device support
WARNING: No blkio throttle.write_bps_device support
WARNING: No blkio throttle.read_iops_device support
WARNING: No blkio throttle.write_iops_device support
WARNING: daemon is not using the default seccomp profile
GPU
Driver
根据Win11机器的显卡安装最新版本驱动（不要在WSL中安装任何Linux版的Nvidia驱动！）
https://www.nvidia.com/Download/index.aspx
输入nvidia-smi，查验是否安装成功。WSL2里面啥都不用做，在WSL2命令行直接就能查看nvidia-smi。
启动docker也能一样查看
1
winse@DESKTOP-BR4MG38:stable-diffusion-taiyi$ docker run -it --rm --gpus all ubuntu nvidia-smi
其实这个启动的container也是一个WSL2。注意：WSL中不需要安装任何Linux版的Nvidia驱动！
验证 WLS2中Docker跑起来的容器 是否能够正常调用GPU：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
winse@DESKTOP-BR4MG38:~$ docker pull nvcr.io/nvidia/pytorch:23.07-py3
23.07-py3: Pulling from nvidia/pytorch
3153aa388d02: Pulling fs layer
...
ee3f0ae6e80f: Pull complete
d4528227b5b8: Pull complete
Digest: sha256:c53e8702a4ccb3f55235226dab29ef5d931a2a6d4d003ab47ca2e7e670f7922b
Status: Downloaded newer image for nvcr.io/nvidia/pytorch:23.07-py3
nvcr.io/nvidia/pytorch:23.07-py3
What's Next?
1. Sign in to your Docker account → docker login
2. View a summary of image vulnerabilities and recommendations → docker scout quickview nvcr.io/nvidia/pytorch:23.07-py3
winse@DESKTOP-BR4MG38:~$ docker run -it --gpus=all --rm nvcr.io/nvidia/pytorch:23.07-py3 nvidia-smi
=============
== PyTorch ==
=============
NVIDIA Release 23.07 (build 63867923)
PyTorch Version 2.1.0a0+b5021ba
Container image Copyright (c) 2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
Copyright (c) 2014-2023 Facebook Inc.
Copyright (c) 2011-2014 Idiap Research Institute (Ronan Collobert)
Copyright (c) 2012-2014 Deepmind Technologies (Koray Kavukcuoglu)
Copyright (c) 2011-2012 NEC Laboratories America (Koray Kavukcuoglu)
Copyright (c) 2011-2013 NYU (Clement Farabet)
Copyright (c) 2006-2010 NEC Laboratories America (Ronan Collobert, Leon Bottou, Iain Melvin, Jason Weston)
Copyright (c) 2006 Idiap Research Institute (Samy Bengio)
Copyright (c) 2001-2004 Idiap Research Institute (Ronan Collobert, Samy Bengio, Johnny Mariethoz)
Copyright (c) 2015 Google Inc.
Copyright (c) 2015 Yangqing Jia
Copyright (c) 2013-2016 The Caffe contributors
All rights reserved.
Various files include modifications (c) NVIDIA CORPORATION & AFFILIATES. All rights reserved.
This container image and its contents are governed by the NVIDIA Deep Learning Container License.
By pulling and using the container, you accept the terms and conditions of this license:
https://developer.nvidia.com/ngc/nvidia-deep-learning-container-license
NOTE: The SHMEM allocation limit is set to the default of 64MB. This may be
insufficient for PyTorch. NVIDIA recommends the use of the following flags:
docker run --gpus all --ipc=host --ulimit memlock=-1 --ulimit stack=67108864 ...
Sat Jan 13 14:01:37 2024
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.146.01 Driver Version: 537.99 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 Quadro T2000 On | 00000000:01:00.0 On | N/A |
| N/A 43C P8 6W / 60W | 856MiB / 4096MiB | 9% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| 0 N/A N/A 27 G /Xwayland N/A |
| 0 N/A N/A 41 G /Xwayland N/A |
| 0 N/A N/A 42 G /Xwayland N/A |
+---------------------------------------------------------------------------------------+
winse@DESKTOP-BR4MG38:~$ docker run --rm --gpus all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
Unable to find image 'nvcr.io/nvidia/k8s/cuda-sample:nbody' locally
nbody: Pulling from nvidia/k8s/cuda-sample
22c5ef60a68e: Pull complete
1939e4248814: Pull complete
548afb82c856: Pull complete
a424d45fd86f: Pull complete
207b64ab7ce6: Pull complete
f65423f1b49b: Pull complete
2b60900a3ea5: Pull complete
e9bff09d04df: Pull complete
edc14edf1b04: Pull complete
1f37f461c076: Pull complete
9026fb14bf88: Pull complete
Digest: sha256:59261e419d6d48a772aad5bb213f9f1588fcdb042b115ceb7166c89a51f03363
Status: Downloaded newer image for nvcr.io/nvidia/k8s/cuda-sample:nbody
Run "nbody -benchmark [-numbodies=<numBodies>]" to measure performance.
-fullscreen (run n-body simulation in fullscreen mode)
-fp64 (use double precision floating point values for simulation)
-hostmem (stores simulation data in host memory)
-benchmark (run benchmark to measure performance)
-numbodies=<N> (number of bodies (>= 1) to run in simulation)
-device=<d> (where d=0,1,2.... for the CUDA device to use)
-numdevices=<i> (where i=(number of CUDA devices > 0) to use for simulation)
-compare (compares simulation results running once on the default GPU and once on the CPU)
-cpu (run n-body simulation on the CPU)
-tipsy=<file.bin> (load a tipsy model file for simulation)
NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.
> Windowed mode
> Simulation data stored in video memory
> Single precision floating point simulation
> 1 Devices used for simulation
GPU Device 0: "Turing" with compute capability 7.5
> Compute 7.5 CUDA device: [Quadro T2000]
16384 bodies, total time for 10 iterations: 64.071 ms
= 41.897 billion interactions per second
= 837.937 single-precision GFLOP/s at 20 flops per interaction
#再跑一遍
winse@DESKTOP-BR4MG38:~$ docker run --rm --gpus all nvcr.io/nvidia/k8s/cuda-sample:nbody nbody -gpu -benchmark
Run "nbody -benchmark [-numbodies=<numBodies>]" to measure performance.
-fullscreen (run n-body simulation in fullscreen mode)
-fp64 (use double precision floating point values for simulation)
-hostmem (stores simulation data in host memory)
-benchmark (run benchmark to measure performance)
-numbodies=<N> (number of bodies (>= 1) to run in simulation)
-device=<d> (where d=0,1,2.... for the CUDA device to use)
-numdevices=<i> (where i=(number of CUDA devices > 0) to use for simulation)
-compare (compares simulation results running once on the default GPU and once on the CPU)
-cpu (run n-body simulation on the CPU)
-tipsy=<file.bin> (load a tipsy model file for simulation)
NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.
> Windowed mode
> Simulation data stored in video memory
> Single precision floating point simulation
> 1 Devices used for simulation
GPU Device 0: "Turing" with compute capability 7.5
> Compute 7.5 CUDA device: [Quadro T2000]
16384 bodies, total time for 10 iterations: 23.398 ms
= 114.724 billion interactions per second
= 2294.490 single-precision GFLOP/s at 20 flops per interaction
WSL2 cuda-toolkit
开发环境/运行环境
* https://zhuanlan.zhihu.com/p/555151725
* https://docs.nvidia.com/cuda/wsl-user-guide/index.html#cuda-support-for-WSL2
* https://developer.nvidia.com/cuda-downloads?target_os=Linux&target_arch=x86_64&Distribution=WSL-Ubuntu&target_version=2.0&target_type=deb_network
1
2
3
4
wget https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-keyring_1.1-1_all.deb
sudo dpkg -i cuda-keyring_1.1-1_all.deb
sudo apt-get update
sudo apt-get -y install cuda-toolkit-12-3
运行安装：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
(demo_env) winse@DESKTOP-BR4MG38:ai$ wget https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-keyring_1.1-1_all.deb
--2024-01-14 23:53:22-- https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-keyring_1.1-1_all.deb
Resolving developer.download.nvidia.com (developer.download.nvidia.com)... 152.199.39.144, 72.21.80.5, 72.21.80.6, ...
Connecting to developer.download.nvidia.com (developer.download.nvidia.com)|152.199.39.144|:443... connected.
HTTP request sent, awaiting response... 301 Moved Permanently
Location: https://developer.download.nvidia.cn/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-keyring_1.1-1_all.deb [following]
--2024-01-14 23:53:23-- https://developer.download.nvidia.cn/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-keyring_1.1-1_all.deb
Resolving developer.download.nvidia.cn (developer.download.nvidia.cn)... 59.36.216.26, 59.36.216.27, 175.4.58.180, ...
Connecting to developer.download.nvidia.cn (developer.download.nvidia.cn)|59.36.216.26|:443... connected.
