View difference between Paste ID: 7p4U0bzm and uE1WcSxD
SHOW: | | - or go back to the newest paste.
1
== https://github.com/ShivamShrirao/diffusers/tree/main/examples/dreambooth ==
2
3
* Make a place for downloads
4
mkdir Downloads
5
cd Downloads
6
7
* Download & Install Anaconda
8
wget https://repo.anaconda.com/archive/Anaconda3-2022.05-Linux-x86_64.sh
9
chmod +x ./Anaconda3-2022.05-Linux-x86_64.sh
10
./Anaconda3-2022.05-Linux-x86_64.sh
11
12
* Nvidia CUDA toolkit for Ubuntu WSL - https://developer.nvidia.com/cuda-downloads
13
wget https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-wsl-ubuntu.pin
14
sudo mv cuda-wsl-ubuntu.pin /etc/apt/preferences.d/cuda-repository-pin-600
15
wget https://developer.download.nvidia.com/compute/cuda/11.7.1/local_installers/cuda-repo-wsl-ubuntu-11-7-local_11.7.1-1_amd64.deb
16
sudo dpkg -i cuda-repo-wsl-ubuntu-11-7-local_11.7.1-1_amd64.deb
17
sudo cp /var/cuda-repo-wsl-ubuntu-11-7-local/cuda-*-keyring.gpg /usr/share/keyrings/
18
sudo apt-get update
19
sudo apt-get -y install cuda
20
21
* New shell with Anaconda & install g++
22
bash
23
sudo apt install build-essential
24
25
* New virtual Python environment
26
conda update -n base -c defaults conda
27
conda create --name diffusers python=3.9
28
conda activate diffusers
29
30
* Make a directory for all your github downloads, then download diffusers
31
mkdir ~/github
32
cd ~/github
33
git clone https://github.com/ShivamShrirao/diffusers.git
34
cd diffusers
35
36
* Install required packages
37
pip install .
38
cd examples/dreambooth
39
pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu116
40
pip install -r requirements.txt
41
pip install -U --pre triton
42
pip install ninja bitsandbytes
43
pip install git+https://github.com/facebookresearch/xformers@1d31a3a#egg=xformers
44
45
* Configure / login
46
accelerate config
47
huggingface-cli login
48
mkdir -p training
49
mkdir -p classes
50
explorer.exe .
51
52
* Edit your training file and add this line to the top or reboot:
53
export LD_LIBRARY_PATH=/usr/lib/wsl/lib:$LD_LIBRARY_PATH
54
55
* Copy your training file from your Windows desktop (or wherever you saved it):
56
cp /mnt/c/Users/YOUR_USERNAME/Desktop/my_training.sh ./
57
chmod +x ./my_training.sh
58
./my_training.sh
59
60
61
* NB: If you get an error like this - 
62
403 Client Error: Forbidden for url: https://huggingface.co/CompVis/stable-diffusion-v1-4/resolve/main/model_index.json
63
64
Then:
65
Make sure you've visited https://huggingface.co/CompVis/stable-diffusion-v1-4 and accepted the licence!
66
NB. If this is the very first time you've used diffusers, it's about a 6GB download
67
68
To train again next time, simply:
69
70
* Start Ubuntu, then:
71
conda activate diffusers
72
cd ~/github/diffusers/examples/dreambooth
73
./my_training.sh
74