Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/camenduru/train-text-to-image-tpu-tutorial
https://github.com/camenduru/train-text-to-image-tpu-tutorial
Last synced: 12 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/camenduru/train-text-to-image-tpu-tutorial
- Owner: camenduru
- License: unlicense
- Created: 2023-01-20T01:32:57.000Z (almost 2 years ago)
- Default Branch: main
- Last Pushed: 2023-06-26T08:29:48.000Z (over 1 year ago)
- Last Synced: 2024-10-04T13:31:23.934Z (about 1 month ago)
- Language: Python
- Size: 64.5 KB
- Stars: 21
- Watchers: 3
- Forks: 7
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Funding: .github/FUNDING.yml
- License: LICENSE
Awesome Lists containing this project
README
🐣 Please follow me for new updates https://twitter.com/camenduru
🔥 Please join our discord server https://discord.gg/k5BwmmvJJU
🥳 Please join my patreon community https://patreon.com/camenduru
This repo contains all codes and commands used in `train text to image with tpu tutorial` https://youtu.be/NGta-t4BoLY
## Prepare TPU VM
use with 🐧 linux or linux inside windows (wsl)
https://console.cloud.google.com
```sh
echo "deb [signed-by=/usr/share/keyrings/cloud.google.gpg] https://packages.cloud.google.com/apt cloud-sdk main" | sudo tee -a /etc/apt/sources.list.d/google-cloud-sdk.list
curl https://packages.cloud.google.com/apt/doc/apt-key.gpg | sudo apt-key --keyring /usr/share/keyrings/cloud.google.gpg add -
sudo apt-get update && sudo apt-get install google-cloud-cli
gcloud init
```
```sh
gcloud alpha compute tpus tpu-vm ssh node-1 --zone us-central1-f
```
```py
pip install -U zipp "jax[tpu]==0.3.23" -f https://storage.googleapis.com/jax-releases/libtpu_releases.html packaging flax==0.6.1 chex==0.1.5 orbax==0.0.13 numpy diffusers==0.10.0 transformers piexif fold_to_ascii discord ftfy dill urllib3 datasets importlib-metadata accelerate==0.16.0 OmegaConf wandb==0.13.4 optax torch torchvision modelcards pytorch_lightning protobuf==3.20.* tensorboard markupsafe==2.0.1 gradio
sudo apt install git-lfs
```
```sh
mkdir tpu
cd tpu
mkdir train
cd train
```
```sh
wget https://raw.githubusercontent.com/camenduru/train-text-to-image-tpu-tutorial/main/train_text_to_image_flax.py
```
```sh
tmux
```
## Prepare Dataset
[](https://colab.research.google.com/github/camenduru/train-text-to-image-tpu-tutorial/blob/main/prepare_dataset_colab.ipynb)
## Train
```py
python3 train_text_to_image_flax.py \
--pretrained_model_name_or_path="flax/sd15-non-ema" \
--dataset_name="camenduru/test" \
--resolution=512 \
--mixed_precision="bf16" \
--train_batch_size=1 \
--num_train_epochs=650 \
--learning_rate=1e-6 \
--max_grad_norm=1 \
--output_dir="test" \
--report_to="wandb"
```
## Push Trained model to 🤗
```sh
wget https://raw.githubusercontent.com/camenduru/train-text-to-image-tpu-tutorial/main/push.py
python3 push.py
```
## Convert Flax model to PyTorch
https://huggingface.co/spaces/camenduru/converter
## Test Flax model or PyTorch model
https://github.com/camenduru/stable-diffusion-diffusers-colab
## Outputs
https://huggingface.co/camenduru/tpu-train-tutorial-flax
https://huggingface.co/camenduru/tpu-train-tutorial-pt
## YouTube Live VOD
https://youtu.be/NGta-t4BoLY
## Main Repo ♥
https://github.com/huggingface/diffusers
## Scripts From ♥
https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image_flax.py