HTTP request sent, awaiting response... 200 OK
Length: 4328 (4.2K) [application/x-deb]
Saving to: ‘cuda-keyring_1.1-1_all.deb’
cuda-keyring_1.1-1_all.deb 100%[====================================================>] 4.23K --.-KB/s in 0s
2024-01-14 23:53:23 (1.61 GB/s) - ‘cuda-keyring_1.1-1_all.deb’ saved [4328/4328]
(demo_env) winse@DESKTOP-BR4MG38:ai$
(demo_env) winse@DESKTOP-BR4MG38:ai$ sudo dpkg -i cuda-keyring_1.1-1_all.deb
(demo_env) winse@DESKTOP-BR4MG38:ai$ sudo apt-get update
(demo_env) winse@DESKTOP-BR4MG38:ai$ sudo apt-get -y install cuda-toolkit-12-3
1
2
3
(base) winse@DESKTOP-BR4MG38:~$ vi .bashrc
export PATH=/usr/local/cuda/bin:$PATH
新打开一个shell：
1
2
3
4
5
6
(base) winse@DESKTOP-BR4MG38:~$ nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2023 NVIDIA Corporation
Built on Wed_Nov_22_10:17:15_PST_2023
Cuda compilation tools, release 12.3, V12.3.107
Build cuda_12.3.r12.3/compiler.33567101_0
cuDNN
https://developer.nvidia.com/cudnn
NVIDIA CUDA® Deep Neural Network library 支持神经网络的推理。
注册下载对应CUDA的版本 https://developer.nvidia.com/rdp/cudnn-download
注意：如果不在WSL2-Ubuntu中直接使用cuDNN，后续通过容器直接拉取包含cuDNN的容器，就可以省略这一部分。
https://docs.nvidia.com/deeplearning/cudnn/install-guide/index.html#installlinux-deb
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
#sudo dpkg -i cudnn-local-repo-ubuntu2004-8.9.6.50_1.0-1_amd64.deb
#sudo dpkg -r cudnn-local-repo-ubuntu2004-8.9.6.50
#sudo rm /etc/apt/sources.list.d/cudnn-local-ubuntu2004-8.9.6.50.list
(base) winse@DESKTOP-BR4MG38:i$ sudo dpkg -i cudnn-local-repo-ubuntu2004-8.9.7.29_1.0-1_amd64.deb
(base) winse@DESKTOP-BR4MG38:i$ sudo cp /var/cudnn-local-repo-ubuntu2004-8.9.7.29/cudnn-local-30472A84-keyring.gpg /usr/share/keyrings/
(base) winse@DESKTOP-BR4MG38:i$ sudo apt install zlib1g
(base) winse@DESKTOP-BR4MG38:i$ sudo apt update
(base) winse@DESKTOP-BR4MG38:i$ apt search libcudnn8
Sorting... Done
Full Text Search... Done
libcudnn8/unknown 8.9.7.29-1+cuda12.2 amd64
cuDNN runtime libraries
libcudnn8-dev/unknown 8.9.7.29-1+cuda12.2 amd64
cuDNN development libraries and headers
libcudnn8-samples/unknown 8.9.7.29-1+cuda12.2 amd64
cuDNN samples
(base) winse@DESKTOP-BR4MG38:i$ sudo apt install libcudnn8 libcudnn8-dev libcudnn8-samples
校验是否安装成功
https://docs.nvidia.com/deeplearning/cudnn/install-guide/index.html#verify
运行报错参考 https://forums.developer.nvidia.com/t/freeimage-is-not-set-up-correctly-please-ensure-freeimae-is-set-up-correctly/66950
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
(base) winse@DESKTOP-BR4MG38:i$ cp -r /usr/src/cudnn_samples_v8 ./
(base) winse@DESKTOP-BR4MG38:i$ cd cudnn_samples_v8/mnistCUDNN/
(base) winse@DESKTOP-BR4MG38:mnistCUDNN$
(base) winse@DESKTOP-BR4MG38:mnistCUDNN$ sudo apt-get install libfreeimage3 libfreeimage-dev
(base) winse@DESKTOP-BR4MG38:mnistCUDNN$ make clean && make
(base) winse@DESKTOP-BR4MG38:mnistCUDNN$ ./mnistCUDNN
Executing: mnistCUDNN
cudnnGetVersion() : 8907 , CUDNN_VERSION from cudnn.h : 8907 (8.9.7)
Host compiler version : GCC 9.4.0
There are 1 CUDA capable devices on your machine :
device 0 : sms 16 Capabilities 7.5, SmClock 1785.0 Mhz, MemSize (Mb) 4095, MemClock 6001.0 Mhz, Ecc=0, boardGroupID=0
Using device 0
Testing single precision
...
Result of classification: 1 3 5
Test passed!
Testing half precision (math in single precision)
...
Result of classification: 1 3 5
Test passed!
nvidia-container-toolkit???
Docker Desktop + WSL2不用安装 nvidia-container-toolkit ???
WSL2 Python - conda
下载安装conda
下载miniconda
https://docs.conda.io/projects/miniconda/en/latest/
1
2
3
4
5
6
mkdir -p ~/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
rm -rf ~/miniconda3/miniconda.sh
~/miniconda3/bin/conda init bash
运行脚本安装：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
winse@DESKTOP-BR4MG38:ai$ mkdir miniconda3
winse@DESKTOP-BR4MG38:ai$ cd miniconda3/
winse@DESKTOP-BR4MG38:miniconda3$
winse@DESKTOP-BR4MG38:miniconda3$ bash miniconda.sh -b -u -p ~/miniconda3
PREFIX=/home/winse/miniconda3
Unpacking payload ...
Installing base environment...
Downloading and Extracting Packages:
Downloading and Extracting Packages:
Preparing transaction: done
Executing transaction: done
installation finished.
winse@DESKTOP-BR4MG38:miniconda3$ ~/miniconda3/bin/conda init bash
no change /home/winse/miniconda3/condabin/conda
no change /home/winse/miniconda3/bin/conda
no change /home/winse/miniconda3/bin/conda-env
no change /home/winse/miniconda3/bin/activate
no change /home/winse/miniconda3/bin/deactivate
no change /home/winse/miniconda3/etc/profile.d/conda.sh
no change /home/winse/miniconda3/etc/fish/conf.d/conda.fish
no change /home/winse/miniconda3/shell/condabin/Conda.psm1
no change /home/winse/miniconda3/shell/condabin/conda-hook.ps1
no change /home/winse/miniconda3/lib/python3.11/site-packages/xontrib/conda.xsh
no change /home/winse/miniconda3/etc/profile.d/conda.csh
modified /home/winse/.bashrc
==> For changes to take effect, close and re-open your current shell. <==
添加源：
1
2
3
4
5
6
7
8
9
10
11
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/cloud/conda-forge/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/msys2/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/bioconda/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/menpo/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/pytorch/
conda config --set show_channel_urls yes
conda config --show channels
#conda config --remove-key channels
运行测试GPU
1
2
3
4
5
6
7
8
9
10
conda create -n demo_env python=3.8
conda activate demo_env
conda install pytorch==1.6.0 cudatoolkit=10.1 torchaudio=0.6.0 -c pytorch
#conda list
#conda deactivate
#conda env list
#conda remove -n demo_env --all
#conda env remove --name old_name
验证是否安装成功
在我们到demo_env环境下，打开Python，输入以下语句：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
(demo_env) winse@DESKTOP-BR4MG38:ai$ python
Python 3.8.18 | packaged by conda-forge | (default, Dec 23 2023, 17:21:28)
[GCC 12.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> x = torch.rand(5,3)
>>> print(x)
tensor([[0.4343, 0.3966, 0.1862],
[0.1502, 0.0788, 0.7713],
[0.3505, 0.7065, 0.9952],
[0.6420, 0.2574, 0.7550],
[0.8292, 0.7714, 0.9014]])
>>> print(torch.cuda.is_available())
True
模型下载
非常重要，不然时间都浪费等待下载上了。模型动辄几G，稍微大一点的就几十G，是需要慎重和反复探索。
一开始用代理和GIT下载的，又慢又浪费时间又浪费空间。放着下了一晚，早起起来磁盘空间不够 o(╥﹏╥)o 。
参考 https://soulteary.com/2024/01/09/summary-of-reliable-download-solutions-for-ai-models.html
国内的modelscope下载
modelscope它还结合了aliyun提供了一定时长的免费环境，在本地折腾折腾后再上去跑跑。这里只通过它去下载模型（下载的方式是没有.git的文件，少占一半多的磁盘空间）。
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
(base) winse@DESKTOP-BR4MG38:ai$ conda activate demo
(demo) winse@DESKTOP-BR4MG38:ai$ pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
(demo) winse@DESKTOP-BR4MG38:ai$ pip install modelscope
(demo) winse@DESKTOP-BR4MG38:ai$ python -c "from modelscope.hub.snapshot_download import snapshot_download;snapshot_download('damo/nlp_xlmr_named-entity-recognition_viet-ecommerce-title', cache_dir='./')"
2024-01-14 16:52:36,017 - modelscope - INFO - PyTorch version 1.11.0+cu113 Found.
