Link Diary March 2023

My march in links.

• While finding resources on GPT ...
  • Tons of training data may come as JSON
  • Text dumps to train a language model
  • GPT Alternatives anywhere?
  • Supporting software
  • A talktive Language Model
• Stable Diffusion WebUI
• InvokeAI
• Deepfakes
• Voice cloning with Tortoise TTS
• Simple Backup with rsync
• Digital Comic Book Readers for Linux
• Learn Ubuntu the latest nodejs version
• Moving a local git repository to remote
• Remember SSH passwords with ssh-agent
• View SSL Certificate
• Video Picks

It's a frickin' rabbit hole …

The famous ChatGPT:
https://openai.com/blog/chatgpt

Training good AI is expensive:
https://techcrunch.com/2022/12/30/theres-now-an-open-source-alternative-to-chatgpt-but-good-luck-running-it/

Slim data storage "JSON Lines"

• https://jsonlines.org/on_the_web/
• https://github.com/raphaelstolt/json-lines/blob/master/src/JsonLines.php

Text collections to train AI

• https://openwebtext2.readthedocs.io/en/latest/
  • https://skylion007.github.io/OpenWebTextCorpus/
• https://pile.eleuther.ai/

ChatGPT Alternatives:
https://nlpcloud.com/chatgpt-open-source-alternatives.html

OpenSource GPT JAX:
https://pile.eleuther.ai/

nanoGPT: A very small GPT waiting to be trained:
https://github.com/karpathy/nanoGPT

Pretrained GPT-2 models are also available:
https://huggingface.co/gpt2/tree/main

Facebook Library to determine language:
https://github.com/facebookresearch/fastText

Facebook's LLaMA, a probably illegal training with ChatGPT becomes Alpaca, the talktive Language model.

Runs at home, but don't expect ChatGPT level.

https://github.com/cocktailpeanut/dalai

https://github.com/AUTOMATIC1111/stable-diffusion-webui

More options than the Simple Diffusion UI.
You can add models like this one: Modern Disney Style and have fun with it.
https://huggingface.co/nitrosocke/mo-di-diffusion

Another stable diffusion frontend.

https://github.com/invoke-ai/InvokeAI/releases/tag/v2.3.1.post2

Images seem to be better. Multiple special pretrained models available directly.
I had to create the "outputs" folder before the WebUI could be started.

I digged a little looking to find a open source program to do deepfakes.

TL;DR: https://github.com/iperov/DeepFaceLive

Software to make Deepfakes:
https://fossbytes.com/best-deepfake-apps-list/

Brief article on DeepFakes technology:
https://www.forbes.com/sites/lutzfinger/2022/09/08/overview-of-how-to-create-deepfakesits-scarily-simple/

Transfer animation
https://github.com/AliaksandrSiarohin/first-order-model
And the paper for it
https://aliaksandrsiarohin.github.io/first-order-model-website/

DeepFaceLab project:
https://github.com/iperov/DeepFaceLab

Animate images:
https://www.d-id.com/

Faceswap:
https://faceswap.dev/
Installing Faceswap:
https://forum.faceswap.dev/viewtopic.php?f=4&t=68

https://www.youtube.com/watch?v=mq9lBd8XMY4
https://github.com/thisserand/tortoise_local

# -> to sync folderX into folderY/folderX
rsync -avc /the/source/folderX /the/target/folderY/

https://wiki.ubuntuusers.de/Skripte/Backup_mit_RSYNC/

Got a comic as part of a bundle and was wonderung with what I can enjoy the CBZ file format.

https://itsfoss.com/best-comic-book-reader-linux/

I wanted to install Strapi on my cloudatcost machine for experimenting.
Unfortunately the nodejs version of my selected Ubuntu 22.04 was at 12.x even after an apt get update; apt get upgrade. 16.x was the minimum requirement, 18.x recommended.

This is what I found:

sudo apt purge nodejs libnode72 libnode-dev
curl -sL https://deb.nodesource.com/setup_18.x -o nodesource_setup.sh
sudo bash ./nodesource_setup.sh
sudo apt install -y nodejs

https://www.digitalocean.com/community/tutorials/how-to-install-node-js-on-ubuntu-22-04

… keeping all commits and branches …

git remote add origin git@github.com:user/repo.git
git push --all origin
git push --tags origin

https://gist.github.com/niksumeiko/8972566

Problem was the existing "main" branch in my case which I could neither merge nor rebase properly… which I then circumvented by renaming mine and pushing this instead.

git branch -m main develop

Prerequisite:

• you can already log in with your SSH key by entering a passphrase - every time.
• You use the .ssh/config file to define which user and key is used for which domain.

sudo apt install "openssh-server"
eval `ssh-agent`
ssh-add ~/.ssh/my-private-key
ssh myserver.com

https://www.ssh.com/academy/ssh/agent

# full
openssl s_client -showcerts -connect keinerweiss.de:443
# only dates
openssl s_client -connect keinerweiss.de:443 2>/dev/null | openssl x509 -noout -dates

https://ma.ttias.be/how-to-read-ssl-certificate-info-from-the-cli/
https://linuxhandbook.com/check-certificate-openssl/

https://developer.apple.com/videos/play/wwdc2022/10037/

Disney Filter with Stable Diffusion

Rice Bread

Blender: fluffy stuff

Erinnerung: Heinz Erhardt

Clean Code vs Performance

Metal for mental health

The truth is funny

AI depth map in video

How to start a speech

Light is a wave

Blender at Studio

Be an idiot!

OEmbed Error

• Provider returned HTTP Status 401 for
• All resolution methods failed

Stable diffusion: Foto to art

OEmbed Error

• Provider returned HTTP Status 401 for
• All resolution methods failed
• Provider returned HTTP Status 403 for
• All resolution methods failed

Learning ARK

100 years men fashion

100 years women fashion

Try a denutter

Powerpoint mistakes