2024-01-14 16:52:36,018 - modelscope - INFO - Loading ast index from /home/winse/.cache/modelscope/ast_indexer
2024-01-14 16:52:36,050 - modelscope - INFO - Loading done! Current index file version is 1.11.0, with md5 85336421feb1dc1ec9dde85ceee20f42 and a total number of 953 components indexed
2024-01-14 16:52:36,757 - modelscope - WARNING - Model revision not specified, use revision: v1.0.0
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1.19k/1.19k [00:00<00:00, 12.3MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 238/238 [00:00<00:00, 1.97MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 238/238 [00:00<00:00, 1.81MB/s]
Downloading: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████▉| 1.04G/1.04G [00:36<00:00, 30.1MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2.68k/2.68k [00:00<00:00, 18.2MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 2.68k/2.68k [00:00<00:00, 27.7MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 4.83M/4.83M [00:00<00:00, 7.97MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 150/150 [00:00<00:00, 1.58MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 8.68M/8.68M [00:00<00:00, 11.3MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 470/470 [00:00<00:00, 4.53MB/s]
(demo) winse@DESKTOP-BR4MG38:ai$ python -c "from modelscope.hub.snapshot_download import snapshot_download;snapshot_download('Fengshenbang/Taiyi-Stable-Diffusion-1B-Chinese-v0.1', cache_dir='./')"
2024-01-15 08:19:52,588 - modelscope - INFO - PyTorch version 1.11.0+cu113 Found.
2024-01-15 08:19:52,589 - modelscope - INFO - Loading ast index from /home/winse/.cache/modelscope/ast_indexer
2024-01-15 08:19:52,741 - modelscope - INFO - Loading done! Current index file version is 1.11.0, with md5 85336421feb1dc1ec9dde85ceee20f42 and a total number of 953 components indexed
2024-01-15 08:19:53,874 - modelscope - WARNING - Model revision not specified, use revision: v1.0.0
Downloading: 100%|████████████████████████████████████████████████████████████████| 257k/257k [00:00<00:00, 1.48MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████| 600/600 [00:00<00:00, 488kB/s]
Downloading: 100%|██████████████████████████████████████████████████████████████████| 793/793 [00:00<00:00, 1.35MB/s]
Downloading: 100%|██████████████████████████████████████████████████████████████████| 884/884 [00:00<00:00, 1.47MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████| 4.56k/4.56k [00:00<00:00, 220kB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████| 146/146 [00:00<00:00, 251kB/s]
Downloading: 100%|█████████████████████████████████████████████████████████████▉| 3.20G/3.20G [01:29<00:00, 38.5MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████| 319M/319M [00:39<00:00, 8.39MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████| 571k/571k [00:00<00:00, 2.35MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████| 583k/583k [00:00<00:00, 2.16MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████| 571k/571k [00:00<00:00, 2.10MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████| 539/539 [00:00<00:00, 924kB/s]
Downloading: 100%|█████████████████████████████████████████████████████████████████| 196k/196k [00:00<00:00, 998kB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████| 226k/226k [00:00<00:00, 1.24MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████| 342/342 [00:00<00:00, 617kB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████| 390M/390M [00:30<00:00, 13.5MB/s]
Downloading: 100%|█████████████████████████████████████████████████████████████▉| 1.13G/1.13G [00:41<00:00, 29.3MB/s]
Downloading: 100%|██████████████████████████████████████████████████████████████| 8.90k/8.90k [00:00<00:00, 3.09MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████| 298/298 [00:00<00:00, 482kB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████| 186/186 [00:00<00:00, 304kB/s]
Downloading: 100%|█████████████████████████████████████████████████████████████▉| 3.89G/3.89G [02:12<00:00, 31.6MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████| 477k/477k [00:00<00:00, 2.02MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████| 477k/477k [00:00<00:00, 1.72MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████| 198k/198k [00:00<00:00, 1.01MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████| 212k/212k [00:00<00:00, 1.10MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████| 555/555 [00:00<00:00, 933kB/s]
Downloading: 100%|█████████████████████████████████████████████████████████████████| 107k/107k [00:00<00:00, 782kB/s]
对比一下git和直接下载空间的占用，时间就更加不用说了。
modelscope
把整个流程跑一下，跑个简单的例子：
环境安装
https://modelscope.cn/docs/%E7%8E%AF%E5%A2%83%E5%AE%89%E8%A3%85
运行时依赖
1
2
3
4
5
6
7
(demo) winse@DESKTOP-BR4MG38:ai$ pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
#pip config set global.index-url https://mirrors.cloud.aliyuncs.com/pypi/simple
#pip config set install.trusted-host mirrors.cloud.aliyuncs.com
(demo) winse@DESKTOP-BR4MG38:ai$ pip3 install torch==1.11.0 torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113
$ pip install transformers sentencepiece pyvi
测试模型：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
https://modelscope.cn/models/damo/nlp_xlmr_named-entity-recognition_viet-ecommerce-title/summary
(demo) winse@DESKTOP-BR4MG38:ai$ python
Python 3.8.18 | packaged by conda-forge | (default, Dec 23 2023, 17:21:28)
[GCC 12.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from modelscope.pipelines import pipeline
2024-01-15 01:22:20,476 - modelscope - INFO - PyTorch version 1.11.0+cu113 Found.
2024-01-15 01:22:20,478 - modelscope - INFO - Loading ast index from /home/winse/.cache/modelscope/ast_indexer
2024-01-15 01:22:20,498 - modelscope - INFO - Loading done! Current index file version is 1.11.0, with md5 85336421feb1dc1ec9dde85ceee20f42 and a total number of 953 components indexed
>>> from modelscope.utils.constant import Tasks
>>> ner_pipeline = pipeline(Tasks.named_entity_recognition, 'damo/nlp_xlmr_named-entity-recognition_viet-ecommerce-title', model_revision='v1.0.1')
2024-01-15 01:22:28,618 - modelscope - INFO - initiate model from damo/nlp_xlmr_named-entity-recognition_viet-ecommerce-title
2024-01-15 01:22:28,620 - modelscope - INFO - initiate model from location damo/nlp_xlmr_named-entity-recognition_viet-ecommerce-title.
2024-01-15 01:22:28,630 - modelscope - INFO - initialize model from damo/nlp_xlmr_named-entity-recognition_viet-ecommerce-title
2024-01-15 01:22:30,945 - modelscope - INFO - head has no _keys_to_ignore_on_load_missing
2024-01-15 01:22:34,599 - modelscope - INFO - All model checkpoint weights were used when initializing ModelForTokenClassificationWithCRF.
2024-01-15 01:22:34,599 - modelscope - INFO - All the weights of ModelForTokenClassificationWithCRF were initialized from the model checkpoint If your task is similar to the task the model of the checkpoint was trained on, you can already use ModelForTokenClassificationWithCRF for predictions without further training.
>>> result = ner_pipeline('Nón vành dễ thương cho bé gái')
>>> print(result)
{'output': [{'type': 'product', 'start': 0, 'end': 8, 'prob': 0.98140895, 'span': 'Nón vành'}, {'type': 'style', 'start': 9, 'end': 18, 'prob': 0.99752563, 'span': 'dễ thương'}, {'type': 'consumer_group', 'start': 23, 'end': 29, 'prob': 0.99895895, 'span': 'bé gái'}]}
>>>
TODO
国外的，后续用到了再补
当然，如果进场要用到各种工具的特定版本来下载依赖，用一个docker镜像来作为下载器，也是不错的方法
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
#Python轻量环境
docker pull python:3.10-slim
@1
#将本地目录挂载到容器里，一会作为模型下载目录使用
docker run --rm -it -v `pwd`:/models python:3.10-slim bash
sed -i 's/snapshot.debian.org/mirrors.tuna.tsinghua.edu.cn/g' /etc/apt/sources.list.d/debian.sources
pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
cd /models
@2
#创建一个持续运行的 Python 容器
docker run -d --name=downloader -v `pwd`:/models python:3.10-slim tail -f /etc/hosts
#使用命令进入容器进行配置和下载模型
docker exec -it downloader bash
太乙模型
[尝试/试错]
开始是参照 使用 Docker 来快速上手中文 Stable Diffusion 模型：太乙 文章里面说的蛮简单的，想着我这个WSL2+GPU应该也是可以的。
开始的时刻还是git clone的，等到怕了，几个小时还不一定成功，后面才改成下载的方式！
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
#git clone https://huggingface.co/IDEA-CCNL/Taiyi-Stable-Diffusion-1B-Chinese-v0.1
winse@DESKTOP-BR4MG38:/mnt/i/ai/stable-diffusion-taiyi$ git clone https://www.modelscope.cn/Fengshenbang/Taiyi-Stable-Diffusion-1B-Chinese-v0.1.git
Cloning into 'Taiyi-Stable-Diffusion-1B-Chinese-v0.1'...
remote: Enumerating objects: 85, done.
remote: Counting objects: 100% (6/6), done.
remote: Compressing objects: 100% (6/6), done.
remote: Total 85 (delta 2), reused 0 (delta 0), pack-reused 79
Unpacking objects: 100% (85/85), 3.61 GiB | 1.98 MiB/s, done.
Filtering content: 100% (5/5), 8.92 GiB | 1.71 MiB/s, done.
#镜像也下载了好几次 才pull下来
winse@DESKTOP-BR4MG38:/mnt/i/ai$ docker pull soulteary/stable-diffusion:taiyi-0.1
taiyi-0.1: Pulling from soulteary/stable-diffusion
a404e5416296: Pull complete
af6d12d8d61a: Pull complete
bc57d500b85c: Pull complete
fcd60060414d: Pull complete
65b27d733eb0: Pull complete
266c4315d44f: Pull complete
7ed4190451a3: Pull complete
975671c72e25: Pull complete
213ba1e17e15: Pull complete
37bbbc68318a: Pull complete
80438d07027f: Pull complete
74c79bc62d3a: Pull complete
f8054e9907fb: Pull complete
dc8d44bb4941: Pull complete
625444b7a83c: Pull complete
0b90667ff465: Pull complete
67d73c5193e1: Pull complete
Digest: sha256:69cc4b5fc890dd7ccffff9dbfc2eb2262a0a727574b8beeeafe621f9ef135d16
Status: Downloaded newer image for soulteary/stable-diffusion:taiyi-0.1
docker.io/soulteary/stable-diffusion:taiyi-0.1
What's Next?
View a summary of image vulnerabilities and recommendations → docker scout quickview soulteary/stable-diffusion:taiyi-0.1
#这也是得在纯Linux机器上的Docker才行的
#wget https://github.com/soulteary/docker-stable-diffusion-taiyi/blob/main/docker-compose.yml
#我这就直接运行
winse@DESKTOP-BR4MG38:stable-diffusion-taiyi$ docker run --gpus all --rm -it -v $(pwd)/Taiyi-Stable-Diffusion-1B-Chinese-v0.1:/stable-diffusion-webui/models/Taiyi-Stable-Diffusion-1B-Chinese-v0.1 -p 7860:7860 soulteary/stable-diffusion:taiyi-0.1
#Windows cmd
#docker run --gpus all --ipc=host --ulimit memlock=-1 --ulimit stack=67108864 --rm -it -v C:/docker-sdxl/stabilityai/:/app/stabilityai -p 7860:7860 soulteary/sdxl:runtime
运行起来后访问 http://localhost:7860/ ，输入 小船，河流，星空，星星，山峦，油画 查看Win11的任务管理器，GPU是打满了的，但生成的图片是全黑，啥都没有！
WSL-Ubuntu部署
试了很多次都不行，最后还是回到原点，不能偷懒，先把效果跑出来：
安装依赖（通过代理）
1
2
3
4
5
6
7
8
9
10
11
#https://stackoverflow.com/questions/37776228/pycharm-python-opencv-and-cv2-install-error
(demo) winse@DESKTOP-BR4MG38:~$ pip3 install opencv-python
$ pip install diffusers
##pip install accelerate
(demo) winse@DESKTOP-BR4MG38:~$ unset all_proxy && unset ALL_PROXY
(demo) winse@DESKTOP-BR4MG38:~$ pip install pysocks
(demo) winse@DESKTOP-BR4MG38:~$ export ALL_PROXY=socks5://172.22.240.1:23333 HTTPS_PROXY=socks5://172.22.240.1:23333 HTTP_PROXY=socks5://172.22.240.1:23333
(demo) winse@DESKTOP-BR4MG38:~$ pip install git+https://github.com/huggingface/accelerate
测试（跑了一个小时，-_-||）（此时还没安装cuDNN）：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
(demo) winse@DESKTOP-BR4MG38:ai$ python
Python 3.8.18 | packaged by conda-forge | (default, Dec 23 2023, 17:21:28)
[GCC 12.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from modelscope.utils.constant import Tasks
2024-01-15 10:07:41,619 - modelscope - INFO - PyTorch version 1.11.0+cu113 Found.
2024-01-15 10:07:41,630 - modelscope - INFO - Loading ast index from /home/winse/.cache/modelscope/ast_indexer
2024-01-15 10:07:41,751 - modelscope - INFO - Loading done! Current index file version is 1.11.0, with md5 85336421feb1dc1ec9dde85ceee20f42 and a total number of 953 components indexed
>>> from modelscope.pipelines import pipeline
>>> import cv2
>>> pipe = pipeline(task=Tasks.text_to_image_synthesis, model='Fengshenbang/Taiyi-Stable-Diffusion-1B-Chinese-v0.1', model_revision='v1.0.0')
Loading pipeline components...: 0%| | 0/7 [00:00<?, ?it/s]
/home/winse/miniconda3/envs/demo/lib/python3.8/site-packages/transformers/models/clip/feature_extraction_clip.py:28:
FutureWarning: The class CLIPFeatureExtractor is deprecated and will be removed in version 5 of Transformers. Please use CLIPImageProcessor instead.
warnings.warn(
Loading pipeline components...: 57%|████████████████████████████▌ | 4/7 [00:32<00:23, 7.83s/it]
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["id2label"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["bos_token_id"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["eos_token_id"]` will be overriden.
Loading pipeline components...: 100%|██████████████████████████████████████████████████| 7/7 [03:23<00:00, 29.02s/it]
>>> prompt = '飞流直下三千尺，油画'
>>> output = pipe({'text': prompt})
/home/winse/miniconda3/envs/demo/lib/python3.8/site-packages/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion.py:889: FutureWarning: `callback_steps` is deprecated and will be removed in version 1.0.0. Passing `callback_steps` as an input argument to `__call__` is deprecated, consider using `callback_on_step_end`
deprecate(
We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked.
You may ignore this warning if your `pad_token_id` (0) is identical to the `bos_token_id` (0), `eos_token_id` (2), or the `sep_token_id` (None), and your input is not padded.
100%|██████████████████████████████████████████████████████████████████████████████| 50/50 [1:03:11<00:00, 75.83s/it]
>>> cv2.imwrite('result.png', output['output_imgs'][0])
True
>>>
再测个快的
有 cuDNN 加持确实快，10分钟就跑出来了！没安装之前时间估计是三个小时的！！！
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
>>> import torch
>>> from diffusers import StableDiffusionPipeline
>>> torch.backends.cudnn.benchmark = True
>>> pipe = StableDiffusionPipeline.from_pretrained("Fengshenbang/Taiyi-Stable-Diffusion-1B-Chinese-v0.1", torch_dtype=torch.float16)
Loading pipeline components...: 57%|████████████████████████████▌ | 4/7 [00:05<00:03, 1.32s/it]`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["id2label"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["bos_token_id"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["eos_token_id"]` will be overriden.
Loading pipeline components...: 100%|██████████████████████████████████████████████████| 7/7 [00:32<00:00, 4.58s/it]
>>> pipe.to('cuda')
StableDiffusionPipeline {
"_class_name": "StableDiffusionPipeline",
"_diffusers_version": "0.25.0",
"_name_or_path": "Fengshenbang/Taiyi-Stable-Diffusion-1B-Chinese-v0.1",
"feature_extractor": [
"transformers",
"CLIPFeatureExtractor"
],
"image_encoder": [
null,
null
],
"requires_safety_checker": true,
"safety_checker": [
"stable_diffusion",
"StableDiffusionSafetyChecker"
],
"scheduler": [
"diffusers",
"PNDMScheduler"
],
"text_encoder": [
"transformers",
"BertModel"
],
"tokenizer": [
"transformers",
"BertTokenizer"
],
"unet": [
"diffusers",
"UNet2DConditionModel"
],
"vae": [
"diffusers",
"AutoencoderKL"
]
}
>>>
>>> prompt = '飞流直下三千尺，油画'
>>> image = pipe(prompt, guidance_scale=7.5).images[0]
100%|████████████████████████████████████████████████████████████████████████████████| 50/50 [09:32<00:00, 11.45s/it]
>>> image.save("飞流.png")
>>>
改一下conda的名字
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
(demo) winse@DESKTOP-BR4MG38:ai$ conda deactivate
(base) winse@DESKTOP-BR4MG38:ai$ conda rename -n demo modelscope
Source: /home/winse/miniconda3/envs/demo
Destination: /home/winse/miniconda3/envs/modelscope
Packages: 22
Files: 33924
Downloading and Extracting Packages:
Downloading and Extracting Packages:
Preparing transaction: done
Verifying transaction: done
Executing transaction: done
(base) winse@DESKTOP-BR4MG38:ai$ conda env list
# conda environments:
#
base * /home/winse/miniconda3
modelscope /home/winse/miniconda3/envs/modelscope
模型下载脚本
比如下载：https://modelscope.cn/models/Fengshenbang/Taiyi-Stable-Diffusion-1B-Chinese-EN-v0.1/summary
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
$ tail -16 ~/.bashrc
function modelscope_download() {
model=$1
conda activate modelscope
#pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
#pip install modelscope
python -c "from modelscope.hub.snapshot_download import snapshot_download;snapshot_download('$model', cache_dir='./')"
conda deactivate
}
$ source ~/.bashrc
(base) winse@DESKTOP-BR4MG38:ai$ modelscope_download "Fengshenbang/Taiyi-Stable-Diffusion-1B-Chinese-EN-v0.1"
2024-01-15 15:17:13,613 - modelscope - INFO - PyTorch version 1.11.0+cu113 Found.
2024-01-15 15:17:13,614 - modelscope - INFO - Loading ast index from /home/winse/.cache/modelscope/ast_indexer
2024-01-15 15:17:13,637 - modelscope - INFO - Loading done! Current index file version is 1.11.0, with md5 85336421feb1dc1ec9dde85ceee20f42 and a total number of 953 components indexed
2024-01-15 15:17:16,879 - modelscope - WARNING - Model revision not specified, use revision: v1.0.0
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████| 2.31M/2.31M [00:00<00:00, 6.01MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████████| 612/612 [00:00<00:00, 436kB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████████| 547/547 [00:00<00:00, 804kB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████████| 743/743 [00:00<00:00, 1.25MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████| 4.46k/4.46k [00:00<00:00, 244kB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████████| 139/139 [00:00<00:00, 231kB/s]
Downloading: 100%|██████████████████████████████████████████████████████████████████████████████████████████▉| 3.20G/3.20G [02:36<00:00, 22.0MB/s]
Downloading: 100%|█████████████████████████████████████████████████████████████████████████████████████████████| 319M/319M [00:39<00:00, 8.40MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████| 2.11M/2.11M [00:00<00:00, 4.05MB/s]
Downloading: 100%|█████████████████████████████████████████████████████████████████████████████████████████████| 512k/512k [00:00<00:00, 1.28MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████████| 543/543 [00:00<00:00, 845kB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████████| 342/342 [00:00<00:00, 521kB/s]
Downloading: 100%|██████████████████████████████████████████████████████████████████████████████████████████▉| 1.13G/1.13G [01:22<00:00, 14.8MB/s]
Downloading: 100%|█████████████████████████████████████████████████████████████████████████████████████████████| 469M/469M [00:12<00:00, 39.0MB/s]
Downloading: 100%|█████████████████████████████████████████████████████████████████████████████████████████████| 5.00/5.00 [00:00<00:00, 6.02kB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████| 7.97k/7.97k [00:00<00:00, 7.06MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████████| 284/284 [00:00<00:00, 441kB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████████| 389/389 [00:00<00:00, 348kB/s]
Downloading: 100%|██████████████████████████████████████████████████████████████████████████████████████████▉| 11.3G/11.3G [04:07<00:00, 49.0MB/s]
Downloading: 100%|████████████████████████████████████████████████████████████████████████████████████████████████| 697/697 [00:00<00:00, 950kB/s]
Downloading: 100%|█████████████████████████████████████████████████████████████████████████████████████████████| 939k/939k [00:00<00:00, 2.50MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████| 2.87M/2.87M [00:00<00:00, 5.79MB/s]
Downloading: 100%|███████████████████████████████████████████████████████████████████████████████████████████| 3.44M/3.44M [00:00<00:00, 6.17MB/s]
#tree -L 2
(base) winse@DESKTOP-BR4MG38:ai$ tree -L 1 Fengshenbang/
Fengshenbang/
├── Taiyi-Stable-Diffusion-1B-Chinese-EN-v0.1
└── Taiyi-Stable-Diffusion-1B-Chinese-v0.1
2 directories, 0 files
再跑一个中英文的模型试试
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
(base) winse@DESKTOP-BR4MG38:ai$ conda activate modelscope
(modelscope) winse@DESKTOP-BR4MG38:ai$
(modelscope) winse@DESKTOP-BR4MG38:ai$ python
Python 3.8.18 | packaged by conda-forge | (default, Dec 23 2023, 17:21:28)
[GCC 12.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from diffusers import StableDiffusionPipeline
>>> pipe = StableDiffusionPipeline.from_pretrained("Fengshenbang/Taiyi-Stable-Diffusion-1B-Chinese-EN-v0.1").to("cuda")
Loading pipeline components...: 43%|█████████████████████████████████▊ | 3/7 [00:17<00:21, 5.46s/it]/home/winse/miniconda3/envs/modelscope/lib/python3.8/site-packages/transformers/models/clip/feature_extraction_clip.py:28: FutureWarning: The class CLIPFeatureExtractor is deprecated and will be removed in version 5 of Transformers. Please use CLIPImageProcessor instead.
warnings.warn(
Loading pipeline components...: 86%|███████████████████████████████████████████████████████████████████▋ | 6/7 [00:17<00:01, 1.90s/it]`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["id2label"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["bos_token_id"]` will be overriden.
`text_config_dict` is provided which will be used to initialize `CLIPTextConfig`. The value `text_config["eos_token_id"]` will be overriden.
Loading pipeline components...: 100%|███████████████████████████████████████████████████████████████████████████████| 7/7 [00:27<00:00, 3.87s/it]
>>>
>>> prompt = '小桥流水人家，Van Gogh style'
>>> image = pipe(prompt, guidance_scale=10).images[0]
100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████| 50/50 [14:19<00:00, 17.19s/it]
>>> image.save("小桥.png")
>>>
太乙webui - 亦步亦趋
这里记录了安装的详细过程，比较繁琐，如果直接安装可以跳到[太乙webui - 纯净版]。
选一个跟我现在用的环境一样的版本和系统：https://hub.docker.com/r/nvidia/cuda/tags
镜像的描述：CUDA and cuDNN images from gitlab.com/nvidia/cuda
试错
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
(modelscope) winse@DESKTOP-BR4MG38:ai$ docker pull nvidia/cuda:12.3.1-devel-ubuntu20.04
12.3.1-devel-ubuntu20.04: Pulling from nvidia/cuda
12.3.1-devel-ubuntu20.04: Pulling from nvidia/cuda
25ad149ed3cf: Pull complete
ba7b66a9df40: Pull complete
520797292d92: Pull complete
c5f2ffd06d8b: Pull complete
1698c67699a3: Pull complete
16dd7c0d35aa: Pull complete
568cac1e538c: Pull complete
6252d19a7f1d: Pull complete
f573e2686be4: Pull complete
0074e75104ac: Pull complete
df35fae9e247: Pull complete
Digest: sha256:befbdfddbb52727f9ce8d0c574cac0f631c606b1e6f0e523f3a0777fe2720c99
Status: Downloaded newer image for nvidia/cuda:12.3.1-devel-ubuntu20.04
docker.io/nvidia/cuda:12.3.1-devel-ubuntu20.04
What's Next?
1. Sign in to your Docker account → docker login
2. View a summary of image vulnerabilities and recommendations → docker scout quickview nvidia/cuda:12.3.1-devel-ubuntu20.04
(modelscope) winse@DESKTOP-BR4MG38:ai$ docker run --rm --gpus all --ipc host --ulimit memlock=-1 --ulimit stack=67108864 -it -v /mnt/i/ai:/app/stabilityai -p 7860:7860 docker.io/nvidia/cuda:12.3.1-devel-ubuntu20.04
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running hook #0: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as 'legacy'
nvidia-container-cli: requirement error: unsatisfied condition: cuda>=12.3, please update your driver to a newer version, or use an earlier cuda container: unknown.
重新下载镜像并配置
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
(modelscope) winse@DESKTOP-BR4MG38:ai$ nvidia-smi
Mon Jan 15 17:04:18 2024
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.146.01 Driver Version: 537.99 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 Quadro T2000 On | 00000000:01:00.0 On | N/A |
| N/A 46C P8 6W / 60W | 589MiB / 4096MiB | 7% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| 0 N/A N/A 39 G /Xwayland N/A |
| 0 N/A N/A 42 G /Xwayland N/A |
| 0 N/A N/A 44 G /Xwayland N/A |
+---------------------------------------------------------------------------------------+
版本不能高于本地CUDA，重新下载镜像：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
(modelscope) winse@DESKTOP-BR4MG38:ai$ docker pull nvidia/cuda:12.2.2-devel-ubuntu20.04
12.2.2-devel-ubuntu20.04: Pulling from nvidia/cuda
12.2.2-devel-ubuntu20.04: Pulling from nvidia/cuda
96d54c3075c9: Pull complete
db26cf78ae4f: Pull complete
5adc7ab504d3: Pull complete
e4f230263527: Pull complete
95e3f492d47e: Pull complete
35dd1979297e: Pull complete
39a2c88664b3: Pull complete
d8f6b6cd09da: Pull complete
fe19bbed4a4a: Pull complete
469ef7e9efe0: Pull complete
e30c6425f419: Pull complete
Digest: sha256:b7074ef6f9aa30c27fe747f3a7e10402ec442f001290718c73e0972d1ee61342
Status: Downloaded newer image for nvidia/cuda:12.2.2-devel-ubuntu20.04
docker.io/nvidia/cuda:12.2.2-devel-ubuntu20.04
What's Next?
1. Sign in to your Docker account → docker login
2. View a summary of image vulnerabilities and recommendations → docker scout quickview nvidia/cuda:12.2.2-devel-ubuntu20.04
运行容器实例
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
(modelscope) winse@DESKTOP-BR4MG38:P15$ docker run --rm --gpus all --ipc host --ulimit memlock=-1 --ulimit stack=67108864 -it -v /mnt/i/ai:/app/stabilityai -p 7860:7860 docker.io/nvidia/cuda:12.2.2-devel-ubuntu20.04
==========
== CUDA ==
==========
CUDA Version 12.2.2
Container image Copyright (c) 2016-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
This container image and its contents are governed by the NVIDIA Deep Learning Container License.
By pulling and using the container, you accept the terms and conditions of this license:
https://developer.nvidia.com/ngc/nvidia-deep-learning-container-license
A copy of this license is made available in this container at /NGC-DL-CONTAINER-LICENSE for your convenience.
root@41af85cb0007:/# nvidia-smi
Mon Jan 15 13:25:43 2024
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.146.01 Driver Version: 537.99 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 Quadro T2000 On | 00000000:01:00.0 On | N/A |
| N/A 43C P8 3W / 60W | 620MiB / 4096MiB | 2% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| No running processes found |
+---------------------------------------------------------------------------------------+
root@41af85cb0007:/# docker ps -a
bash: docker: command not found
root@41af85cb0007:/# python -V
bash: python: command not found
尽管用的是WSL 2 based engine，但是不是Windows管理的。
配值webui
[解决被官方忽视的 AI 容器应用问题] https://soulteary.com/2022/12/09/use-docker-to-quickly-get-started-with-the-chinese-stable-diffusion-model-taiyi.html
https://github.com/soulteary/docker-stable-diffusion-taiyi/blob/main/docker/Dockerfile
是对官方的依赖安装的拆解。可以参考，还是不建议这么搞，如果能促成源头修正那就是另一种说法了。
由于他的镜像我也跑不起来，所以直接按照官方来安装，参考借鉴他遇到解决过的问题。
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
root@41af85cb0007:/# sed -i.bak -e 's|archive.ubuntu.com/ubuntu/|mirrors.tuna.tsinghua.edu.cn/ubuntu/|' -e 's|security.ubuntu.com/ubuntu/|mirrors.tuna.tsinghua.edu.cn/ubuntu/|' /etc/apt/sources.list
root@41af85cb0007:/# apt update
root@41af85cb0007:/# apt install -y git wget curl iputils-ping iproute2 traceroute
root@41af85cb0007:/# wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
root@41af85cb0007:/# bash Miniconda3-latest-Linux-x86_64.sh -b
PREFIX=/root/miniconda3
Unpacking payload ...
Installing base environment...
Downloading and Extracting Packages:
Downloading and Extracting Packages:
Preparing transaction: done
Executing transaction: done
installation finished.
root@41af85cb0007:/# /root/miniconda3/bin/conda init bash
no change /root/miniconda3/condabin/conda
no change /root/miniconda3/bin/conda
no change /root/miniconda3/bin/conda-env
no change /root/miniconda3/bin/activate
no change /root/miniconda3/bin/deactivate
no change /root/miniconda3/etc/profile.d/conda.sh
no change /root/miniconda3/etc/fish/conf.d/conda.fish
no change /root/miniconda3/shell/condabin/Conda.psm1
no change /root/miniconda3/shell/condabin/conda-hook.ps1
no change /root/miniconda3/lib/python3.11/site-packages/xontrib/conda.xsh
no change /root/miniconda3/etc/profile.d/conda.csh
modified /root/.bashrc
==> For changes to take effect, close and re-open your current shell. <==
root@41af85cb0007:/# source ~/.bashrc
(base) root@41af85cb0007:/#
#https://github.com/IDEA-CCNL/stable-diffusion-webui/zipball/master/
#https://github.com/IDEA-CCNL/stable-diffusion-webui/tarball/master/
#https://docs.github.com/en/repositories/working-with-files/using-files/downloading-source-code-archives#source-code-archive-urls
(base) root@41af85cb0007:/opt# wget -c https://github.com/IDEA-CCNL/stable-diffusion-webui/archive/refs/heads/master.tar.gz -O stable-diffusion-webui.tgz
(base) root@41af85cb0007:/opt# tar zxf stable-diffusion-webui.tgz
(base) root@41af85cb0007:/opt# mv stable-diffusion-webui-master stable-diffusion-webui
(base) root@41af85cb0007:/opt# cd stable-diffusion-webui
#@@ 反正走代理，没必要
#(webui) root@41af85cb0007:/opt/stable-diffusion-webui# pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
#Writing to /root/.config/pip/pip.conf
#https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/4345
#https://stackoverflow.com/questions/75099182/stable-diffusion-error-couldnt-install-torch-no-matching-distribution-found
#ERROR: Ignored the following versions that require a different python version: 1.6.2 Requires-Python >=3.7,<3.10; 1.6.3 Requires-Python >=3.7,<3.10; 1.7.0 Requires-Python >=3.7,<3.10; 1.7.1 Requires-Python >=3.7,<3.10
(base) root@41af85cb0007:/opt/stable-diffusion-webui# conda create -n py39 python=3.9
(base) root@41af85cb0007:/opt/stable-diffusion-webui# conda activate py39
(py39) root@41af85cb0007:/opt/stable-diffusion-webui#
#@@ 用到github，得socks代理一下，@@先去掉代理不然又解析不了@@
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# unset all_proxy && unset ALL_PROXY && unset https_proxy && unset HTTPS_PROXY
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# pip install pysocks
Collecting pysocks
WARNING: Retrying (Retry(total=4, connect=None, read=None, redirect=None, status=None)) after connection broken by 'ConnectTimeoutError(<pip._vendor.urllib3.connection.HTTPSConnection object at 0x7fca218ae370>, 'Connection to files.pythonhosted.org timed out. (connect timeout=15)')': /packages/8d/59/b4572118e098ac8e46e399a1dd0f2d85403ce8bbaad9ec79373ed6badaf9/PySocks-1.7.1-py3-none-any.whl
Downloading PySocks-1.7.1-py3-none-any.whl (16 kB)
Installing collected packages: pysocks
Successfully installed pysocks-1.7.1
WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv
#?? ImportError: libGL.so.1: cannot open shared object file: No such file or directory
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# apt-get install ffmpeg libsm6 libxext6 -y
...
Setting up tzdata (2023c-0ubuntu0.20.04.2) ...
debconf: unable to initialize frontend: Dialog
debconf: (No usable dialog-like program is installed, so the dialog based frontend cannot be used. at /usr/share/perl5/Debconf/FrontEnd/Dialog.pm line 76.)
debconf: falling back to frontend: Readline
Configuring tzdata
------------------
Please select the geographic area in which you live. Subsequent configuration questions will narrow this down by presenting a list of cities, representing the time zones in which they are located.
1. Africa 2. America 3. Antarctica 4. Australia 5. Arctic 6. Asia 7. Atlantic 8. Europe 9. Indian 10. Pacific 11. SystemV 12. US 13. Etc
Geographic area: 6
Please select the city or region corresponding to your time zone.
1. Aden 9. Baghdad 17. Chita 25. Dushanbe 33. Irkutsk 41. Kashgar 49. Macau 57. Omsk 65. Rangoon 73. Taipei 81. Ujung_Pandang 89. Yekaterinburg
2. Almaty 10. Bahrain 18. Choibalsan 26. Famagusta 34. Istanbul 42. Kathmandu 50. Magadan 58. Oral 66. Riyadh 74. Tashkent 82. Ulaanbaatar 90. Yerevan
3. Amman 11. Baku 19. Chongqing 27. Gaza 35. Jakarta 43. Khandyga 51. Makassar 59. Phnom_Penh 67. Sakhalin 75. Tbilisi 83. Urumqi
4. Anadyr 12. Bangkok 20. Colombo 28. Harbin 36. Jayapura 44. Kolkata 52. Manila 60. Pontianak 68. Samarkand 76. Tehran 84. Ust-Nera
5. Aqtau 13. Barnaul 21. Damascus 29. Hebron 37. Jerusalem 45. Krasnoyarsk 53. Muscat 61. Pyongyang 69. Seoul 77. Tel_Aviv 85. Vientiane
6. Aqtobe 14. Beirut 22. Dhaka 30. Ho_Chi_Minh 38. Kabul 46. Kuala_Lumpur 54. Nicosia 62. Qatar 70. Shanghai 78. Thimphu 86. Vladivostok
7. Ashgabat 15. Bishkek 23. Dili 31. Hong_Kong 39. Kamchatka 47. Kuching 55. Novokuznetsk 63. Qostanay 71. Singapore 79. Tokyo 87. Yakutsk
8. Atyrau 16. Brunei 24. Dubai 32. Hovd 40. Karachi 48. Kuwait 56. Novosibirsk 64. Qyzylorda 72. Srednekolymsk 80. Tomsk 88. Yangon
Time zone: 70
Current default time zone: 'Asia/Shanghai'
Local time is now: Tue Jan 16 02:00:49 CST 2024.
Universal Time is now: Mon Jan 15 18:00:49 UTC 2024.
Run 'dpkg-reconfigure tzdata' if you wish to change it.
Setting up libxcb-present0:amd64 (1.14-2) ...
Setting up libglib2.0-data (2.64.6-1~ubuntu20.04.6) ...
Setting up libslang2:amd64 (2.3.2-4) ...
....
Setting up libavdevice58:amd64 (7:4.2.7-0ubuntu0.1) ...
Setting up ffmpeg (7:4.2.7-0ubuntu0.1) ...
Processing triggers for libc-bin (2.31-0ubuntu9.12) ...
/sbin/ldconfig.real: /lib/x86_64-linux-gnu/libcudadebugger.so.1 is not a symbolic link
/sbin/ldconfig.real: /lib/x86_64-linux-gnu/libcuda.so.1 is not a symbolic link
Processing triggers for libgdk-pixbuf2.0-0:amd64 (2.40.0+dfsg-3ubuntu0.4) ...
#@@ 改改改：1 可以root跑，2 直接用当前项目，下载没带.git， 3 使用conda管理环境
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# diff -u webui.sh.back webui.sh
--- webui.sh.back 2024-01-16 02:10:15.775277218 +0800
+++ webui.sh 2024-01-16 02:41:22.748823781 +0800
@@ -64,23 +64,23 @@
if [[ $(id -u) -eq 0 ]]
then
printf "\n%s\n" "${delimiter}"
- printf "\e[1m\e[31mERROR: This script must not be launched as root, aborting...\e[0m"
- printf "\n%s\n" "${delimiter}"
- exit 1
+# printf "\e[1m\e[31mERROR: This script must not be launched as root, aborting...\e[0m"
+# printf "\n%s\n" "${delimiter}"
+# exit 1
else
printf "\n%s\n" "${delimiter}"
printf "Running on \e[1m\e[32m%s\e[0m user" "$(whoami)"
printf "\n%s\n" "${delimiter}"
fi
-if [[ -d .git ]]
-then
+#if [[ -d .git ]]
+#then
printf "\n%s\n" "${delimiter}"
printf "Repo already cloned, using it as install directory"
printf "\n%s\n" "${delimiter}"
install_dir="${PWD}/../"
clone_dir="${PWD##*/}"
-fi
+#fi
# Check prerequisites
for preq in "${GIT}" "${python_cmd}"
@@ -120,19 +120,20 @@
cd "${install_dir}"/"${clone_dir}"/ || { printf "\e[1m\e[31mERROR: Can't cd to %s/%s/, aborting...\e[0m" "${install_dir}" "${clone_dir}"; exit 1; }
if [[ ! -d "${venv_dir}" ]]
then
- "${python_cmd}" -m venv "${venv_dir}"
+# "${python_cmd}" -m venv "${venv_dir}"
+ mkdir -p "${venv_dir}"
first_launch=1
fi
# shellcheck source=/dev/null
-if [[ -f "${venv_dir}"/bin/activate ]]
-then
- source "${venv_dir}"/bin/activate
-else
- printf "\n%s\n" "${delimiter}"
- printf "\e[1m\e[31mERROR: Cannot activate python venv, aborting...\e[0m"
- printf "\n%s\n" "${delimiter}"
- exit 1
-fi
+#if [[ -f "${venv_dir}"/bin/activate ]]
+#then
+# source "${venv_dir}"/bin/activate
+#else
+# printf "\n%s\n" "${delimiter}"
+# printf "\e[1m\e[31mERROR: Cannot activate python venv, aborting...\e[0m"
+# printf "\n%s\n" "${delimiter}"
+# exit 1
+#fi
printf "\n%s\n" "${delimiter}"
printf "Launching launch.py..."
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# export HTTPS_PROXY=socks5://172.22.240.1:23333
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# bash webui.sh
################################################################
Install script for stable-diffusion + Web UI
Tested on Debian 11 (Bullseye)
################################################################
################################################################
################################################################
Repo already cloned, using it as install directory
################################################################
################################################################
Launching launch.py...
################################################################
Python 3.9.18 (main, Sep 11 2023, 13:41:44)
[GCC 11.2.0]
Commit hash: <none>
Installing torch and torchvision
Installing gfpgan
Installing clip
Cloning Stable Diffusion into repositories/stable-diffusion...
Cloning Taming Transformers into repositories/taming-transformers...
Cloning K-diffusion into repositories/k-diffusion...
Cloning CodeFormer into repositories/CodeFormer...
Cloning BLIP into repositories/BLIP...
Installing requirements for CodeFormer
Installing requirements for Web UI
repositories/Taiyi-Stable-Diffusion-1B-Chinese-v0.1/feature_extractor/preprocessor_config.json | File missing.
repositories/Taiyi-Stable-Diffusion-1B-Chinese-v0.1 does not exist or file is missing. (1)Do you want to redownload the Taiyi model? Or (2)move your downloaded Taiyi model path? 1/2: 2
Detection failed, please reconfirm that the model has been moved to: repositories/Taiyi-Stable-Diffusion-1B-Chinese-v0.1
Please move the Taiyi model to: repositories/Taiyi-Stable-Diffusion-1B-Chinese-v0.1. Completed? y: y
???
File "/root/miniconda3/envs/py39/lib/python3.9/site-packages/httpx/_transports/default.py", line 275, in __init__
self._pool = httpcore.AsyncConnectionPool(
TypeError: __init__() got an unexpected keyword argument 'socket_options'
#https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/13236
#pip install -U httpcore
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# pip3 install httpx==0.24.1
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# unset all_proxy && unset ALL_PROXY && unset https_proxy && unset HTTPS_PROXY && unset http_proxy && unset HTTP_PROXY
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# bash webui.sh
???
ImportError: cannot import name '_compare_version' from 'torchmetrics.utilities.imports' (/root/miniconda3/envs/py39/lib/python3.9/site-packages/torchmetrics/utilities/imports.py)
#https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/11648
#conda list torchmetrics
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# conda install --force-reinstall torchmetrics==0.11.4
pip install torchmetrics==0.11.4 torch==1.13.1+cu117 torchvision==0.14.1+cu117 torchtext==0.14.1 torchaudio==0.13.1 torchdata==0.5.1 --extra-index-url https://download.pytorch.org/whl/cu117
???
export COMMANDLINE_ARGS="--lowvram --precision full --no-half --skip-torch-cuda-test"
???
RuntimeError: Cannot add middleware after an application has started
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# pip install fastapi==0.90.1
@@终于启动了
(py39) root@41af85cb0007:/opt/stable-diffusion-webui# bash webui.sh
################################################################
Install script for stable-diffusion + Web UI
Tested on Debian 11 (Bullseye)
################################################################
################################################################
################################################################
Repo already cloned, using it as install directory
################################################################
################################################################
Create and activate python venv
################################################################
################################################################
Launching launch.py...
################################################################
Python 3.9.18 (main, Sep 11 2023, 13:41:44)
[GCC 11.2.0]
Commit hash: <none>
Installing requirements for Web UI
Obtaining file:///opt/stable-diffusion-webui
ERROR: file:///opt/stable-diffusion-webui does not appear to be a Python project: neither 'setup.py' nor 'pyproject.toml' found.
Launching Web UI with arguments: --lowvram --precision full --no-half --ckpt /opt/stable-diffusion-webui/repositories/Taiyi-Stable-Diffusion-1B-Chinese-v0.1/Taiyi-Stable-Diffusion-1B-Chinese-v0.1.ckpt --listen --port 12345
LatentDiffusion: Running in eps-prediction mode
DiffusionWrapper has 859.52 M params.
making attention of type 'vanilla' with 512 in_channels
Working with z of shape (1, 4, 32, 32) = 4096 dimensions.
making attention of type 'vanilla' with 512 in_channels
Loading weights [e2e75020] from /opt/stable-diffusion-webui/repositories/Taiyi-Stable-Diffusion-1B-Chinese-v0.1/Taiyi-Stable-Diffusion-1B-Chinese-v0.1.ckpt
Applying cross attention optimization (Doggettx).
Model loaded.
Loaded a total of 0 textual inversion embeddings.
Embeddings:
Running on local URL: http://0.0.0.0:12345
To create a public link, set `share=True` in `launch()`.
启动镜像的时刻忘了挂数据U盘了，直接全部拷贝到容器里面：
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
(base) root@41af85cb0007:/opt/stable-diffusion-webui/repositories# mkdir Taiyi-Stable-Diffusion-1B-Chinese-v0.1
#(base) winse@DESKTOP-BR4MG38:Taiyi-Stable-Diffusion-1B-Chinese-v0.1$ tar tf 1.tar *
#(base) winse@DESKTOP-BR4MG38:Taiyi-Stable-Diffusion-1B-Chinese-v0.1$ docker cp 1.tar 41af85cb0007:/opt/stable-diffusion-webui/repositories/Taiyi-Stable-Diffusion-1B-Chinese-v0.1/
(base) root@41af85cb0007:/opt/stable-diffusion-webui/repositories/Taiyi-Stable-Diffusion-1B-Chinese-v0.1# tar xf 1.tar
(base) root@41af85cb0007:/opt/stable-diffusion-webui/repositories/Taiyi-Stable-Diffusion-1B-Chinese-v0.1# rm -rf 1.tar
(base) root@41af85cb0007:/opt/stable-diffusion-webui/repositories/Taiyi-Stable-Diffusion-1B-Chinese-v0.1# ll
total 4084196
drwxr-xr-x 9 root root 4096 Jan 15 17:57 ./
drwxrwxr-x 9 root root 4096 Jan 15 17:42 ../
-rwxrwxrwx 1 webui webui 4182159787 Jan 15 00:26 Taiyi-Stable-Diffusion-1B-Chinese-v0.1.ckpt*
-rwxrwxrwx 1 webui webui 146 Jan 15 00:19 configuration.json*
drwxrwxrwx 2 webui webui 4096 Jan 15 17:50 feature_extractor/
-rwxrwxrwx 1 webui webui 539 Jan 15 00:22 model_index.json*
drwxrwxrwx 2 webui webui 4096 Jan 15 17:50 safety_checker/
drwxrwxrwx 2 webui webui 4096 Jan 15 17:50 scheduler/
drwxrwxrwx 2 webui webui 4096 Jan 15 17:50 text_encoder/
drwxrwxrwx 2 webui webui 4096 Jan 15 17:50 tokenizer/
drwxrwxrwx 2 webui webui 4096 Jan 15 17:50 unet/
drwxrwxrwx 2 webui webui 4096 Jan 15 17:50 vae/
太乙webui - 纯净版
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
(base) winse@DESKTOP-BR4MG38:P15$ cat /etc/os-release
(base) winse@DESKTOP-BR4MG38:P15$ cat /etc/issue
Ubuntu 20.04.6 LTS \n \l
(base) winse@DESKTOP-BR4MG38:P15$ nvidia-smi
Tue Jan 16 07:09:11 2024
+---------------------------------------------------------------------------------------+
| NVIDIA-SMI 535.146.01 Driver Version: 537.99 CUDA Version: 12.2 |
|-----------------------------------------+----------------------+----------------------+
| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|=========================================+======================+======================|
| 0 Quadro T2000 On | 00000000:01:00.0 On | N/A |
| N/A 50C P8 4W / 60W | 537MiB / 4096MiB | 1% Default |
| | | N/A |
+-----------------------------------------+----------------------+----------------------+
+---------------------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=======================================================================================|
| 0 N/A N/A 32 G /Xwayland N/A |
| 0 N/A N/A 39 G /Xwayland N/A |
| 0 N/A N/A 41 G /Xwayland N/A |
+---------------------------------------------------------------------------------------+
选一个跟我现在用的环境一样的版本和系统：https://hub.docker.com/r/nvidia/cuda/tags CUDA and cuDNN images from gitlab.com/nvidia/cuda
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
(base) winse@DESKTOP-BR4MG38:P15$ docker pull nvidia/cuda:12.2.2-devel-ubuntu20.04
(base) winse@DESKTOP-BR4MG38:P15$ docker run --gpus all --ipc host --ulimit memlock=-1 --ulimit stack=67108864 -it -v /mnt/i/ai:/app/stabilityai -p 7860:7860 docker.io/nvidia/cuda:12.2.2-devel-ubuntu20.04
root@c65a73d918b1:/# sed -i.bak -e 's|archive.ubuntu.com/ubuntu/|mirrors.tuna.tsinghua.edu.cn/ubuntu/|' -e 's|security.ubuntu.com/ubuntu/|mirrors.tuna.tsinghua.edu.cn/ubuntu/|' /etc/apt/sources.list
root@c65a73d918b1:/# apt update
root@c65a73d918b1:/# apt install -y git wget vim
root@c65a73d918b1:/# cd
root@c65a73d918b1:~# wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
root@c65a73d918b1:~# bash Miniconda3-latest-Linux-x86_64.sh -b -u
root@c65a73d918b1:~# ~/miniconda3/bin/conda init bash
root@c65a73d918b1:~# source ~/.bashrc
(base) root@c65a73d918b1:~#
(base) root@c65a73d918b1:~# wget -c https://github.com/IDEA-CCNL/stable-diffusion-webui/archive/refs/heads/master.tar.gz -O - | tar zxf -
(base) root@c65a73d918b1:~# mv stable-diffusion-webui-master stable-diffusion-webui
(base) root@c65a73d918b1:~# cd stable-diffusion-webui
(base) root@c65a73d918b1:~/stable-diffusion-webui# conda create -n py3.10 python=3.10
(base) root@c65a73d918b1:~/stable-diffusion-webui# conda activate py3.10
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# mkdir .git
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# diff -u webui.sh.back webui.sh
--- webui.sh.back 2024-01-15 23:39:24.271256229 +0000
+++ webui.sh 2024-01-15 23:42:17.070404354 +0000
@@ -64,9 +64,6 @@
if [[ $(id -u) -eq 0 ]]
then
printf "\n%s\n" "${delimiter}"
- printf "\e[1m\e[31mERROR: This script must not be launched as root, aborting...\e[0m"
- printf "\n%s\n" "${delimiter}"
- exit 1
else
printf "\n%s\n" "${delimiter}"
printf "Running on \e[1m\e[32m%s\e[0m user" "$(whoami)"
@@ -120,19 +117,11 @@
cd "${install_dir}"/"${clone_dir}"/ || { printf "\e[1m\e[31mERROR: Can't cd to %s/%s/, aborting...\e[0m" "${install_dir}" "${clone_dir}"; exit 1; }
if [[ ! -d "${venv_dir}" ]]
then
- "${python_cmd}" -m venv "${venv_dir}"
+# "${python_cmd}" -m venv "${venv_dir}"
+ mkdir -p "${venv_dir}"
first_launch=1
fi
# shellcheck source=/dev/null
-if [[ -f "${venv_dir}"/bin/activate ]]
-then
- source "${venv_dir}"/bin/activate
-else
- printf "\n%s\n" "${delimiter}"
- printf "\e[1m\e[31mERROR: Cannot activate python venv, aborting...\e[0m"
- printf "\n%s\n" "${delimiter}"
- exit 1
-fi
printf "\n%s\n" "${delimiter}"
printf "Launching launch.py..."
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# pip install pysocks
#https://github.com/invoke-ai/InvokeAI/issues/3560#issuecomment-1689474997
#https://blog.csdn.net/shark1357/article/details/131238924
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# pip install tb-nightly -i https://mirrors.aliyun.com/pypi/simple
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# pip install gfpgan==1.3.8
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# apt-get install ffmpeg libsm6 libxext6 -y
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# pip install httpcore httpx==0.24.1 torchmetrics==0.11.4 fastapi==0.90.1
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# vi webui-user.sh
export COMMANDLINE_ARGS="--lowvram --precision full --no-half --skip-torch-cuda-test"
(py3.10) root@c65a73d918b1:~# cd stable-diffusion-webui/repositories/
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui/repositories# ln -s /app/stabilityai/Fengshenbang/Taiyi-Stable-Diffusion-1B-Chinese-v0.1/
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui/repositories# ll Taiyi-Stable-Diffusion-1B-Chinese-v0.1
lrwxrwxrwx 1 root root 69 Jan 16 07:58 Taiyi-Stable-Diffusion-1B-Chinese-v0.1 -> /app/stabilityai/Fengshenbang/Taiyi-Stable-Diffusion-1B-Chinese-v0.1//
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# pip install socksio httpx[socks]
#@@ 加代理下载部署时需要的github代码
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# export HTTPS_PROXY=socks5://172.22.240.1:23333
(py3.10) root@c65a73d918b1:~/stable-diffusion-webui# bash webui.sh --port 7860
然后在Windows浏览器访问： http://localhost:7860/
TODO
汉化： https://github.com/VinsonLaro/stable-diffusion-webui-chinese
–END
